Thursday, August 15, 2013

My brother was trying to tell me that the lack of ice in the Arctic is not unusual, bullshit.

I ran into this recently, I found it interesting how the recent sea ice quantities have approximated the "falling off a cliff" scenario so well...

What do you think?

Here is an overview...

And here is a link to the original if you want to look at the fancy expensive version...

In other news who wants to buy me a subscription to nature?

Monday, August 12, 2013

The Relation Between Intelligence and Religiosity: A Meta-Analysis and Some Proposed Explanations

I Posted this here for reference purposes only, if you want to read the full article with intact tables go download the original from

----------------------- Page 1-----------------------

     Personality and Social Psychology                                                                       Review                                                                                                                                                                                                                                                  The Relation Between Intelligence and Religiosity: A Meta-Analysis and Some Proposed Explanations                                                 Miron Zuckerman, Jordan Silberman and Judith A. Hall                                                 Pers Soc Psychol Rev published online 6 August 2013                                                                DOI: 10.1177/1088868313497266                                                                                                                                               The online version of this article can be found at:                                                                                                             Published by:                                                                                                                                                                                                                                     On behalf of:                                                           Society for Personality and Social Psychology                                                                                                                                                                                                                                                                                                                                                                                                 Additional services and information for Personality and Social Psychology Review can be found at:                                                                                                                                                       Email Alerts:                                                                                                                                                    Subscriptions:                                                                                                                                                 Reprints:                                                                                                                                            Permissions:                                                       >> OnlineFirst Version of Record - Aug 6, 2013                                                                                                                                                                         What is This?                                                        Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 2-----------------------

497266PSRXXX 10.1177/1088868313497266<italic>Personality and Social Psychology Review</italic>Zuckerman et al. research-article2013                     Article                                                                                                                                                                    Personality and Social Psychology Review                                                                                                                                                                   XX(X) 1–30                     The Relation Between Intelligence and                                                                                                         © 2013 by the Society for Personality                                                                                                                                                                   and Social Psychology, Inc.                      Religiosity: A Meta-Analysis and Some                                                                                                         Reprints and permissions:                                                                                                                                                                                Proposed Explanations                                                                                                                         DOI: 10.1177/1088868313497266                                                                                                                                                                                                                     1                                        1                                       2                      Miron Zuckerman , Jordan Silberman , and Judith A. Hall                     Abstract                     A meta-analysis of 63 studies showed a significant negative association between intelligence and religiosity. The association                     was stronger for college students and the general population than for participants younger than college age; it was also                     stronger for religious beliefs than religious behavior. For college students and the general population, means of weighted                     and unweighted correlations between intelligence and the strength of religious beliefs ranged from −.20 to −.25 (mean r                     −.24). Three possible interpretations were discussed. First, intelligent people are less likely to conform and, thus, are more                     likely to resist religious dogma. Second, intelligent people tend to adopt an analytic (as opposed to intuitive) thinking style,                     which has been shown to undermine religious beliefs. Third, several functions of religiosity, including compensatory control,                     self-regulation, self-enhancement, and secure attachment, are also conferred by intelligence. Intelligent people may therefore                     have less need for religious beliefs and practices.                     Keywords                     intelligence, religiosity, meta-analysis                     For more than eight decades, researchers have been investi-                                          Religiosity can be defined as the degree of involvement in                     gating the association between intelligence levels and mea-                                     some       or    all  facets     of   religion.      According        to   Atran      and                     sures   of   religious   faith.   This   association   has   been   studied                     Norenzayan (2004), such facets include beliefs in supernatu-                     among individuals of all ages, using a variety of measures.                                     ral agents, costly commitment to these agents (e.g., offering                     Although a substantial body of research has developed, this                                     of property), using beliefs in those agents to lower existential                     literature        has     not     been      systematically          meta-analyzed.              anxieties such as anxiety over death, and communal rituals                     Furthermore,           proposed       explanations         for   the    intelligence–           that   validate   and   affirm   religious   beliefs.   Of   course,   some                     religiosity association have not been systematically reviewed.                                  individuals may express commitment or participate in com-                     In the present work, our goal was to meta-analyze studies on                                    munal   rituals   for   reasons   other   than   religious   beliefs. This                     the relation between intelligence and religiosity and present                                   issue was put into sharp relief by Allport and Ross (1967),                     possible explanations for this relation.                                                        who drew a distinction between intrinsic and extrinsic reli-                          Following Gottfredson (1997), we define intelligence as                                    gious orientations. Intrinsic orientation is the practice of reli-                     the “ability to reason, plan, solve problems, think abstractly,                                 gion for its own sake; extrinsic religion is the use of religion                     comprehend   complex   ideas,   learn   quickly   and   learn   from                            as a means to secular ends. This distinction will be referred                     experience” (p. 13). This definition of intelligence is often                                   to in later sections.                     referred to as analytic intelligence or the g factor—the first                                       Since the inception of IQ tests early in the 20th century,                     factor   that   emerges   in   factor   analyses   of   IQ   subtests   (e.g.,                  intelligence has continuously occupied a central position in                     Carroll,       1993;     Spearman,         1904).      Other     newly       identified         psychological           research      (for    a  summary         of   the   field,    see                     types of intelligence, such as creative intelligence (Sternberg,                      1999,   2006)   or   emotional   intelligence   (Mayer,   Caruso,   &                           1                                                                                                                      University of Rochester, NY, USA                     Salovey,   1999),   are   out   of   the   scope   of   the   present   work                    2                                                                                                                      Northeastern University, Boston, MA, USA                     because the available studies on the relation between intelli-                     gence and religiosity examined only analytic intelligence. In                                   Corresponding Author:                                                                                                                      Miron Zuckerman, Department of Clinical and Social Sciences in                     addition, there are still disputes about the nature of nonana-                                                                                                                      Psychology, University of Rochester, PO Box 270266, Meliora 431,                     lytic     intelligence        (see    recent     exchange         between       Mayer,           Rochester, NY 14627, USA.                     Caruso, Panter, & Salovey, 2012, and Nisbett et al., 2012a).                                     Email:                                                                       Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 3-----------------------

2                                                                                                   Personality and Social Psychology Review XX(X) Nisbett et al., 2012b). Religion, on the other hand, has a more                  these results supported the Kosa and Schommer prediction. intermittent history. Gorsuch (1988) noted that interest in the                  Unfortunately, measures of religious beliefs were not used. psychology       of   religion   was    strong   before    1930,    almost          Hoge (1969, 1974) tracked changes in religious attitudes extinct between 1930 and 1960, and on the rise after 1960.                       on 13 American campuses. He compared survey data, most This latter trend has accelerated in recent years. Indeed, it is                 of which were collected between 1930 and 1948, with data safe to say that the bulk of the present content of psychology                   that he collected himself in 1967 and 1968. On four cam- of religion has been constructed over the last 20 years (see,                    puses, Hoge also examined the relation between SAT scores for example, Atran & Norenzayan, 2004, or the special reli-                      and religious attitudes. Correlations were small and mostly giosity issues of Personality and Social Psychology Review,                      negative. Hoge (1969) concluded that “no organic or psychic February 2010, and of the Journal of Social Issues , December                    relationship   exists   between   intelligence   and   religious   atti- 2005).                                                                           tudes and . . . the relationships found by researchers are either     Given the importance of both intelligence and religious                      due   to   educational   influences   or   biases   in   the   intelligence beliefs in psychological research, the relation between them                     tests” (p. 215). Hoge acknowledged that range restrictions of constitutes an intriguing question. Indeed, as shown below,                      college   students’  intelligence   scores   may   decrease   correla- this question attracted attention very early in the history of                   tions between intelligence and other variables. Nevertheless, psychological       research,    and   it  continues   to  foster   debate       he   concluded   that   the   low   negative-intelligence-religiosity today.   The   nature   of   the   relation   between   intelligence   and       correlations implied that there is no relation between intelli- religiosity can advance our knowledge about both constructs:                     gence and religiosity. We   might   learn   who   holds   religious   beliefs   and   why;   we             Seventeen years later, in a revision of Argyle’s 1958 book, might also learn how and why intelligent people do (or do                        Argyle and Beit-Hallahmi (1975) again reviewed the litera- not) develop a particular belief system.                                         ture   on   the  relation   between     intelligence    and    religiosity.                                                                                  Unlike the first edition, the revised monograph did not offer The Relation Between Intelligence and                                            a   conclusion   regarding   the   magnitude   or   direction   of   this                                                                                  relation.    This    waning     conviction     continued      during    the Religiosity: A Brief History                                                                                  remainder   of   the   century.   For   example,   Beit-Hallahmi   and To our knowledge, the first studies on intelligence and religi-                  Argyle (1997) suggested that “there are no great differences osity   appeared   in   1928,   in   the  University   of   Iowa   Studies       in   intelligence    between      the  religious    and   non-religious, series, Studies in Character (Howells, 1928; Sinclair, 1928).                    though fundamentalists score a little lower” (p. 183). They These studies examined sensory, motor, and cognitive cor-                        also noted the lack of large-scale studies that controlled for relates of religiosity. Intelligence tests were included in the                  demographic   variables,   “or   any   studies   which   make   clear battery    of   administered     tasks.   Both    Howells     (1928)    and      what the direction of causation is, if there is any effect at all” Sinclair (1928) found that higher levels of intelligence were                    (p. 177). In an introduction to his own study, Francis (1998) related to lower levels of religiosity.                                          reviewed the published evidence and stated that the number     Accumulation   of   additional   research   during   the   subse-            of studies reporting a negative relation exceeded the number quent three decades prompted Argyle (1958) to review the                         reporting a positive relation or no relation. However, his own available evidence. He concluded that “intelligent students                      findings   from   that   study   as   well   as   others   (e.g.,   Francis, are much less likely to accept orthodox beliefs, and rather                      1979) showed no relation, posing a clear challenge to “the less   likely   to   have   pro-religious   attitudes”   (p.   96). Argyle       research consensus formulated in the late 1950s by Argyle also noted that, as of 1958, all available evidence was based                    (1958)” (Francis, 1998, p. 192). Ironically, Francis worked on children or college student samples. He speculated, how-                      exclusively      with   children    and   adolescents—precisely         the ever, that the same results might be observed for adults of                      population   that,   according   to Argyle   (1958),   does   show   a post-college age.                                                                negative relation between intelligence and religiosity.     In   the   subsequent   decade,   the   pendulum   swung   in   the             As if in response to Beit-Hallahmi and Argyle’s (1997) opposite   direction.   Kosa   and   Schommer   (1961)   and   Hoge              call, the last decade has seen a number of large-scale studies (1969) drew conclusions from their data that were inconsis-                      that examined the relation between intelligence and religios- tent   with   those   of Argyle   (1958). According   to   Kosa   and            ity   (Kanazawa,       2010a;    Lewis,    Ritchie,    &   Bates,    2011; Schommer, “social environment regulates the relationship of                      Nyborg, 2009; Sherkat, 2010). Kanazawa (2010a), Sherkat mental   abilities   and   religious   attitudes   by   channeling   the         (2010), and Lewis et al. (2011) all found negative relations intelligence into certain approved directions: a secular-oriented                between   intelligence   and   religiosity   in   post-college   adults. environment may direct it toward skepticism, a church-ori-                       Nyborg   (2009)   found   that   young   atheists   (age   12   to   17) ented environment may direct it toward increased religious                       scored significantly higher on an intelligence test than reli- interest” (p. 90). They found that in a Catholic college, more                   gious youth. intelligent students knew more about religious doctrine and                         The last decade also saw studies on the relation between participated more in strictly religious organizations. To the                    intelligence and religiosity at the group level. Using data extent   that   such   participation   is   an   indicator   of   religiosity,   from 137 nations, Lynn, Harvey, and Nyborg (2009) found                                               Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 4-----------------------

Zuckerman et al.                                                                                                                                            3 a negative relation between mean intelligence scores (com-                       The Present Investigation puted for each nation) and mean religiosity scores. However, IQ scores from undeveloped and/or non-Westernized coun-                          The purpose of the present investigation was twofold. First, tries   might   have   limited   validity   because   most   tests   were        we aimed to conduct a quantitative assessment of the nature developed for Western cultures. Low levels of literacy and                       and magnitude of the relation between intelligence and religi- problems       in  obtaining     representative      samples      in  some       osity. Embedded in this purpose was also the intent to exam- countries may also undermine the validity of these findings                      ine a tripartite division of research participants—precollege, (Hunt, 2011; Richards, 2002; Volken, 2003). In response to                       college, and non-college (non-college refers to individuals of these critiques, Reeve (2009) repeated the analysis but set                      college age or older who are not in college)—as a moderator all   national   IQ   scores   lower   than   90   to   90. The   resulting      of this association. Francis’s studies (e.g., Francis, 1998) sug- IQ-religiosity correlation was not lower than what had been                      gest   that   in   the   precollege   population,   intelligence   is   only reported in prior studies (see Reeve, 2009, for a discussion                     weakly related to religiosity. Because the college experience of    his  truncating     procedure).      In  the   same    vein,   Pesta,      has many unique characteristics (e.g., first time away from McDaniel,   and   Bertsch   (2010)   found   a   negative   relation             home, exposure to new ideas, higher levels of freedom and between intelligence and religiosity scores that were com-                       independence),        and   because     the  range    of   intelligence    is puted for all 50 states in the United States. These results are                  restricted in the college population, this demographic group less susceptible to the problems (e.g., cultural differences)                    was considered separately. This left the non-college popula- that plagued studies at the country level. Thus, the current                     tion as the third category to be examined. literature suggests that aggregates may also exhibit a nega-                         The second purpose of the investigation was to examine tive relation between intelligence and religiosity. However,                     explanations for any observed associations between intelli- the   reasons   for   relations   at   the   group   level   may   be   quite    gence and religiosity. Most extant explanations (of a nega- different   from   reasons   for   the   same   relations   at   the   indi-     tive    relation)   share   one   central   theme—the        premise    that vidual level.                                                                    religious beliefs are irrational, not anchored in science, not     Finally, parallel to studies on intelligence and religiosity,                testable and, therefore, unappealing to intelligent people who psychologists have also examined a related issue—the prev-                       “know better.” As Bertsch and Pesta (2009) put it, “people alence of religiosity among scientists. This line of research                    who are less able to acquire the capacity for critical thought also started early (Leuba, 1916) and the topic continued to                      may   rely   more   heavily   on   comfortable   belief   systems   that attract attention in more recent years (e.g., Larson & Witham,                   provide uncontested (and uncontestable) answers” (p. 232). 1998).   Studies   in   this   area   have   found   that,   relative   to   the Nyborg (2009) offered a similar view: “High IQ-people are general public, scientists are less likely to believe in God.                    able to curb magical, supernatural thinking and tend to deal For example, Leuba (1916) reported that 58% of randomly                          with the uncertainties of life on a rational-critical-empirical selected scientists in the United States expressed disbelief in,                 basis” (p. 91). Some investigators adopted this approach but, or doubt regarding the existence of God; this proportion rose                    as Hoge (1969) had done earlier, added education as a pos- to nearly 70% for the most eminent scientists. Larson and                        sible    mediating     variable.    Reeve    and    Basalik    (2011),   for Witham (1998) reported similar results, as evidenced by the                      example, suggested that “populations with higher average IQ title of their article—“Leading scientists still reject God.” Of                 are likely to gravitate away from religious social conventions course, higher intelligence is only one of a number of factors                   and   towards   more   rational   .   .   .   systems   conferred   by   the that can account for these results.                                              higher   (average)   educational   achievement   of   that   popula-     Despite the recent uptick of research on the intelligence–                   tion” (p. 65). religiosity connection, we are not aware of any recent schol-                        We   identified   three   other   explanations   of   the   negative arly   reviews   besides   those   listed   hereinbefore.   Outside   of         intelligence–religiosity         association,      offered    by    Argyle academic   journals,   however,   there   have   been   at   least   two         (1958),     Kanazawa       (2010a,     2010b),    and   Sherkat     (2010). reviews (Beckwith, 1986; Bell, 2002). Beckwith (1986) con-                       Argyle (1958) suggested briefly and without elaboration that cluded   that   39   of   the   43   studies   that   he   summarized   sup-     more   intelligent   people   tend   to   rebel   against   conventions, ported a negative relation between intelligence and religiosity,                 including      orthodox     religious    beliefs.   Kanazawa       (2010a, and Bell (2002) simply repeated this tally. However, some of                     2010b) posited that religious beliefs developed early in our the studies reviewed by Beckwith were only indirectly rele-                      ancestral     environment       because     they   were    evolutionarily vant  (e.g., comparisons between more and less prestigious                       adaptive;   atheism,   in   contrast,   is   evolutionarily   novel.   He universities), and some relevant studies were excluded.                          also   proposed   that   intelligence   developed   as   a   capacity   to     In summary, the relation between intelligence and religi-                    cope   more   effectively   with   evolutionarily   novel   problems. osity   has   been   examined   repeatedly,   but   so   far   there   is   no   Given that atheism is evolutionarily novel, it is more likely to clear consensus on the direction and/or the magnitude of this                    be adopted by more intelligent people. Sherkat (2010), focus- association. There is a hint that age might moderate the rela-                   ing on Christian fundamentalism (as opposed to general reli- tionship, but this issue has not been put to test. Finally, there                giosity) and verbal ability (as opposed to general intelligence), is also no consensus on what might explain this relation.                        proposed that fundamentalism has a negative effect on verbal                                           Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 5-----------------------

4                                                                                                    Personality and Social Psychology Review XX(X) ability. The reason is that very conservative Christians scorn                    university entrance exams (UEEs; e.g., SAT, GRE), which are secular   education,   the   search   for   knowledge,   information              highly correlated with standard IQ measures (correlations in from the media, and anything emanating from the scientific                        the   .60-.80   range   are   typical   for   college   students).   Indeed, method.       Furthermore,       conservative      Christians     maintain        these   tests   are   often   viewed   as   measures   of   general   intelli- homogeneous social networks, shun nonadherents, and avoid                         gence (Frey & Detterman, 2004; Koenig, Frey, & Detterman, information from external sources. The overall effect is that                     2008). We also included studies that administered tests of cog- fundamentalist Christian beliefs as well as ties to sectarian                     nitive   abilities   (e.g.,   synonym   tests,   working   memory   tests) denominations have a negative effect on verbal ability.                           that could reasonably serve as proxies for IQ measures.     Religion may indeed be a set of beliefs in the supernatural                       We   also   examined   the   relations   between   school   perfor- (e.g., Nyborg, 2009) that was adaptive in an ancestral envi-                      mance       (grade     point    average,      GPA)      and    religiosity. ronment (Kanazawa, 2010a). However, we believe that reli-                         Intelligence and GPA correlate only moderately (the .25-.40 gion is much more than that and, as such, the interpretations                     range   is   typical   for   college   students;   e.g.,   Feingold,   1983; of its inverse relation with intelligence are more complicated                    Pesta    &   Poznanski,      2008;    Ridgell    &   Lounsbury,      2004). than those offered so far. Specifically, recent theoretical and                   Indeed, Coyle and Pillow (2008; Study 1) reported that cor- empirical work on the role and functions of religion in human                     relations   between   intelligence   and   SAT/ACT   scores   were life (e.g., Sedikides, 2010) allow a new look at the relation                     substantially      higher   than   those    between     intelligence    and between intelligence and religiosity. However, we will take                       GPA. However, while GPA is probably a poor indicator of that look only after we establish the nature of this relation.                    intelligence, it can also be seen as a measure of educational                                                                                   achievement. As noted hereinbefore, some investigators saw                                                                                   education   as   the   mediator   of   the   relation   between   intelli- Method                                                                                   gence and religion, a view that will get some support if GPA Selection of Studies                                                              (viewed as a measure of educational performance) is nega-                                                                                   tively    related   to  religiosity.   Accordingly,      we    planned    to We searched for relevant articles in PsycINFO, using the fol-                     examine        the   relation     between       GPA      and    religiosity lowing   intelligence-related   search   terms:   intelligence   quo-             separately. tient,   IQ,   intelligence,   and  cognitive   ability.   Search   terms             The     religiosity   measures      included     belief   scales    that relating to religiosity were also entered, including religion,                    assessed various themes related to religiosity (e.g., belief in spirituality,     religiosity,   and   religious     beliefs.   A   Google        God     and/or    the  importance      of   church).    In  addition,    we Scholar search was conducted for articles that contained the                      included studies that measured frequency of religious behav- word religion and either IQ or intelligence. In addition, arti-                   iors   (e.g.,   church   attendance,   prayer),   participation   in   reli- cles from the Journal for the Scientific Study of Religion and                    gious organizations, and membership in denominations. Review      of  Religious     Research     in   years   not   indexed     by          Table 1 presents the studies that were analyzed. If results PsycINFO were inspected one-by-one. The Archive for the                           for more than one independent sample of participants were Psychology of Religion, which is not covered by PsycINFO,                         reported in a single article, they were considered as separate was   also   reviewed.   Finally,   reference   lists   of   studies   that       studies in the analysis. Altogether there were 63 studies from were identified by any of the aforementioned methods were                         52   sources.   Articles   from   which   data   were   extracted   are searched for additional relevant studies.                                         marked by an asterisk in the Reference section. Table 1 pres-     Studies were included in the meta-analysis if they exam-                      ents a number of characteristics for each study; these will be ined the relation between intelligence and religiosity at the                     explained in greater detail in the following section. Finally, individual level, and if the effect size (Pearson r) of that rela-                Table 1 presents an effect size for each study—the zero-order tion was provided directly or could be computed from other                        correlation between intelligence and religiosity. statistics.   For   several    studies,   intelligence    and   religiosity were   measured,   but   the   authors   did   not   report   the   relation between these two variables. Authors of such studies were                         Data Extraction and Coding contacted to obtain the relevant information. If authors did                      The first author extracted an effect size r for each study; a not respond to our first request, two more reminders were                         negative   correlation   indicated   that   higher   intelligence   was sent.   When   necessary,   second   and/or   third   coauthors   were            associated with lower religiosity. When several correlations also contacted. Studies that examined the relation between                        were available due to the use of multiple religiosity and/or intelligence   and   religiosity   indirectly   (e.g.,   comparisons   at         intelligence     measures,     the   average    correlation     was   com- group levels, comparisons between scientists and the general                      puted. However, the separate correlations were retained for population) were excluded.                                                        moderation   analyses   if   each   correlation   corresponded   to   a     Studies included in the present meta-analysis used a variety                  different level of a moderator (see the following for details). of intelligence and religiosity measures. Most of the intelli-                    The   first   author   also   coded   all   of   study   attributes   that   are gence   tests   are   widely   used   (e.g.,   Wechsler   tests,   Peabody        described in the following. The third author recomputed all Picture   Vocabulary   Test,   etc.).   A   subgroup   of   studies   used        effect sizes and recoded all study attributes. Only one coding                                                Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 6-----------------------

                Table 1.  Overview of Studies Included in the Meta-Analysis.                                                                                       Proportion                                                                Religiosity          Number of items in                                                          Effect size                 Study                                                 Total n           of males                     Intelligence measure                        measure             religiosity measure               Sample                   Bias                 (r)                 Bender (1968)                                              96               1.0           UEE and GPA                                          Attendance                        1                Non-college            Time gap                  −.10a                 Bertsch and Pesta (2009)                                  278              0.42           Wonderlic Personnel Test                                Beliefs                      >2                 College                     n/a                  −.15                 Blanchard-Fields, Hertzog, Stein, and Pak (2001); C. Hertzog, personal communications, October 2011:                    Study 1                                                 96              0.60           Shipley Vocabulary Test                                 Mixed                        >2                 College                     n/a                    .00b                    Study 2                                                219              0.42           Shipley Vocabulary Test                                 Mixed                        >2                 Non-college                 n/a                  −.32c                 Bloodgood, Turnley, and Mudrack                           230              0.63           UEE                                                  Attendance                        1                College                     n/a                  −.15                   (2008) D o w n               D. G. Brown and Lowe (1951)                               108              n/a            UEE                                                     Beliefs                      >2                 College                Extreme                   −.43 l o a d                                                                                                                                                                                                                                          groups e d   f               Carlson (1934)                                            100              n/a            UEE                                                     Beliefs                      >2                 College                Time gap                  −.19 r o m                                                                                                                                                                                                                                                d   p               Carothers, Borkowski, Burke                               101                 0           WAIS-R Vocabulary and Block                          Attendance                      >2                 Non-college                 n/a                  −.25 s r . s                 Lefever, and Whitman (2005); S. S.                                                         design a g e                 Carothers, personal communication, p u b c.                September 2011 o m                 Ciesielski-Kaiser (2005)                                  216              0.36           Shipley Institute for Living Scale                      Beliefs                      >2                 College                     n/a                  −.14 a t   F U               Corey (1940)                                              234              n/a            UEE                                                     Beliefs                      >2                 College                     n/a                  −.03 N D                                                                                                                                                                                                                                                                          e                 Cottone, Drucker, and Javier (2007)                       123              0.35           WAIS III comprehension and                              Beliefs                      >2                 College                     n/a                  −.14 C O O                                                                                                              similarities and GPA R                                                                                                                                      f D               Crossman (2001)                                            75                 0           Immediate free recall                                   Beliefs                      >2                 Non-college                 n/a                  −.36 E   A P               Deptula, Henry, Shoeny, and Slavick                   11,963               n/a            Modified Peabody Picture                                Beliefs                        1                Precollege                  n/a                  −.10 R F O                 (2006)                                                                                    Vocabulary Test   P E               Dodrill (1976)                                             44              0.54           WAIS                                                    Beliefs                      >2                 Non-college            Extreme                     .05 S S L                                                                                                                                                                                                                                            groups N I V                                                                                                                                                                                                                                                g E               Dreger (1952)                                              60              0.50           Wonderlic Personnel Test                                Beliefs                      >2                 Non-college                 n/a                  −.13   o n                 Feather (1964)                                            165               1.0           Syllogisms                                              Beliefs                      >2                 College                     n/a                  −.16 A u g u               Feather (1967)                                             40              0.50           Syllogisms                                          Membership                         1                College                     n/a                  −.09 s t    1  2              Foy (1975)                                                 36              0.50           WAIS                                                    Beliefs                      >2                 Non-college                 n/a                  −.50 ,    2                                                                                                                                                                                                                                                                       h  0  1              Francis (1979)                                         2,272               n/a            IQ from school records                                  Mixed                        >2                 Precollege                  n/a                    .04  3                                                                                                                                                                                                                                                                       i                 Francis (1998); Francis (1997)                            711              0.40           Raven Progressive Matrices                              Mixed                 Both 1 and >2             Precollege                  n/a                  −.04                 Francis, Pearson, and Stubbs (1985)                       290              0.72           IQ (not specified)                                      Beliefs                      >2                 Precollege                  n/a                  −.13                 Franzblau (1934)                                          354              0.44           Terman Test of Mental Abilities                         Beliefs                      >2                 Precollege                  n/a                  −.15                 Gilliland (1940)                                          326              n/a            Not specified                                           Beliefs                      >2                 College                     n/a                    .00                 Gragg (1942)                                              100              0.50           UEE                                                     Beliefs                      >2                 College                     n/a                  −.02                 Hadden (1963)                                             261              n/a            GPA                                                     Mixed                        >2                 College                     n/a                  −.06                 Hoge (1969)                    Study 1                                                179              n/a            UEE                                                     Mixed                        >2                 College                     n/a                  −.12j                    Study 2                                                135              n/a            UEE                                                     Mixed                        >2                 College                     n/a                  −.08j                    Study 3                                                327              n/a            UEE                                                     Mixed                        >2                 College                     n/a                  −.07j                 Horowitz and Garber (2003)                                172              0.46           WISC Vocabulary and Block design                        Mixed                          1                Precollege             Time gap                    .05k                                                                                                                                                                                                                                                                  (continued)       5
----------------------- Page 7-----------------------

       6                          Table 1. (continued)                                                                                         Proportion                                                                  Religiosity          Number of items in                                                            Effect size                 Study                                                   Total n           of males                      Intelligence measure                         measure              religiosity measure               Sample                    Bias                  (r)                 Howells (1928)                    Study 1                                                  461             0.43             Thorndike Intelligence Test, Iowa                        Beliefs                       >2                  College                     n/a                   −.25l                                                                                                                Comprehension Test, and GPA                    Study 2                                                   n/a              n/a            Not specified                                            Beliefs                       >2                  College                     n/a                   −.29                 Inzlicht, McGregor, Hirsh, and Nash                          22             0.41             Wonderlic Personnel Test                                 Beliefs                         1                 College                     n/a                   −.13                   (2009; Study 2)                 V. Jones (1938)                                             268               n/a            UEE                                                      Beliefs                       >2                  College                     n/a                   −.24 D o w               Kanazawa (2010a); S. Kanazawa, personal communications, 2011: n l o a                  Study 1                                              14,277              0.47             Peabody Picture Vocabulary Test                          Beliefs                         1                 Non-college            Time gap                   −.12 d e d  f                 Study 2                                               7,160              0.44             Verbal synonyms                                          Beliefs                         1                 Non-college                 n/a                   −.14 r o m                 Kosa and Schommer (1961)                                    361               1.0            Assorted tests and GPA                                Membership                       >2                  College                     n/a                     .09m p s r                                                                                                                                                                                                                                                                                n . s               Lewis, Ritchie, and Bates (2011)                         2,155                n/a            Assorted tests                                            Mixed                        >2                  Non-college                 n/a                   −.16 a g                                                                              o e p               McCullough, Enders, Brion, and Jain                       951                 n/a            Stanford–Binet                                           Beliefs                       >2                  Non-college            Extreme                    −.45 u b . c                 (2005)                                                                                                                                                                                                                         groups o m a               Nokelainen and Tirri (2010);                                 20             0.45             WAIS III                                                 Beliefs                       >2                  Precollege             Attenuated                 −.20 t   F U                  P. Nokelainen, personal                                                                                                                                                                                                        range N D                 communication, December 2011   C O O               Nyborg (2009)                                            3,742                n/a            Assorted tests                                        Membership                         1                 Precollege                  n/a                   −.05 R   D               Pennycook, Cheyne, Seli, Koehler, and E   A P                  Fugelsang (2012) R F                                                                                                                                                                                                                                                                                p O                  Study 1                                                  223               .41            Assorted tests                                            Mixed                        >2                  Non-college                 n/a                   −.19   P                                                                                                                                                                                                                                                                              q E                  Study 2                                                  267               .22            Assorted tests                                            Mixed                        >2                  Non-college                 n/a                   −.17 S S L                 Poythress (1975)                                            195               n/a            UEE                                                      Beliefs                       >2                  College                     n/a                   −.19 N I V E               Räsänen, Tirri, and Nokelainen (2006)                       142               n/a            Assorted tests                                           Beliefs                       >2                  precollege                  n/a                   −.17   o n  A              Salter and Routledge (1974) u g u                  Study 1                                                  339               n/a            UEE                                                      Beliefs                         1                 College                     n/a                   −.15 s t    1  2                 Study 2                                                  241               n/a            UEE                                                      Beliefs                         1                 College                     n/a                   −.18 ,    2  0  1              Saroglou and Fiasse (2003)                                  120             0.56             GPA                                                      Beliefs                         1                 College                     n/a                     .07  3                 Saroglou and Scariot (2002;                                  94             0.41             GPA                                                       Mixed                        >2                  Precollege                  n/a                     .13                   Study 2); V. Saroglou, personal                   communications, March 2012                 Shenhav, Rand, and Greene (2011;                            306             0.35             Shipley Vocabulary Test, and WAIS                        Beliefs                       >2                  College                     n/a                   −.06                   Study 2)                                                                                     III Matrix Reasoning Test                 Sherkat (2010); D. E. Sherkat, personal                 12,994              0.43             Vocabulary Test                                           Mixed                          1                 Non-college                 n/a                   −.15r                   communications, October 2011                 Sherkat (2011)                                            1,780               n/a            Scientific Literacy Scale from                           Beliefs                         1                 Non-college                 n/a                   −.34                                                                                                                General Social Survey                                                                                                                                                                                                                                                                         (continued)
----------------------- Page 8-----------------------

               Table 1. (continued)                                                                                     Proportion                                                             Religiosity          Number of items in                                                        Effect size                Study                                                Total n          of males                    Intelligence measure                       measure             religiosity measure              Sample                  Bias                 (r)                Sinclair (1928)                                           67            0.48            UEE                                                    Beliefs                     >2                 College               Extreme                 − .44                                                                                                                                                                                                                                      groups                Southern and Plant (1968)                                 72              .58           Mensa membership                                       Beliefs                     >2                 Non-college           Extreme                  −.75                                                                                                                                                                                                                                      groups                Stanovich and West (2007); R. F. West, personal communication, October 2011:                   Study 1                                              439             0.24            UEE                                                    Beliefs                      1                 College                   n/a                  −.24                   Study 2                                             1,045            0.31            UEE                                                    Beliefs                      1                 College                   n/a                  −.18 D o w              Symington (1935) n l o a                 Study 1                                              200               n/a           Otis Test of Mental Ability                            Beliefs                     >2                 College               Attenuated               −.24 d e d  f                                                                                                                                                                                                                                   range r o m                 Study 2                                               160              n/a           Otis Test of Mental Ability                            Beliefs                     >2                 College               Attenuated               −.47 p s r s.                                                                                                                                                                                                                                   range a g e              Szobot et al. (2007); C. M. Szobot,                     236               1.0           WAIS III block design and                          Attendance                       1                 Precollege                n/a                    .15 p u b .                personal communication, December                                                        Vocabulary c o m                2011   a t   F              Turner (1980) U N D                 Study 1                                              200               1.0           Thurstone Primary Mental Abilities                     Beliefs                     >2                 Precollege                n/a                  −.04 C O                                                                                                        Scale O R   D                 Study 2                                              200               1.0           Thurstone Primary Mental Abilities                     Beliefs                     >2                 Precollege                n/a                  −.02 E   A                                                                                                        Scale P R F              Verhage (1964)                                         1,538              n/a           Groninger Intelligence Test                        Membership                       1                 Non-college               n/a                  −.12 O   P E              Young, Dustin, and Holtzman (1966) S S L                 Study 1                                              481             0.69            GPA                                                    Beliefs                     >2                 College                   n/a                    .03 N I V E                 Study 2                                               574            0.57            GPA                                                    Beliefs                     >2                 College                   n/a                  −.11   o n   A u g              Note. UEE     University Entrance Exams; GPA       grade point average; n/a     not applicable; WAIS     Wechsler Adult Intelligence Scale; WISC        Wechsler Intelligence Scale for Children. u              a s t              Average of .00 correlation with GPA and −.20 correlation with the intelligence measure.  1  2                b ,               Average of .04 correlation with religious beliefs and −.03 correlation with religious behavior.  2  0                c  1             Average of −.33 correlation with religious beliefs and −.30 correlation with religious behavior.  3             dThe procedure of the study included “time gap” and “extreme groups” (for further explanation, see bias section under Data Extraction and Coding below); these features were expected to decrease and increase,                respectively, the effect size, resulting in zero net bias.                eAverage of −.15 correlation with GPA and −.14 with the intelligence measure.                f The Immediate Free Recall, a component of Gudjonnson’s (1987) suggestibility scale, correlated .69 with the WAIS-R (Gudjonsson & Clare, 1995).                g Using extreme groups in religiosity but eliminating participants with low intelligent scores or inconsistent responses to the religiosity scales were assumed to cancel one another.                hAverage of .03 correlation with religious belief and .05 correlation with religious behavior.                iThe −.04 effect size is the average of −.02 correlation with a religious beliefs scale and −.02 and −.08 correlations with two one-item religious behavior questions.                j The three effect sizes in Hoge’s (1969) studies are each an average of a correlation with religious beliefs and a correlation with religious behavior: −.08 and −.17, −.13 and –.04, and −.08 and −.06, respectively.                kAverage of −.05 correlation with religious beliefs and .15 correlation with religious behavior.                lAverage of −.15 correlation with GPA and −.36 with the intelligence measure.                mAverage of .00 correlation with GPA and .18 correlation with the intelligence measure.                nAverage of −.10 and −.25 correlations with two religious beliefs measures and −.14 and −.15 correlations with two religious behavior measures.                oAverage of Ns of several comparisons between Terman’s sample and the general population (see Table 9).                pAverage of −.20 correlation with religious beliefs and −.18 correlation with a mixed measure of religious belief and behavior.                qAverage of −.18 correlation with religious beliefs and −.16 correlation with a mixed measure of religious belief and behavior.                rAverage of −.29 correlation with religious beliefs and −.01 correlation with religious behavior.       7
----------------------- Page 9-----------------------

8                                                                                                   Personality and Social Psychology Review XX(X) variable (goal of study) involved a subjective judgment, and                     Americans, and mixed or not available. However, for reli- the two discrepancies for this variable were resolved by dis-                    gion and race, the resulting distributions were too skewed cussion.   Discrepancies   in   either   the   computation   of   effect         to allow meaningful analysis. sizes or the coding of study attributes indicated mistakes and were corrected.                                                                  Bias.  We coded studies as biased if their methodology could                                                                                  artificially   attenuate    or  inflate   the  intelligence–religiosity Gender.  We coded the percentage of study participants who                       correlation.   There   were   two   potential   causes   of   bias   that were males, for studies that provided the gender distribution.                   could   attenuate   correlations:   restriction   of   range   and   time                                                                                  gap between measurements. Restricted range studies exam- Intelligence     measures.  For     each    study,    we   coded    which        ined samples that were limited in their range of either religi- intelligence test was used; a separate coding category was                       osity   (e.g.,   because   only   very   religious   participants   were included   for   GPA.   However,   with   the   exception   of   GPA             included)   or   intelligence   (e.g.,   because   only   highly   intelli- and UEE, the number of studies associated with a single                          gent participants were included). Note that we used this cod- intelligence measure was too small (ranging from one to                          ing   in   addition   to   the   aforementioned   college   category   in four,   see Table   1)   to   allow   a   meaningful   analysis   of   this      which all studies were assumed to be restricted in their range variable.                                                                        of intelligence scores. In time gap studies, researchers admin-                                                                                  istered the intelligence measure some time (usually a number Religiosity measures.  We coded whether the religiosity mea-                     of years) before the religiosity measure. There were no time sure involved beliefs, frequency of church attendance and/or                     gap    studies    in   which    religiosity    was    measured      before prayer,   or   participation/membership   in   religious   organiza-             intelligence. tions. A “mixed” category included studies that reported cor-                       A third bias category—extreme groups—was comprised relations    for   more    than   one   type   of  religiosity   measure.        of studies in which investigators compared participants very Although       measures     of   beliefs   were    heterogeneous       with      high in intelligence (or religiosity) with participants very low respect to the focus of the belief (e.g., belief in God, belief in               in intelligence (or religiosity). This design was expected to scriptures, beliefs in spirits), there were not enough studies to                inflate intelligence−religiosity correlations. The fourth cate- allow a more detailed classification. We also coded the num-                     gory included all remaining studies. ber of items in the religiosity measures, expecting measures with more items to be more reliable. However, this variable                      Published or unpublished.  We coded whether or not the study did not produce any results of interest.                                         was published. However, this variable did not influence the                                                                                  results. Goal    of   study.  We    coded    whether     assessing    the   relation between intelligence and religiosity was the main goal of the                                                                                  Analysis study, one of several goals, or not a goal at all.                                                                                  Random-effects and fixed-effects analyses were performed. Sample type.  Studies were classified as investigating precol-                   Random-effects models produce results that can be general- lege, college, or non-college samples, as defined hereinbe-                      ized to future studies not having designs that are identical to fore. Note that this variable is related to age and education.                   those of studies included in the meta-analysis; fixed-effects The precollege participants were almost exclusively between                      models limit generalization to new participants in the meta- 12   and   18   years   of   age.   Only   one   study   in   this   category    analyzed study designs (Hedges & Vevea, 1998; Rosenthal, (Francis, 1979) included participants younger than 12. Col-                      1995; Schmidt, Oh, & Hayes, 2009). Fixed-effects models lege participants were undergraduates and, very infrequently,                    give   weight   to   studies   proportional   to   their   sample   sizes; a   mixture     of  undergraduates       and   graduate    students.    We       these models can therefore be misleading when methodolog- assumed that intelligence scores of college participants were                    ical   features   of   larger   studies   differ   from   those   of   smaller restricted in range relative to those of the general population.                 ones. On the other hand, random-effects models can be lack- Non-college participants were recruited outside of academic                      ing in statistical power. contexts and tended to be older than participants in the col-                       In the present meta-analysis, there was a sufficient num- lege group.                                                                      ber of studies to permit meaningful random-effects analyses                                                                                  of central tendency and moderators, using the PASW 18 sta- Religion    and    race.  Participants’   religions     were    coded    as      tistical package (e.g., correlation, analysis of variance). All Protestant,   Catholic,   “Christian”   (a   term   that   often   went          random-effects       analyses    used    unweighted      effect  sizes   as undifferentiated in the studies), Jewish, or unspecified. For                    dependent   variables   and   studies   as   the   units   of   analysis. each religion, a study was coded as “all” (90% or more) or                       Fixed-effects      analyses    of  central   tendency,     homogeneity, “mostly”   (more   than   50%). There   were   no   studies   in   the           publication bias, and moderators were performed using the “mostly Jewish” category. We coded race according to four                        Comprehensive             Meta-Analysis          software         package categories:      Mostly     Caucasians,      all  Caucasians,     African        (Borenstein, Hedges, Higgins, & Rothstein, 2005).                                               Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 10-----------------------

Zuckerman et al.                                                                                                                                                    9 Table 2.  Correlations Between Religiosity and Intelligence.                                                 Random-effects results                                            Fixed-effects results                                         Unweighted                 t test against   Weighted                                          Heterogeneity          File Studies                                  M r (SD)       Median           0              M r           95% CI        Combined Z           chi-square       drawer ka All (k   63)                             −.16 (.18)       −.14       −6.83***          −.13b        [−.14, −.12]      −34.36***          554.00***          10,129 No GPA (k      58)                       −.18 (.19)       −.15       −7.16***          −.13c        [−.14, −.12]      −34.70***          566.99***          10,240 No GPA and no extreme groups             −.15 (.14)       −.15       −7.77***          −.12d        [−.13, −.12]      −32.70***          378.98***           7,250 (k   53) Note. GPA    grade point average; CI  confidence interval. aTwo-tail test. bk   62 because sample size was missing for one study. The effect size for that study was −.29. ck   57 (see Footnote a). dk   52 (see Footnote a). ***p < .001. Results                                                                               omitting   four   studies   that   used   both   kinds   of   measures.   A                                                                                      random-effects (unweighted effect sizes) ANOVA showed a Overall Relation of Intelligence to Religiosity                                       significant difference, F(1, 57)            5.22, p  <  .05 (M            .01,                                                                                                                                                         GPA The first row of Table 2 presents basic statistics describing                        Mother   tests    −.18).   A   fixed-effects   (weighted   effect   sizes) the   relation   between   intelligence   and   religiosity   for   all   63          comparison   was   also   significant, p         <  .001   (MGPA         −.03, studies.   Results   are   presented   for   random-effects   analyses               Mother tests    −.13). When GPA–religiosity correlations from (unweighted   mean   correlations)   and   fixed-effects   analyses                  the five studies using only GPA are combined with GPA– (weighted       mean     correlations).     Fifty-three     studies    showed        religiosity correlations from the four studies using GPA as negative correlations while 10 studies showed positive cor-                          well as other intelligence measures, the mean GPA–religios- relations.   Thirty-seven   studies   showed   significant   correla-                ity    correlation     was   not   significantly     different    from    zero, tions; of these, 35 were negative and 2 were positive. The                           MGPA       −.027, p        .33. It was concluded that GPA had no unweighted   mean   correlation   (r)   between   intelligence   and                 meaningful relation to religiosity and, accordingly, all subse- religiosity was −.16, the median r was −.14, and the weighted                         quent analyses omitted the five studies that used only GPA. mean r was −.13. The similarity of these three indicators of                         For   the   four   studies   that   used   GPA   and   other   intelligence central tendency indicates that the distribution was approxi-                        tests,   we   used   only   their   non-GPA   results   in   subsequent mately   symmetrical   and   was   not   skewed   by   several   very                 analyses. large studies that were in the database. Random- and fixed-                              After excluding all findings for GPA, 58 studies remained effects models yielded significant evidence that the higher a                         for   analysis.    The    effect    size   for  these    non-GPA       studies person’s intelligence, the lower the person scored on the reli-                       (shown in the second row of Table 2) was more negative than giosity measures.                                                                    that of the full data set in the random-effects analysis but did     The     distribution     of  intelligence–religiosity        correlations        not change in the fixed-effects analysis. was highly heterogeneous as indicated by the significant chi- square statistic (Table 2). The file drawer calculation indi-                         Statistical artifacts. As noted hereinbefore, two methodologi- cated that 10,129 studies with average effects of r                  0 would          cal   features   were   considered   likely   to   attenuate   the   intelli- have to be added to nullify the two-tailed test of the com-                           gence–religiosity        relation—restriction         of   range     and    the bined probability. Application of a fixed-effects trim and fill                      presence of a time gap between measurements. A third meth- procedure (Duval & Tweedie, 2000) for detecting gaps in the                           odological feature—extreme groups—was expected to inflate distribution of effect sizes suggested that two positive effects                     the intelligence–religiosity relation. Any effects that distorted would have to be added to make the distribution symmetri-                            the “true” intelligence–religiosity relation should be removed. cal;   even   with   these   imputed   studies   added,   however,   the                 In   an ANOVA  of   the   unweighted   effect   sizes   (random- overall estimated effect size did not change.                                         effects model), the following groups of studies were com-                                                                                      pared: restricted range studies (M                −.31, k      3), time gap                                                                                       studies (M       −.12, k     4), extreme groups studies (M              −.43, Biases in the Intelligence–Religiosity Relation                                                                                      k     5), and the remaining studies (M                 −.14, k      46). The GPA.  As   stated   hereinbefore,   GPA   is   a   poorer   measure   of              omnibus F was significant, F(3, 54)               7.06, p  < .001. intelligence than are standardized cognitive tests. To exam-                              Surprisingly, studies with restriction of range yielded an ine whether GPA had a weaker correlation with religiosity                             effect size (M       −.31) that was more negative than that of the than did other measures of intelligence, studies using GPA                           46     studies   with    no   evident     source    of   bias   (M       −.14). (k    5) were compared with studies using other tests (k                   54),       Therefore,   range-restricted   studies   were   retained.   Because                                             Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 11-----------------------

 10                                                                                                      Personality and Social Psychology Review XX(X) Table 3.  Sample Type as a Moderator of the Religiosity–Intelligence Relation (k                   53).                                       Random-effects results                                                 Fixed-effects results                               Unweighted                         t test      Weighted                           Combined          Heterogeneity             File Sample type                     M r (SD)         Median        against 0         M r            95% CI               Z              chi-square          drawer k Precollege (k      12)          −.06 (.10)         −.05       −1.86†            −.07          [−.08, −.06]        −9.97***            60.54***              123 College (k      27)             −.14 (.13)         −.15       −5.59***          −.15a         [−.17, −.12]       −12.00***           118.72***              775 Non-college (k       14)        −.23 (.13)         −.18       −6.92***          −.15          [−.16, −.14]       −30.25***           110.10***            2,245 Note. CI    confidence interval. ak   26 (see Note a, Table 2). †p < .10. ***p < .001. time gap studies yielded an effect size (M                  −.12) close in           Type     of  religiosity   measure.  Type       of   religiosity    measure magnitude to the no-bias studies, they also were left in. As                         (behavior, beliefs, group membership, or a combination of expected, the extreme groups effect size (M                 −.43) that was           measure   types)   had   a   marginally   significant   effect   in   the significantly more negative than that of the unbiased studies                       unweighted (random-effects) analysis, F(3, 49)                     2.31, p  <  (p < .001 by post hoc least significant difference [LSD] test).                      .09 (M             −.06, k     5; M           −.18, k     33; M                                                                                              behavior                     beliefs                     group member- The     five    studies    with    extreme      groups     were     therefore                −.09,  k      3;   and M                            −.10,  k      12).                                                                                      ship                             combination   of   measures excluded, leaving 53 studies for further analysis. The overall                       Religious   behavior   and   membership   in   religious   organiza- results   for   these   53   studies   are   shown   in   the   third   row   of     tions are conceptually similar in that both can be motivated Table 2; the effect size changed very little from the corre-                        by either religious beliefs or by extrinsic reasons. It is not sponding effect size for all 63 studies.                                             surprising,     then,   that   the   difference     in  mean     effect   size                                                                                     between these two groups was not significant, t                  .26. On the                                                                                      other hand, the comparison of these two groups of studies Moderators of the Intelligence–Religiosity Relation                                                                                     with   the   studies   measuring   beliefs   was   significant,  t(39) Sample type.  We expected that the relation between intelli-                         2.15, p  < .05. gence and religiosity would differ by sample type (precol-                              Omitting   studies   that   used   combination   of   measures,   a lege,   college,   and   non-college).   As   shown   in   Table   3,   the          fixed-effects analysis also showed that the effect for belief unweighted and the weighted effect sizes were highest at the                         measures was stronger than the effect for behavior or group non-college level, intermediate at the college level, and low-                       membership measures, p            <  .001 (M             −.14, M                                                                                                                                     beliefs              behavior est at the precollege level. An ANOVA on the unweighted                              .02, M                     −.07). Given these results, studies mea-                                                                                              group membership effect sizes (random-effects approach) was significant, F(2,                         suring   behavior   and   studies   measuring   membership   were 50)     6.41, p  < .01; post hoc LSD tests showed that all pair-                     combined into one enlarged behavior category. wise comparisons among these means were significant at p ≤                              For the 10 studies using a combination of measures for .05. The fixed-effects analysis also showed a highly signifi-                       which separate belief and behavior effect sizes could be cal- cant overall between-groups effect,p < .001. However, when                           culated, a random-effects analysis was conducted that com- weighted   by   sample   size,   the   non-college   group   no   longer            pared measure type within studies. The difference between showed   the   most   negative   correlation.   Table   3   also   shows            beliefs and behavior was in the same direction as the differ- that the effects were significantly below zero and were het-                         ence found in the between-studies analysis, though due to erogeneous for all three sample types.                                               the small number of studies it was not significant, matched     The fixed-effects trim and fill method for detecting pos-                        t(9)    1.32, p     .22 (M            −.11, M               −.06).                                                                                                                   beliefs            behavior sible publication bias yielded negligible impact for the pre-                           Given that the intelligence–religiosity relation differed by college     and    non-college      groups.     For   the   college    group,        sample   type   and   the   beliefs/behavior   distinction,   it   was   of however, there was evidence of publication bias, such that                           interest to examine their combined effects. Excluding the 10 nine negative effect sizes would need to be added to yield a                         studies   that   used   a   combination   of   beliefs/behavior   mea- symmetrical   distribution.   The   imputation   of   these   effects                sures, the remaining 43 studies were examined in a 2 (beliefs/ resulted in an adjusted mean effect of −.21, noticeably quite                       behavior)   ×   3   (sample   type) ANOVA.   The   results   showed different from the observed weighted mean effect of −.15.                            significant effects for the beliefs/behavior factor, F(1, 37) Because the adjusted effect size is hypothetical, it will not be                     6.44, p   <  .02,   and   sample   type, F(2,   37)         4.56, p   <  .02. incorporated into subsequent analyses. However, this result                          Importantly, the interaction was not significant, F                 .42. and the range restriction in intelligence scores in this group                          To get as accurate picture as possible of the results, rather suggest that the true intelligence–religiosity relation in the                       than presenting means only for the 43 studies that measured college population may be more negative than the literature                          either beliefs or behaviors, we looked at all available data. indicates. We return to this issue below.                                            Accordingly,   we   added   the   10   studies   that   provided   effect                                                  Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 12-----------------------

Zuckerman et al.                                                                                                                                               11 Table 4.  The Intelligence–Religiosity Relation by Sample Type and the Belief/Behavior Distinction (k                     63).                                             Unweighted mean correlations                                              Weighted mean correlations Sample                                Behavior                                Beliefs                            Behavior                            Beliefs Precollege                              .05 (5)                              −.08 (10)                             −.01                               −.08 College                                −.05 (7)                              −.16 (24)                             −.02                               −.17a Non-college                            −.18 (6)                              −.25 (11)                             −.04                               −.20 ak   23 (see Note a, Table 2). sizes for beliefs and behaviors, displaying each study twice                        of gender as a moderator of the intelligence–religiosity rela- (once   in   the   beliefs   column   and   once   in   the   behavior   col-       tion remains a topic for future research. umn). Table 4 presents unweighted and weighted mean cor- relations   for   the   63   data   entries.   These   mean   correlations          Goal of the study.  Study goal (i.e., whether investigating the were extremely similar to the corresponding mean correla-                           intelligence–religiosity   relation   was   the   main   goal   of   the tions observed for only the 43 data entries.                                        study, one of several goals, or not a goal) was not related to     Note that data entries in the Behavior column in Table 4                        effect   sizes   in   a   random-effects   ANOVA   (p        <  .34).   In   a (k    18) represent independent studies as do data entries in                       fixed-effects   analysis,   the   between-groups   effect   was   sig- the Beliefs column (k           45). The nonindependence involves                   nificant, p  <  .001 (M                −.09, M                         −.17,                                                                                                                main goal              one of several goals the fact that 10 studies appear in both columns because they                        M             −.14). Thus, the negative relation between intel-                                                                                       not a goal had behavior and belief effects. The differences among the                          ligence   and   religiosity   was   more   negative   for   studies   in three sample groups within each column were significant in                          which this relation was not the main question of interest. random-   and   fixed-effects   analyses   (p        <  .05).   In   absolute           Having estimated the overall intelligence–religiosity rela- terms, the mean correlations in the Behavior columns were                           tion, and having tested a number of moderators of this rela- weak, particularly when means were unweighted. Because                              tion,   we   now   proceed   to   ancillary   analyses.   We   begin   by of nonindependence, this enlarged data set of k                  63 results         correcting r values for range restriction, converting r values was not used in any subsequent analysis (unless otherwise                           to Cohen’s d scores, and using these d scores to estimate IQ noted).                                                                             differences   between   believers   and   nonbelievers.   We   then                                                                                     examine whether the observed relation between intelligence Percentage of male participants.  As an exploratory analysis,                       and religiosity might be accounted for by a number of “third we   examined   the   relation   between   percentage   of   males   in             variables.”   Finally,   we   examine   evidence   that   might   shed each study and effect size of the intelligence–religiosity rela-                    light   on   the   causal   direction   of   the   intelligence–religiosity tion. In the 34 studies in which it could be determined, per-                       relation. centage of males was positively correlated with unweighted effect sizes, r(32)        .50, p  <  .01. This correlation indicates                                                                                     Effect Size of the Intelligence–Religiosity Relation: that   the   negative   intelligence–religiosity   relation   was   less negative   in   studies   with   more   males. This   relation   held   in          r, Corrected r, Cohen’s d, and IQ Points terms   of   magnitude   for   the   precollege   and   college   groups,           Clearly,     the   size   of   the   intelligence–religiosity        relation r(6)     .48, ns, and r(12)       .51, p     .06, but was weaker at the             depends on sample group and type of religiosity measure. In non-college level, r(10)          .19, ns. When analyzed as a fixed-                the precollege group, the best estimate of the size of this rela- effects regression, the relation between percentage of males                        tion is r    −.08; this effect size was obtained in random- and and effect size was also markedly positive, p  < .001.                              fixed-effects   analyses   of   studies   with   religiosity   measures     A more direct test of the possibility that the intelligence–                    that assessed religious beliefs (see Table 4). religiosity relation is less negative for males is a within-study                       For   the   college   group,   the   effect   sizes   presented   so   far comparison   between   males   and   females.   Kanazawa1                con-       were not corrected for range restriction of intelligence scores. ducted this test for two studies (Kanazawa, 2010a; combined                         Below, we present the uncorrected and the corrected rs for N     21,437). If anything, the results pointed in the opposite                     this   group.   Computation   of   the   corrected  rs   was   based   on direction.      The     intelligence–religiosity         correlations      for      Thorndike’s (1949) Case 2 formula, which requires use of females and males, respectively, were −.11 and −.12 in Study                        the ratio between the unrestricted and restricted SDs of intel- 1,   and   −.14   and   −.16   in   Study   2. Although   the   difference          ligence     scores.2    Sackett,    Kuncel,     Arneson,      Cooper,     and between females and males was not significant, even when                            Waters (2009; P. R. Sackett, personal communication, May combined   meta-analytically   across   studies   (Z              1.39, p           2012) calculated a 1/.67 ratio between SDs of SAT scores for .16), the direction of this difference is inconsistent with the                     students who applied to but did not attend college, and stu- between-studies finding of the meta-analysis. Thus, the issue                       dents who applied to and attended college. These estimates                                            Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 13-----------------------

 12                                                                                                  Personality and Social Psychology Review XX(X) Table 5.  Effect Size of the Relation Between Intelligence and Religiosity for Selected Groups.                                                               College uncorrected                College corrected                     Non-college Effect size                                               Unweighted        Weighted        Unweighted         Weighted        Unweighted        Weighted All studies    r                                                          −.14              −.15             −.21             −.22             −.23             −.15    d                                                          −.28              −.30             −.43             −.45             −.47             −.30    IQ points                                                  4.2               4.5              6.4              6.8              7.1              4.5 Studies with religiosity measure assessing religious beliefs    r                                                          −.16              −.17             −.24             −.25             −.25             −.20    d                                                          −.32              −.34             −.49             −.52             −.52             −.41    IQ points                                                  4.8               5.1              7.4              7.8              7.8              6.2 pertain to three cohorts (1995-1997) of students who applied                      non-college level, the corrected college effect sizes are simi- to   41   colleges   and   universities   in   the   United   States.   The       lar in the random-effects analysis and somewhat larger in the schools were diverse in location and size, and included pri-                      fixed-effects analysis. A cautious conclusion is that the two vate and public institutions. This ratio is conservative in that                  populations do not differ in the degree to which intelligence it is based on the population of SAT test takers rather than the                  and religiosity are negatively related.5 entire population.                                                                   Random- and fixed-effects analyses produced more nega-     One of the studies included in the present meta-analysis                      tive effect sizes when religiosity measures assessed religious (Bertsch & Pesta, 2009) used the ratio of 1/.71 to correct a                      beliefs. Effect sizes (rs) for the college (corrected) and non- correlation between college students’ scores on the Wonderlic                     college groups ranged from −.20 to −.25 (mean r                  −.24); in                                      3 Personnel Test and religiosity.  We chose to use only the ratio                   IQ points, the effect sizes ranged from 6.2 to 7.8 (M                7.3). computed by Sackett et al. (2009), because it was based on a far larger sample; still, the similarity between the two ratios                                                                                   Testing Third Variable Effects: Gender, Age, and was reassuring.                                                                                   Education     In   addition   to   correcting   the   correlations   at   the   college level, we also examined (at college and non-college levels),                      A  correlation   between   intelligence   and   religiosity   may   be                                                            4 the Cohen’s  d equivalents of the observed  rs.  Conversion                       due   to   a   third   variable   such   as   gender,   age,   or   education. from r to  d is informative when one of the two correlated                        Gender may act as a third variable because of its relation to variables can be dichotomized meaningfully. In the present                        religiosity—women           tend   to  be  more    religious    than   men analysis, religiosity dichotomizes conceptually to believers                      (McCullough,        Enders,     Brion,    &   Jain,   2005;    Sherkat    & and nonbelievers. Because the correlations used in the meta-                      Wilson, 1995; Stark, 2002). Age is also related to religiosity, analysis   were   based   on   the   entire   population   (studies   of          although this relation is not consistent across the life span or extreme groups were excluded), conceptualizing the equiva-                        across   cultures   (Argue,   Johnson,   &   White,   1999;   Sherkat, lent ds as the difference between believers and nonbelievers                      1998). Education, as noted earlier, is related to intelligence is   extremely   conservative.   Still,   the  d   of   the   difference   in     and religiosity. Accordingly, we identified studies that pro- intelligence scores between these two groups is highly infor-                     vided the relevant information (usually a correlation matrix mative; multiplying d by 15—the standard deviation of the                         of all variables) that allowed us to compute partial correla- most   widely   used   intelligence   tests   such   as   the   Wechsler          tions   between   intelligence   and   religiosity,   controlling   for Adult     Intelligence    Scale–Third      Edition    [WAIS–III]—pro-             each of the hypothesized third variables. vides   an   estimate   of   the   number   of   IQ   points   separating             Table 6 presents zero-order and partial correlations con- believers from nonbelievers.                                                      trolling for gender for 13 studies. The absolute differences     Table 5 presents the results for all studies at the college                   between the zero-order and partial correlations ranged from and non-college levels (top panel) and for studies utilizing                      .00 to .03, with a median difference of .01. Thus, controlling religiosity measures that targeted religious beliefs. (In Table 5,                for   gender     neither    augmented       nor   reduced     correlations we   used   all   the   studies   from   the   college   and   non-college        between intelligence and religiosity. groups from the enlarged data set presented in Table 4; k                             Table 7 presents zero-order and partial correlations con- 63.) Not surprisingly, the corrected effect sizes at the college                  trolling for age for 10 studies. Excluding Franzblau (1934), level (middle part of the table) are more negative than the                       absolute   differences   between   the   two   types   of   correlations uncorrected effect sizes (left side of the table). In addition,                   ranged   from   .00   to   .02,   with   a   median   of   .00.   Franzblau’s the corrected effect sizes at the college level appear compa-                     data yielded zero-order and partial correlations of −.15 and rable   with   those   at   the   non-college   level.   Relative   to   the      −.21, respectively. On balance, it seems that controlling for                                                Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 14-----------------------

Zuckerman et al.                                                                                                                                           13 Table 6.  Zero-Order and Partial (Controlling for Gender) Correlations Between Intelligence and Religiosity. Study                                                                                  Zero-order correlations                         Partial correlations Blanchard-Fields, Hertzog, Stein, and Pak (2001)    Study 1: Student sample                                                                            .00                                          .01    Study 2: Adult sample                                                                            −.32                                         −.30 Bloodgood, Turnley, and Mudrack (2008)                                                              −.15                                         −.16 Cottone, Drucker, and Javier (2007)                                                                 −.14                                         −.14 Deptula, Henry, Shoeny, and Slavick (2006)                                                          −.10                                         −.10 Foy (1975)                                                                                          −.50                                         −.52a Francis (1979)                                                                                        .04                                          .04 Francis (1997, 1998)                                                                                −.04                                         −.04 Francis, Pearson, and Stubbs (1985)                                                                 −.13                                         −.11 Hadden (1963)                                                                                       −.06                                         −.05 Kanazawa (2010a; personal communication, January 2012)    Study 1                                                                                          −.12                                         −.11    Study 2                                                                                          −.14                                         −.15 Lewis, Ritchie, and Bates (2011)                                                                    −.16                                         −.16 Pennycook, Cheyne, Seli, Koehler, and Fugelsang (2012)    Study 1                                                                                          −.19                                         −.22    Study 2                                                                                          −.17                                         −.16 aAverage of zero-order correlations that were computed separately for men and women. Table 7.  Zero-Order and Partial (Controlling for Age) Correlations Between Intelligence and Religiosity. Study                                                                                   Zero-order correlations                        Partial correlations Blanchard-Fields, Hertzog, Stein, and Pak (2001)    Study 1: Student sample                                                                          .00                                           .00    Study 2: Adult sample                                                                          −.32                                          −.30 Ciesielski-Kaiser (2005)                                                                          −.14                                          −.14 Cottone, Drucker, and Javier (2007)                                                               −.14                                          −.14 Deptula, Henry, Shoeny, and Slavick (2006)                                                        −.10                                          −.10 Francis, Pearson, and Stubbs (1985)                                                               −.13                                          −.14 Franzblau (1934)                                                                                  −.15                                          −.21 Hadden (1963)                                                                                     −.06                                          −.06 Kanazawa (2010a; personal communication, January 2012)    Study 1                                                                                        −.12                                          −.12    Study 2                                                                                        −.14                                          −.15 Lewis, Ritchie, and Bates (2011)                                                                  −.16                                          −.14 Pennycook, Cheyne, Seli, Koehler, and Fugelsang (2012)    Study 1                                                                                        −.19                                          −.19    Study 2                                                                                        −.17                                          −.18 age has little effect on correlations between intelligence and                    by   Blanchard-Fields,   Hertzog,   Stein,   and   Pak   (2001;   first religiosity.                                                                      row in Table 8) can be excluded because of range restriction     As   previously   noted,   some   investigators   suggested   that            for   intelligence   and   education   (indeed,   all   correlations   for education   mediates   the   relation   between   intelligence   and              that study were weak). The results of the remaining six stud- religiosity (Hoge, 1974; Reeve & Basalik, 2011). Interestingly,                   ies indicate that education does not mediate the intelligence– Kanazawa (S. Kanazawa, personal communication, January                            religiosity relation. 2012) espouses an opposing view, namely that intelligence                            To begin with, intelligence was more negatively related to accounts   for   any   negative   relation   between   education   and            religiosity   than   was   education   (unweighted   mean   correla- religiosity. Table 8 presents results that address the two com-                   tions were −.18 and −.06, respectively). We tested the sig- peting hypotheses. The analyses are based on seven studies                        nificance of this difference separately for each study, using a from three sources. Results from the student sample studied                       procedure       for   comparing        nonindependent         correlations                                           Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 15-----------------------

 14                                                                                                    Personality and Social Psychology Review XX(X) Table 8.  Zero-order and Partial Correlations Among Intelligence, Religiosity and Education.                                                       Intelligence and religiosity          Intelligence and education            Education and religiosity                                                     Zero-order              Partial                 Zero-order                 Zero-order             Partial Study                                               correlations        correlationsa               correlations               correlations       correlationsb Blanchard-Fields, Hertzog, Stein, and Pak (2001)    Study 1: Student samplec                              .00                 .02                        .21*                      −.06                −.06    Study 2: Adult sampled                              −.32***             −.31***                      .30***                    −.08                 .03 Kanazawa (2010a; personal communication, 2012)    Study 1e                                            −.12***             −.14***                      .32***                      .05***             .10***    Study 2f                                            −.14***             −.08***                      .50***                    −.17***             −.09*** Lewis, Ritchie, and Bates (2011)g                      −.16***             −.12***                      .41***                    −.14***             −.08** Pennycook, Cheyne, Seli, Koehler, and Fugelsang (2012)    Study 1                                             −.19**              −.20**                       .22***                      .01                .06    Study 2                                             −.17**              −.16**                       .27***                    −.05                 .00 aControlling for education. bControlling for intelligence. cN    96. dN    219. eN ranges from 14,265 to 14,987. f N ranges from 6,030 to 10,971 except for the correlation between intelligence and education (N        23,026). gN ranges from 1851 to 2307. *p < .05. **p < .01. ***p < .001. (Meng, Rosenthal, & Rubin, 1992); the combined difference                           an antecedent of) religiosity. The first set concerns the afore- across the six studies was highly significant, Z                  9.32, p  <       mentioned time gap studies; the second is based on studies of .001.   Furthermore,   controlling   for   education   did   not   have             gifted children, primarily Terman’s (1925-1959) study and, much of an effect on the intelligence–religiosity relation—                         indirectly,   the   Hunter   study   (Subotnik,   Karp,   &   Morgan, unweighted means of the six zero-order and partial correla-                         1989). tions were −.18 and −.17, respectively.     In contrast, controlling for intelligence led to a somewhat                     Time gap studies.  In four studies, intelligence was measured greater     change     in  the   education–religiosity        relation;    the      long before religiosity, with time gaps ranging from 3 to 25 unweighted means for the six zero-order and partial correla-                       years   (Bender,   1968;   Carlson,   1934;   Horowitz   &   Garber, tions were −.06 and .00, respectively. This finding is consis-                      2003; Kanazawa, 2010a, Study 1). If intelligence measured tent with S. Kanazawa’s (personal communication, January                            on   one   occasion   influences   religiosity   that   is   measured   a 2010) view that intelligence accounts for the education–reli-                      number of years later, then a significant correlation between giosity relation. However, given that the analysis is based on                     these   two   variables   is   consistent   with   a   model   in   which only six studies, our conclusions are tentative.                                    intelligence   drives   religiosity.   A   fifth   time   gap   study   by     Perhaps it is not how long people have been in school but                       Carothers, Borkowski, Burke Lefever, and Whitman (2005) rather    how     much     they   learnt   that   mediates     the   relation      was not included in this analysis. These investigators stud- between intelligence and religiosity. GPA can be viewed as                          ied   only   participants   who   were   very   high   or   very   low   on an indicator of how much knowledge one acquired in school.                         religiosity. This “extreme groups” design could have inflated If amount of knowledge mediates the relationship between                           the effect size (r        −.25, see Table 1) despite the time gap intelligence      and    religiosity,    intelligence      and    religiosity      procedure. should be correlated with GPA. While GPA is indeed corre-                              The four time gap studies yielded a mean effect size of lated with intelligence, it was shown earlier that GPA is not                       r    −.12 for unweighted and weighted analyses. This value related to religiosity. We again conclude that there is no evi-                    was   marginally   significant   in   the   random-effects   analysis, dence to support the notion that education mediates the intel-                      t(3)    1.99, p      .07, one-tailed, and highly significant (p  <  ligence–religiosity relation.                                                       .001)   in   the   fixed-effects   analysis. The   results   are   actually                                                                                    more impressive than they first appear. First, the Horowitz                                                                                     and   Garber   (2003)   study   used   behavior-   and   belief-based Does Intelligence Drive Religiosity?                                                                                    measures of religiosity. If we consider only the belief-based The present findings are correlational and cannot support any                      measure,       the   unweighted       average   r    of  the   four   studies causal relation. However, two sets of results are consistent                       becomes −.14, t(3)          3.98, p < .05. In addition, Bender (1968) with the hypothesis that intelligence influences (or at least is                    and Carlson (1934) studied college students and, if corrected                                                 Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 16-----------------------

Zuckerman et al.                                                                                                                                         15 Table 9.  Comparing Religiosity in Terman’s Sample with Religiosity in USA General Public. Terman’s sample                                                                              General public Year of data collection        Age         n        M        SD      Year of data collection        Age         n       M        SD         t test       r  1960                         44-60      894       1.79     1.46                1965               45-64      927      3.42      1.11     26.84***      .53  1977                         61-77      650       1.75     1.51                1978                65+        117     3.45      1.16     11.57***      .39  1986                         70-86      606       1.48     1.47                1986                65+        135     3.13      1.42     11.87***      .40  1991                         75-91      399       1.45     1.48                1991                65+         76     3.50      1.04     11.54***      .47 Note. The Terman’s sample data were available from McCullough, Enders, Brion, and Jain (2005). The general public data were available from the following surveys: Gallup Organization (1965), Yankelovich, Skelly, and White, Inc (1978), Gallup Organization (1986), and Yankelovich Clancy Shulman, Inc. (1991). ***p < .001. for range restriction, their effect sizes (−.20 and −.19, respec-                scores to the 0-to-4 scale that was used by McCullough et al. tively) become −.29 and −.26, respectively.                                      (2005).6     On the other hand, two limitations of these studies should                      Table   9   presents   the   findings.   In   all   four   comparisons, be noted. In all four studies, religiosity was predicted from                    Terman’s   sample   scored   significantly   lower   on   religiosity an earlier measure of intelligence without controlling for an                    than the general public (the average of these effects was used earlier measure of religiosity. In addition, intelligence was                    in the meta-analysis as one of the extreme groups’ studies). not predicted from an earlier measure of religiosity. Still, it is               Admittedly, the years of data collection and ages of the two remarkable   that   intelligence   can   predict   religiosity   scores          groups do not match perfectly. However, the results are so that are obtained years later.                                                   strong that it is difficult to imagine that more exact matching                                                                                  would make a difference. The   Terman      study.  The   Terman      (1925-1959)      longitudinal           These results are even more striking if the Termites’ reli- study of bright children initially included 1,528 participants                   gious   upbringing   is   considered.   Terman   and   Oden   (1959) who were identified at approximately 10 years of age as hav-                     reported   that   close   to   60%   of   Termites   reported   that   they ing    IQs   that   in  general    exceeded     135.    Religiosity    was       received   “very   strict”   or   “considerable”   religious   training; assessed among “Termites” in several subsequent waves of                         approximately   33%   reported   receiving   little   training,   and data collection. Holahan and Sears (1995) and McCullough                         about 6% reported no religious training. This suggests that et al. (2005) have noted that Termites grew up to be less reli-                  the Termites underwent changes in their religiosity after their gious than the general public. Until recently, however, it was                   childhood. difficult   to   conduct   a   systematic   comparison   among   Ter- mites’ religiosity scores at each wave of data collection, and                   The Hunter study.  Subotnik et al. (1989) compared findings between Termites’ religiosity and that of the general public.                    from Terman’s studies with findings from another group of This is because religiosity measures administered to the Ter-                    gifted children. The latter sample consisted of graduates of mites    varied    across   data   collection   points,   and   were    not      Hunter College Elementary School who were 38 to 50 years always comparable with measures administered to the gen-                         old at the time of the comparison. The Hunter participants eral public.                                                                     were tested approximately at the age of nine with the 1937     This    problem     was    addressed     by   McCullough       et  al.’s     edition    of  the  Stanford–Binet.   Termites’   intelligence         was (2005) study of religious development among the Termites.                        assessed     with    the   1916    edition   of   the   Stanford–Binet. These investigators rescored Termites’ religiosity data from                     Because IQ scores based on the 1937 version are comparable six time points on a uniform 0-to-4 scale (0             no importance           with somewhat lower IQ scores based on the 1916 version, or   being   actively   antireligious,   4     high   importance,   high         Subotnik et al. limited the Hunter group to individuals scor- interest, and high satisfaction gained from religion). Because                   ing 140 or greater (range: 140-196, M             159); Termites’ IQs public surveys (e.g., Gallup Organization, 1965) also mea-                       ranged from 120 to 180, M           148. sure   the   importance   of   religion,   we   were   able   to   compare          For the Hunter group, researchers administered a number religiosity levels scores from Terman’s sample at four time                      of   questions    that  were    used   earlier   in  the   Terman    study points (as presented in McCullough et al., 2005, article) to                     (Terman & Oden, 1959). Although these questions were not religiosity scores obtained at approximately the same years                      comparable with measures used in surveys of the U.S. public for   age-matched       individuals    from    the  general    public.   To      (R. F. Subotnik, personal communication, January 2012), the accomplish   this   comparison,   we   reverse   scored   the   three-           Hunter–Terman comparison is still informative. Because the point religiosity measure (1         very important, 3        not impor-         religiosity measures did not show any gender differences, we tant) used in four public surveys (Gallup Organization, 1965,                    present results only for the combined groups. 1986; Yankelovich, Skelly, & White, Inc, 1978; Yankelovich                          The Hunter and the Terman samples were asked to choose Clancy Shulman, Inc., 1991), and then rescaled the reversed                      any number of possible sources of personal satisfaction from                                           Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 17-----------------------

 16                                                                                                  Personality and Social Psychology Review XX(X) a list that included religion. In Terman’s sample (N                 428),        derived from data indicating that intelligence develops ear- 13.1%   chose   religion;   15.6%   chose   religion   in   the   Hunter          lier than does religiosity. Intelligence can be reliably mea- sample (N       147), Z of the difference < 1. Both samples were                  sured    at  a  very   early   age   while    religiosity   cannot    (e.g., also asked to identify any number of variables related to suc-                    Jensen, 1998; Larsen, Hartmann, & Nyborg, 2008). In their cess   from   a   list   that   included   religious/spiritual   values.   In     classic study, for example, H. E. Jones and Bayley (1941) Terman’s   sample   (N           410),   1.2%   checked   the   religious         showed that the mean of intelligence scores assessed at ages option, compared with .4% in the Hunter group (N                     139),        17 and 18 (a) correlated .86 with the mean scores assessed at Z < 1. These results suggest that on an absolute level, religion                  ages 5, 6, and 7; and (b) correlated .96 with the mean of intel- was relatively unimportant to middle-aged adults who were                         ligence scores assessed at ages 11, 12, and 13. Because intel- identified as gifted in childhood in both samples. In addition,                   ligence can be measured at an early age, it can be used to we   speculate   that   if   the   Hunter   sample   is   similar   to   the      predict outcomes observed years later. For example, Deary, Terman sample with respect to religiosity, it too may be less                     Strand, Smith, and Fernandes (2007) reported a .69 correla- religious than the general population.                                            tion between intelligence measured at age 11 and educational     In the Terman and the Hunter samples, a high intelligence                     achievement at age 16. level at an early age preceded lower religiosity many years                          Unlike intelligence, religiosity assessed at an early age is later.   However,   our   analyses   of   these   results   neither   con-        a   weak    predictor    of  religiosity    assessed    years    later.  For trolled   for   possible   relevant   factors   at   an   early   age   (e.g.,    example, Willits and Crider (1989) found only small to mod- socioeconomic         status)   nor   examined      possible     mediators        erate correlations between religiosity at age 16 and that at 27 (e.g., occupation) of this relation.                                              (.28   for   church   attendance   and   .36   for   beliefs).   O’Connor,                                                                                   Hoge, and Alexander (2002) found no relationship between Discussion                                                                        measures of church involvement at ages 16 and 38.                                                                                       The assumption that intelligence affects religiosity is also Results of the present meta-analysis demonstrated a reliable                      consistent with two of our findings: (a) In time gap studies, negative   relation   between   intelligence   and   religiosity.   The           intelligence measured on one occasion predicted religiosity size of the relation varied according to sample type and the                      that was measured years later; and (b) Terman’s study par- nature of the religiosity measure. The relation was weakest at                    ticipants, who were selected at an early age on the basis of the precollege level, although even in that group it was sig-                     high intelligence scores, reported years later lower religiosity nificantly different from zero. After correlations observed in                    than the general public (participants in the Hunter study also college   populations   were   corrected   for   range   restriction   of         showed this trend but the evidence in this case is indirect). intelligence scores, the magnitude of the intelligence–religi-                       Below, we discuss three proposed reasons for the inverse osity relation at the college level was comparable with that at                   relation between intelligence and religiosity. The first two— the non-college level.                                                            “atheism   as   nonconformity”   and   “cognitive   style”—repeat     The    relation    was   also   more    negative    when    religiosity       (with   some   elaboration)   explanations   that   were   previously measures      assessed     religious    beliefs   rather   than   religious       proposed      in  the  literature.   The    third  reason—“functional behavior.     This    difference    brings    to  mind    the   distinction       equivalence”—is (to the best of our knowledge) new. between intrinsic religious orientation (religion practiced for its own sake) and extrinsic religious orientation (using reli-                                                                                   Atheism as Nonconformity gion   as   a   means   to   secular   ends;   Allport   &   Ross,   1967). Because   religious   behavior   (e.g.,   attending   church)   can   be          As    noted   hereinbefore,   Argyle       (1958)    implied    that  more enacted for reasons extrinsic to faith, they are more aligned                     intelligent   people   are   less   likely   to   conform   to   religious with the concept of extrinsic religious orientation. Religious                    orthodoxy.   This   notion   incorporates   two   implicit   assump- beliefs, which are held privately, appear more aligned with                       tions. The first is that atheism can be characterized as non- the concept of intrinsic religious orientation. In Allport and                    conformity in the midst of religious majority. The second is Ross’s (1967) view, intrinsic religious orientation represents                    that more intelligent people are less likely to conform. There more   normative   or   “truer”   religiosity   (for   a   review   of   this     is   qualified   empirical   support   for   the   first   assumption   and issue,     see    Cohen,     Hall,    Koenig,      &    Meador,      2005).       strong support for the second. Accordingly, the finding that intelligence is more negatively                        First, although the prevalence of religiosity varies widely related   to   religious   beliefs   than   to   religious   behavior   sup-      among countries and cultures, more than 50% of the world ports the conclusion of a negative relation between the con-                      population consider themselves religious. Using survey data structs     of   intelligence     and   religiosity.    However,      some        collected by P. Zuckerman (2007) from 137 countries, Lynn limitations on this conclusion are noted below.                                   et   al.   (2009)   and   Reeve   (2009)   observed   a   prevalence   of     With   one   exception   (Sherkat,   2010),   the   interpretations           89.9%   believers   in   the   world   and   89.5%   believers   in   the that follow focus on the assumption that intelligence affects                     United States. However, a recent Win-Gallup International religiosity rather than the reverse. To be sure, this assump-                     (2012) poll of 59,927 persons in 57 countries found that only tion is not derived from our correlational data. Rather, it is                    59% of the respondents (60% in the United States) consider                                                Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 18-----------------------

Zuckerman et al.                                                                                                                                            17 themselves   religious,   a   decline   of   9%   (13%   in   the   United        that   are   not   subject   to   empirical   tests   or   logical   reasoning States) from a similar 2005 poll. Atheism might be consid-                        (e.g., Nyborg, 2009). But why would intelligent people know ered a case of nonconformity in societies where the majority                      better? It does not take a great deal of cognitive ability to is   religious.   This   is   not   so,   however,   if   one   grows   up   in   understand   that   religion   does   not   arise   from   scientific   dis- largely atheist societies, such as those that exist in Scandinavia                course. One does not generally hear from believers that their (P. Zuckerman, 2008).                                                             faith is based on fact or logic, but they continue to believe     People   who   do   grow   up   in   a   religious   environment   are        anyhow.   What        is  it  exactly   about   intelligent    people    that likely to believe and practice what is supported and espoused                     makes them more resistant to religion? in    their  social   environment       (Gervais     &   Henrich,     2010;           The answer to this question may be related to cognitive Gervais, Willard, Norenzayan, & Henrich, 2011). Still, what                       style.   Dual   processing   models   of   cognition   (e.g.,   Epstein, makes an atheist in a religious society a nonconformist is not                    1994;   Kahneman,   2003;   Stanovich   &   West,   2000)   distin- only that most people are religious, but also that religion is                    guish between analytic and intuitive styles (this distinction more than a privately held belief. According to Graham and                        has also been called “system 2” vs. “system 1,” “think” vs. Haidt     (2010),    religious   practices    can   serve   to  strengthen        “blink,”   etc.).   Analytic   thinking   is   controlled,   systematic, social    bonds     and   ensure    a  group’s     continued     existence.       rule-based,   and   relatively   slow;   intuitive   thinking,   in   con- Ysseldyk,   Matheson,   and   Anisman   (2010)   suggested   that                 trast, is reflexive, heuristic-based, spontaneous, mostly non- religion     provides    social   identity    and   an   “eternal”    group       conscious,       and    relatively    fast.  We     propose     that   more membership.   There   is   also   empirical   evidence   suggesting               intelligent people tend to think analytically and that analytic that religiosity may be an in-group phenomenon, reinforcing                       thinking leads to lower religiosity. There is empirical support prosocial     tendencies     within    the   group    (see   a  review    by      for both these hypotheses. Norenzayan & Gervais, 2012), but also predisposing believ-                            A common test of the tendency to use analytic thinking is ers   to   reject   out-groups   members   (see   meta-analysis   by              Frederick’s      Cognitive      Reflection     Test    (CRT;     Frederick, D.   L.   Hall,   Matz,   &   Wood,   2010).   To   become   an   atheist,        2005). This instrument assesses the ability to choose correct therefore, it may be necessary to resist the in-group dogma of                    but    intuitively   unattractive     answers,     which    is  thought    to religious beliefs. Not surprisingly, there is evidence of anti-                   reflect   reliance   on   analytic   thinking.   CRT  scores   are   posi- atheist      distrust    and     prejudice      (Gervais,     Shariff,     &      tively   associated   with   better   performance   on   a   number   of Norenzayan,   2011;   Gervais   &   Norenzayan,   2012b;   for   a                heuristic   problems   (i.e.,   lesser   susceptibility   to   misleading review, see Norenzayan & Gervais, 2012).                                          intuitions; Cokely & Kelley, 2009; Frederick, 2005; Toplak,     Intelligent people may be more likely to become atheists                      West,   &   Stanovich,   2011;   but   see   Campitelli   &   Labollita, in religious societies, because intelligent people tend to be                     2010). nonconformists. In a meta-analysis of seven studies, Rhodes                           CRT scores are also positively related to intelligence, with and Wood (1992) found that more intelligent people are more                       correlations in the .40-.45 range (Frederick, 2005; Obrecht, resistant to persuasion and less likely to conform. In addition                   Chapman, & Gelman, 2009; Toplak et al., 2011). There is to   the   studies   reviewed   by   Rhodes   and   Wood,   three   other         also   evidence   linking   higher   intelligence   to   better   perfor- investigations       reported      a   significant     negative     relation      mance on a variety of other heuristics and biases tasks (there between   intelligence   and   conformity   (Long,   1972;   Smith,               are    exceptions,     however;     see   Stanovich     &    West,   2008). Murphy, & Wheeler, 1964; Osborn, 2005). Rhodes and Wood                           Importantly, Stanovich and West (2008) proposed that intel- (1992) proposed that the greater knowledge that intelligent                       ligent people are more able to override cognitive biases, not people possess allows them to be more critical and less yield-                    so   much   because   they   realize   that   the   appealing   intuition ing when presented with arguments or claims (cf. W. Wood,                         might be wrong or because they have the ability to find the 1982; W. Wood, Kallgren, & Preisler, 1985). Recently, Millet                      more time-consuming logical solution. Instead, more intelli- and    Dewitte     (2007)    reported    a  positive    relation   between        gent people may be more capable of sustaining the cognitive intelligence and self-perceived uniqueness; this led them to                      effort needed for good performance on heuristics tasks. propose that more intelligent people conform less because of                          There is strong evidence that analytic style, as measured their ability to be self-sufficient and to secure resources in                    by performance on heuristic tasks (e.g., CRT) or induced by isolation.                                                                        priming is related to lower religiosity (Gervais & Norenzayan,     If more intelligent people are less likely to conform, they                   2012a;     Pennycook,       Cheyne,     Seli,  Koehler,     &   Fugelsang, also    may    be   less  likely   to  accept    a  prevailing    religious       2012; Shenhav, Rand, & Greene, 2011). Interestingly, both dogma.                                                                            Shenhav et al. (2011) and Gervais and Norenzayan (2012a)                                                                                   argued   that   religious   beliefs   are   a   matter   of   intuitive   pro-                                                                                   cesses that can be overridden through analytic approach. In Atheism and Cognitive Style                                                                                   contrast,   Pennycook   et   al.   (2012)   proposed   that   religious As noted hereinbefore, the most common explanation for the                        beliefs are actually counterintuitive (unwarranted on either inverse relation between intelligence and religiosity is that                     logical or empirical grounds), and thus require more analytic the intelligent person “knows better” than to accept beliefs                      scrutiny if they are to be rejected. Independent of the exact                                            Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 19-----------------------

 18                                                                                                   Personality and Social Psychology Review XX(X) mediating mechanism, we propose that intelligent people are                       also provides these four benefits and, therefore, lowers one’s less   religious   because   they   are   more   likely   to   use   analytic     need to be religious. processes.     There is some empirical support for our proposition. Both                     Religiosity as compensatory control.  Religiosity can provide a  Shenhav et al. (2011) and Pennycook et              al. (2012) adminis-          sense of external control, that is, the perception that the world tered intelligence measures in addition to measures of religios-                  is   orderly   and   predictable   (as   opposed   to   random   and   cha- ity and cognitive style. Using a college sample, Shenhav et al.                   otic); religiosity can also provide a sense of personal control (2011) found a negative but not significant relation between                      by empowering believers directly through their personal rela- intelligence and religiosity (mean r = −.06; see table 1). Using                  tions   with   God.   In   a   series   of   studies,   Kay   and   colleagues non-college samples in two studies, Pennycook et al. (2012)                       (Kay, Gaucher, Napier, Callan, & Laurin, 2008; Kay, Mosco- found significant correlations between two measures of intel-                     vitch,   &   Laurin,   2010;   Laurin,   Kay,   &   Moscovitch,   2008) ligence and a measure of religious beliefs. Controlling for two                   showed that threatening a sense of personal control increased measures   of   analytic   style,   these   correlations   were   reduced         beliefs   in   God,   particularly   when   the   controlling   nature   of from −.24 and −.15, ps  < .05, to −.13, p  < .05, and −.04, ns,                   God was emphasized. According to these investigators, peo- respectively, in Study 1, and from −.13, and −.22, ps  < .05, to                  ple    who    lose   personal    control    take   comfort     in  religion, −.02 and −.08, ps  > .28, respectively, in Study 2.                               because it suggests to them that the world is under God’s con-     In   contrast,   correlations   between   measures   of   analytic            trol   and,   therefore,   predictable   and   nonrandom.   Kay,   Gau- style and religiosity were lower but mostly remained signifi-                     cher,     McGregor,       and   Nash     (2010)    also    suggested     that cant when controlling for two measures of intelligence. In                        religiosity   can   confer   a   specific   form   of   personal   control study 1, these correlations were reduced from −.33 and −.19,                      when other forms of personal control are decreased. They cite ps  < .05, to −.26, p  < .05, and −.11, p  <.10, respectively; in                 evidence indicating that individuals whose personal control is  Study 2, these correlations were reduced from −.29 and −.31,                     threatened   become   more   certain   of   the   superiority   of   their ps  < .05 to −.25 and −.23, ps  < .05, respectively. These results                religious beliefs, more determined to live in accordance with are consistent with the proposition that cognitive style medi-                    their faith, and more convinced that others would agree with ates,   at   least   in   part,   the   negative   relation   between   intelli-  their beliefs if they tried to understand them (McGregor, Haji, gence and religiosity.                                                            Nash, & Teper, 2008; McGregor, Nash, & Prentice, 2009). In     Our   proposition   is   also   consistent   with   Stanovich   and           sum, religiosity provides compensatory control when an indi- West’s (2008) model, which links intelligence to bias over-                       vidual’s personal control beliefs are undermined. ride. We suggested hereinbefore that the rejection of religion                        Intelligence also confers a sense of personal control. We does     not   necessarily    require    superior    cognitive    skills.  In     identified   eight   studies   that   reported   correlations   between other   words,   neither   bias   override,   according   to   Stanovich          intelligence and belief in personal control (Grover & Hertzog, and West, nor religion override as stated above, depends that                      1991;   Lachman,   1983;   Lachman,   Baltes,   Nesselroade,   & much   on   tools   or   ability   that   intelligent   people   are   more       Willis, 1982; Martel, McKelvie, & Standing, 1987; Miller likely to have. Instead, if one grows up in a religious com-                      &     Lachman,      2000;    Prenda     &   Lachman,      2001;   Tolor    & munity, rejecting theism probably requires a sustained cogni-                     Reznikoff, 1967; P. Wood & Englert, 2009). All eight cor- tive effort. Intelligence may confer the ability to sustain such                  relations were positive, with a mean correlation (weighted an effort (Stanovich & West, 2008).                                               by df of each study) of .29. In addition, higher intelligence is                                                                                   associated   with   greater   self-efficacy—the   belief   in   one’s                                                                                   own ability to achieve valued goals (Bandura, 1997). This Functional Equivalence                                                                                   construct is similar to personal control beliefs but has been In   his   introduction   to   the   special  Personality   and   Social          examined separately in the literature. In a meta-analysis of Psychology        Review     issue   on   religion,    Sedikides     (2010)       26   studies,   the   mean   correlation   between   intelligence   and described a functional approach to religion that posits a “spe-                   self-efficacy was .20 (Judge, Jackson, Shaw, Scott, & Rich, cific motive or need driving religious belief and practice” (p.                   2007). 4). In this approach, religious beliefs and practices satisfy a                       If more intelligent people are higher in personal control number of needs, and need-fulfillment is a potential reason                       beliefs or self-efficacy, then they may have less need for the for adopting and maintaining religious beliefs. It is possible,                   sense of control offered by religion. however, that needs typically fulfilled through religion can also be fulfilled through other means. Specifically, some of                      Religiosity    as   self-regulation.  McCullough        and    Willoughby the   functions   of   religion   may   also   be   conferred   by   intelli-     (2009)      presented     evidence     that   religiosity    is  associated gence; that is, in some respects, religion and intelligence may                   (albeit weakly) with positive outcomes, including well-being be functionally equivalent.                                                       and academic achievement. They suggested that self-regula-     We describe hereafter four functions that religion may pro-                   tion (adjusting behavior in the pursuit of goals) and self-con- vide:    compensatory        control,   self-regulation,     self-enhance-        trol    (forgoing    small,    immediate      rewards    to   increase    the ment,   and   attachment.   We   propose   that   higher   intelligence           likelihood   of   obtaining   larger,   but   delayed   rewards)   might                                                Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 20-----------------------

Zuckerman et al.                                                                                                                                           19 mediate the association between religiosity and positive out-                     correlations  between  intelligence  and  impulsiveness.  In  the comes. The researchers presented evidence from cross-sec-                         Dolan and Fullam (2004) study, intelligence was negatively tional, longitudinal, and experimental studies showing that                       related to two other impulsiveness scales besides the BIS and religiosity promotes self-control. They marshaled additional                      to   three   behavioral   measures   of   impulsiveness   besides   the evidence indicating that religiosity facilitates the completion                   delay of gratification task. of each component of the self-regulation process, including                           Shamosh and Gray (2008) offered a number of explana- goal setting, monitoring discrepancies between one’s present                      tions for the relation between intelligence and self-control. state   and   one’s   goals,   and   correcting   behavior   to   make   it       They argued that delay of gratification may require working more compatible with one’s goals. Finally, McCullough and                         memory       to  maintain     representations      of  delayed    rewards Willoughby   presented   evidence   indicating   that   self-control              while processing other types of information (e.g., opportu- and/or self-regulation mediate the relation between religios-                     nity costs of forgoing immediate rewards). More intelligent ity   and    positive   outcomes.      Consistent     with   that   review,       people   have   better   working   memories   (for   a   review,   see Rounding, Lee, Jacobson, and Ji (2012) found that partici-                        Ackerman, Beier, & Boyle, 2005), which may explain why pants primed with religious concepts exercised better self-                       they   have   better   self-control.   Alternatively,   Shamosh   and control;     in  addition,   priming     religious    concepts    renewed         Gray   proposed   that   delay   of   gratification   requires   “cool” self-control   in   participants   whose   ability   to   exercise   self-        (more rational) executive functioning rather than “hot” (more control had been depleted.                                                        affective)    executive     functioning.     More    intelligent   people,     A more nuanced model of the relation between religiosity                      suggested Shamosh and Gray, are more likely to engage the and self-control was proposed by Koole, McCullough, Kuhl,                         cool system and may therefore be better able to exercise self- and Roelofsma (2010). They proposed that intrinsic religios-                      control.   Regardless   of   the   mechanism,   if   more   intelligent ity facilitates implicit self-regulation whereas extrinsic reli-                  people have better self-regulation and/or self-control capa- giosity     (as   well   as   fundamentalism)        facilitates    explicit      bilities, then they may have less need for the self-regulatory self-regulation.      Focusing      on   the   implicit   aspect    of   this     function of religiosity. dichotomy, Koole et al. argued that the components of intrin- sic religiosity (holistic approach to well-being, integration of                  Religiosity   as   self-enhancement. As  stated   by   Sedikides   and cognitive   processing,   and   embodiment)   draw   on   the   same              Gebauer   (2010),   “people   are   motivated   to   see   themselves processes that are used in the service of implicit self-regula-                   favorably . . . Stated differently, people are motivated to self- tion. They   reviewed   a   large   number   of   findings   consistent           enhance”      (p.  17).   Meta-analyses       by  Trimble     (1997)    and with this model. For example, the relation between intrinsic                      Sedikides and Gebauer indicated that intrinsic religiosity is religiosity   and   implicit   self-regulation   of  action  was   illus-         positively     related    to  self-enhancing       responses     although trated by evidence showing that priming religious concepts                        extrinsic     religiosity    is  not.   To   explain     these   findings, increases     prosocial     behavior    (Randolph-Seng        &    Nielsen,       Sedikides      and   Gebauer      proposed     that   religious    cultures 2007);     and   the   relation   between     intrinsic   religiosity   and       approve of being religious as an end in itself, which can turn implicit self-regulation of affect was illustrated by evidence                    intrinsic religiosity into a source of self-worth. Religious cul- showing that praying for someone reduced anger after provo-                       tures disapprove, however, of using religion as a means to cation (Bremner, Koole, & Bushman, 2011).                                         secular ends, which may explain the disassociation between     Intelligence is also associated with better self-regulation                   extrinsic religiosity and self-enhancement. In support of this and self-control abilities. The classic test of such abilities is                 model, Sedikides and Gebauer showed that in more religious the    delay   of  gratification    paradigm     in  which    participants        cultures, (a) the positive relation between intrinsic religiosity choose between a small immediate reward and a large delayed                       and   self-enhancement   was   more   positive,   whereas   (b)   the reward (Block & Block, 1980). Choosing the large delayed                          low   or   negative   relation   between   extrinsic   religiosity   and reward serves as an indicator of self-control. Shamosh and                        self-enhancement was more negative. Yet another reason for Gray (2008) meta-analyzed the relation between intelligence                       the    association     between      intrinsic    religiosity    and    self- and delay discounting (the latter construct is identical to delay                 enhancement may be the elevated status that believers can of gratification except that high delay discounting indicates                     derive   from   personal   relationships   with   God   (Sedikides   & poor self-control). Their analysis, based on 26 studies, yielded                  Gebauer, 2010; see also Batson, Schoenrade, & Ventis, 1993; a   mean r  of   −.23. This   suggests   that   intelligent   people   are        Exline, 2002; Reiss, 2004). more likely to delay gratification (i.e., less likely to engage in                   Like religiosity, intelligence may provide a sense of higher delay discounting).                                                               self-worth.     Evidence      for  this   comes     from    two   lines   of     Two   of   the   studies   included   in   the   Shamosh   and   Gray         research.   First,   a   number   of   studies   examined   the   relation (2008) meta-analysis (de Wit, Flory, Acheson, McCloskey, &                        between      intelligence     and   self-esteem.     While     one    study Mannuck, 2007; Dolan & Fullam, 2004), and a third study by                        (Gabriel,     Critelli,  &    Ee,   1994)    reported     no   association Dom,   De   Wilde,   Hulstijn,   and   Sabbe   (2007),   utilized   the           between   the   two   constructs   (r       −.02),   three   other   studies Barratt     Impulsiveness      Scale    (BIS;   Barratt,   1985;    Patton,       (Judge, Hurst, & Simon, 2009; Lynch & Clark, 1985; Pathare Stanford, & Barratt, 1995). All three studies reported negative                   & Kanekar, 1990) reported significant albeit small positive                                           Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 21-----------------------

20                                                                                                   Personality and Social Psychology Review XX(X) correlations (rs       .18, .27, and .16, respectively). A second                early but, as they become older, they are more likely to be line   of   research   linking   higher   intelligence   to   higher   self-     married than less intelligent individuals. worth concerns the relation between intelligence and general                         Turning to divorce rates, Herrnstein and Murray (1994) factors of personality. Harris (2004) reduced 10 personality                     and Holley, Yabiku, and Benin (2006) found negative asso- scales to two factors—openness and achievement—that cor-                         ciations between intelligence and the likelihood of divorce. related .15 and .26, respectively, with intelligence. Schermer                   Blazys   (2009)   reported   the   same   negative   association   for and Vernon (2010) reduced 20 personality scales to a single                      Caucasians       but   found    no  significant    relation   for   African general factor of personality that correlated .27 with intelli-                  Americans and Hispanics. Finally, Dronkers (2002) found a gence. These authors proposed that high scores on the general                    negative      relation   between     intelligence    and    likelihood    of personality factor represent high self-esteem, emotional sta-                    divorce for a Dutch cohort born in 1958, but a positive rela- bility,   agreeableness,   conscientiousness,   and   openness—all               tion for a cohort born in 1940. When the data on marriage strongly     positive    attributes.   If  intelligent   individuals     see     and divorce rates are considered together, on balance more themselves   as   possessing   such   attributes,   then   they   might          intelligent people appear more likely to be married. have less need for the self-enhancement function of religion.                        Most     explanations     for   the  association     between     intelli-                                                                                  gence   and   marital   status   focus   on   the   ability   of   intelligent Religiosity   as   attachment.  Kirkpatrick   (2005)   proposed   that           people to plan more effectively, to act less impulsively, to religious   beliefs   can   be   conceptualized   as   an   attachment           adapt to changes, and so on. Regardless of the mediator, if system      (Bowlby,     1980),    which    can    confer   security    and      intelligent people are more likely to be married, then they safety in times of distress. Believers, suggested Kirkpatrick,                   may have less of a need to seek religion as a refuge from experience personal love of God (or some other supernatu-                        loneliness. ral   entity)   whose   omnipresence   serves   as   refuge   and   safe haven. There are two models of the association between reli-                                                                                  The Focus on Intrinsic Religiosity giosity   and   attachment   (Granqvist,   Mikulincer,   &   Shaver, 2010; Kirkpatrick, 1998). According to the first, the com-                       A common theme in most, although not all, of the interpreta- pensation model, people turn to God as an attachment figure                      tions hereinbefore is that they focus on intrinsic religiosity. when they experience loss due to separation, death of loved                      Sometimes   this   focus   is   stated   explicitly.   Other   times,   the ones, and other dire circumstances. According to the second,                     focus is described as religious beliefs, which (as noted here- the correspondence model, people extend to God the same                          inbefore)   are   more   strongly   related   to   intrinsic   religiosity attachment system that they have developed with close oth-                       than is religious behavior. For example, analytic thinking (a ers. This latter model does not posit a clear religious func-                    possible   mediator   of   the   relation   between   intelligence   and tion and, therefore, is not relevant to the notion of functional                 religiosity) has been shown to undermine religious beliefs. equivalence.                                                                     Of the four functions that are associated with religiosity and     There is strong support for the compensation model. For                      intelligence,     self-enhancement        was   exclusively     related   to example, S. L. Brown, Nesse, House, and Utz (2004) found                         intrinsic   religiosity.   The   remaining   functions—compensa- that religiosity increased following bereavement and, in turn,                   tory    control,   self-regulation,     and   attachment—are        mostly was associated with less grief. Evidence also indicates that                     functions of religious beliefs. Only the conceptualization of religiosity   increases   after   people   are   exposed   to   threat   of      atheism as nonconformity stands out in this regard. One can loneliness (Epley, Akalis, Waytz, & Cacioppo, 2008). In two                      certainly resist religious practice as much as one can resist other    studies    (Kirkpatrick     &   Shaver,     1992;   Kirkpatrick,        religious beliefs. Of course, resisting religious practice can Shillito, & Kellas, 1999), participants reporting a secure per-                  signal a search for a higher level, “purer” form of religion as sonal relationship with God also reported less loneliness.                       much as it can signal a step toward atheism.     Intelligence also lowers loneliness through its effects on                       Measures of religious behavior do not elucidate the rea- marital relations. Specifically, evidence suggests that more                     son   for   attending   church,   belonging   to   religious   organiza- intelligent people are more likely to marry and less likely to                   tions,   or   engaging   in   other   religious   practices.   We   argued get divorced. Terman and Oden (1947) reported that, as of                        earlier that because these reasons can be extrinsic to faith, the the 1930s, the prevalence of marriage in their high IQ sample                    behaviors       measured      might     reflect   extrinsic     religiosity. exceeded that of the general population. Similarly, Quensel                      Because most of the proposed interpretations focus on intrin- (1958) found that as intelligence increases, so does the likeli-                 sic religiosity and/or religious beliefs, they lead to the pre- hood of being married. Herrnstein and Murray (1994) found                        diction that intelligence would be more negatively related to that by age 30, marriage rates are lower at high and low ends                    religious beliefs than to religious behavior. The results are of the intelligence spectrum. However, the low marriage rate                     consistent with this prediction. observed among highly intelligent people could reflect a ten-                        However, it has been noted that viewing religious beliefs dency of more intelligent people to marry late. Blazys (2009)                    as the more genuine or as the intrinsic component of religios- examined   marriage   rates   up   to   an   average   age   of   43   and       ity characterizes American Protestant religion (Cohen et al., found that more intelligent people are less likely to marry                      2005).     Judaism     and   “catholic”    Christianity,    Cohen     et  al.                                                Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 22-----------------------

Zuckerman et al.                                                                                                                                           21 (2005) argued, view religious rituals and practice at least as                   those     with   access   to  knowledge.   A   test     of  these    effects important or central as religious beliefs. Perhaps, then, the                    requires a longitudinal study. stronger negative relation between intelligence and religious beliefs (relative to religious behavior) may be less true for                     Trajectory of the Intelligence–Religiosity Judaism       and   Catholicism.      That    is,  when     Judaism     and                                                                                  Connection Catholicism are concerned, perhaps the concept of functional equivalence might encompass not only the function of reli-                       The mechanisms through which intelligence affects religios- gious   beliefs,   but   also   the   functions   of   religious   practice.     ity may vary across the life span. At college, for example, This issue is left for future research.7                                         more   intelligent   students   may   be   more   likely   to   embrace                                                                                  atheism as a form of nonconformity; at a more advanced age,                                                                                  intelligent   people   may   be   more   likely   to   embrace   atheism Other Interpretations                                                                                  because they are more likely to be married and, therefore, As   mentioned   in   the   introduction,   Kanazawa   (2010a)   and             may be less reliant on the attachment function that religion Sherkat (2010) proposed two additional interpretations of the                    provides. We address in the following section, the question negative      relation    between      intelligence     and    religiosity.      of   when   mediators   of   the   intelligence–religiosity   relation Kanazawa   (2010a)   argued   that   more   intelligent   people   are           come into play. Importantly, this section does not review the better equipped to deal with evolutionarily novel phenom-                        life span trajectory of religiosity; rather, we focus only on the ena,    including     atheism.    Sherkat    suggested     that   sectarian      relation of religiosity with intelligence. In addition, much of affiliations   and   Christian   fundamentalism   block   access   to            the following discussion is speculative. secular   knowledge   and,   thereby   negatively   impact   verbal                  Human beings are psychologically predisposed to develop ability. We comment briefly on these views below.                                religious beliefs (Barrett, 2004; Boyer, 2001; Guthrie, 1993).     Kanazawa’s        (2010a)     interpretation      is  based    on    the     Biases or tendencies of the human mind that support religios- assumption that evolution favored the development of reli-                       ity   include   misattributions   of   intent   to   naturally   occurring gion. This assumption is readily acceptable, particularly in                     events (Kelemen, 2004; Kelemen & Rosset, 2009) and belief view of the functions that religion seems to provide. He also                    in disembodied mind as an attribute of supernatural deities argued that atheism is evolutionarily novel because, except                      (Bering,      2006;    Bloom,      2007;    Norenzayan,       Gervais,     & for   former   communist   societies,   it   is   not   mentioned   in   the     Trzesniewski,        2012).    As    noted     hereinbefore,     however, description   of   any   culture   in   The   Encyclopedia   of   World          Gervais, Willard, et al. (2011) argued convincingly that vari- Cultures. However, it is rather difficult to write about athe-                   ations   in   beliefs   across   societies   depend   heavily   on   social ism because, unlike theism, it does not produce (religious)                      contexts. That is, an individual is likely to believe only in relics and is not associated with (religious) customs. Thus,                     supernatural entities that are espoused in that person’s sur- although   it   is   not   mentioned   in   the Encyclopedia,   atheism          roundings; in religious societies, those who do otherwise risk could have existed all along, together with theism. In addi-                     being     labeled    as   heretics.   This   context-bound       approach tion, it is possible to consider monotheism as evolutionarily                    might   explain   the   weak   relation   between   intelligence   and novel instead of part and parcel of all preceding beliefs in the                 religiosity in precollege populations. supernatural;       this   will    negate    the    basic    rationale    of         During   adolescence,   there   is   a   strong   relation   between Kanazawa’s   (2010a)   approach.   Finally,   it   is   not   clear   that       religiosity   of   parents   and   that   of   their   children   (Cavalli- atheism belongs to the category of evolutionarily novelprob -                    Sforza,      Feldman,      Chen,    &    Dornbusch,       1982;    Gibson, lems  that   intelligence   addresses   (unless   atheism   is   consid-         Francis, & Pearson, 1990; Hoge, Petrillo, & Smith, 1982). As ered problematic because it does not provide the functions                       adolescents grow older, these associations decrease such that that religion does).                                                             correlations   between   childhood   religious   socialization   and     On the other hand and in line with Kanazawa’s (2010a)                        religiosity in adulthood are weak or nonexistent (Arnett & model, genetic influences have been implicated not only in                       Jensen,   2002;   Hoge,   Johnson,   &   Luidens,   1993; Willits   & intelligence (cf., Nisbett et al., 2012b), but also in religiosity               Crider, 1989). If religiosity in adolescence is largely a func- (D’Onofrio, Eaves, Murrelle, Maes, & Spilka, 1999; Koenig,                       tion of parental instructions and example, then it will be only McGue, & Iacono, 2008). Furthermore, the model was used                          minimally influenced by attributes of the person, including to predict other correlates of intelligence (e.g., political liber-              intelligence. alism     and,   for  men,    monogamy),        and   those    predictions           College      exposes    people    to  new    ideas   and   influences, received      empirical     support.    In   conclusion,      Kanazawa’s         which can impact religious beliefs. Students’ beliefs become (2010a) interpretation remains an intriguing possibility.                        more secular in college (Funk & Willits, 1987; Madsen &     Sherkat’s (2010) interpretation, while limited to Christian                  Vernon,   1983),   and   religious   service   attendance   decreases fundamentalism and verbal ability, alerts us to some poten-                      (Hunsberger,   1978;   Lefkowitz,   2005).   However,   there   are tial effects of religiosity on intelligence. It is likely that such              also   reports   of   an   increase   in   religious   commitment   and effects take time to develop, as those who are denied learning                   intrinsic     religiosity    during     this   period     (De    Haan     & fall more and more behind, over time, in comparison with                         Schulenberg, 1997; Stolzenberg, Blair-Loy, & Waite, 1995).                                           Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 23-----------------------

22                                                                                                 Personality and Social Psychology Review XX(X) These changes are often a consequence of the self-explora-                      (for a review, see Vail et al., 2010). This function was not tion    that  typifies   emerging     adulthood     and    that  is  often      included in our discussion of functional equivalence because, observed   in   college   students   (Arnett,   1997,   1998;   Greene,         to the best of our knowledge, there is no evidence pertaining Wheatley, & Aldava, 1992; Lefkowitz, 2005). The separa-                         to   the   relation   between     intelligence    and    death   anxiety. tion from home and the exposure to a context that encour-                       Although      this   logic   suggests    that  the   negative    relation ages questioning may allow intelligence to impact religious                     between intelligence and religiosity might decline at the end beliefs.   Using   analytic   (as   opposed   to   intuitive)   thinking,       of life, the relevant evidence we have indicates otherwise. more     intelligent   college   students    may    be  more    likely   to     The highly intelligent members of Terman’s sample retained eschew religion. If atheism is disapproved of at home, higher                   lower religiosity scores (relative to the general population) intelligence may facilitate resistance to conformity pressure.                  even at 75 to 91 years of age (Table 9). Additional research is These mechanisms might explain why the negative relation                        needed to resolve this issue. between      intelligence    and   religiosity   increases    in  college. However, as noted by Kosa and Schommer (1961), religious                        Limitations colleges may offer an exception to this trend.     The exploration that characterizes the college years con-                   The available data did not allow adequate consideration of the tinues later (Arnett & Jensen, 2002). However, those who                        role of religion type and of culture. As mentioned hereinbe- transition to atheism during college may face unanticipated                     fore, the articles included in the meta-analysis did not provide challenges.   Outside   of   academic   contexts,   most   societies            enough information to code religion type as a potential mod- are religious, and atheists are viewed with distrust (Gervais,                  erator. There was also not enough information to consider the Shariff,   et   al.,   2011).   We   speculate   that   more   intelligent      role of culture in the intelligence–religiosity association. Of people are better able to address these challenges through                      the 41 studies in the college and no-college groups (the popu- some of the aforementioned intelligence-related functions.                      lations on which we base most of our conclusions), 33 were These   functions   may   take   time   to   develop.   For   example,          conducted in the United States; the remainder were conducted intelligent people typically spend more time in school—a                        in Canada (3), Australia (2), Belgium and Holland (1 each); form of self-regulation that may yield long-term benefits.                      finally, one study was conducted in several countries but pri- More intelligent people get higher level jobs (Herrnstein &                     marily   (87%   of   participants)   in   the   United   States,   Canada, Murray, 1994), and better employment (and higher salary)                        and the United Kingdom. Clearly, the present results are lim- may   lead   to   higher   self-esteem,   and   encourage   personal            ited to Western societies. control     beliefs.   Last,   more    intelligent   people    are   more          Earlier   we   alluded   to   some   possible   effects   of   religion likely to get and stay married (greater attachment), though                     type   and   culture.   Specifically,   it   was   mentioned   that   the for intelligent people, that too comes later in life (Blazys,                   emphasis on beliefs as the intrinsic component of religiosity 2009). We therefore suggest that as intelligent people move                     (and, as such, the component with stronger negative relation from young adulthood to adulthood and then to middle age,                       to intelligence) might be an attribute of American Protestant the benefits of intelligence may continue to accrue. Thus,                      religion, and may be less true of Judaism and Catholicism after college, the degree to which intelligence obviates the                    (Cohen et al., 2005). Stated differently, the stronger negative functions of religion may gradually increase over time.                         relation   of   intelligence   with   religious   beliefs   may   also   be     The religious practices and beliefs adopted during college                  limited to American Protestant population. and in subsequent years are often retained for the remainder                       We also mentioned above that atheism is not likely to be of   the   life   span.   McCullough   et   al.   (2005)   reported   that      considered  nonconformist  in  majority  atheist  societies,  like (unlike the weak relation between religiosity in the precol-                    Scandinavian societies (P. Zuckerman, 2008). Atheism may lege years and religiosity in adulthood) there is considerable                  also lose its association with nonconformity in majority athe- rank-order stability in religiosity from the early 20s to the                   ist subcultures, such as the subculture of scientists (Larson & end of life. However, these investigators also noted that in                    Witham,   1998).   One   might   even   speculate   that   in   majority addition to interindividual stability, there are also intraindi-                atheist societies, atheism is associated with conformity rather vidual changes as people increase and decrease in religiosity                   than nonconformity. Even in these societies, however, several over time. For example, most people become more religious                       other proposed causes of the negative relation between intel- when they get married and have children, but become less                        ligence and religiosity remain intact. First, religion remains religious when their children leave home (Ingersoll-Dayton,                     negatively linked to analytic style, which characterized more Krause, & Morgan, 2002; McCullough et al., 2005; Sherkat                        intelligent people. Second, although religion in atheist society & Wilson, 1995; Stolzenberg et al., 1995). If the rank order                    is   not   likely   to   be   self-enhancing,   it   probably   continues   to of religiosity is stable, then its relation to intelligence should              provide functions such as compensatory control, better self- also be stable.                                                                 regulation, and a means of reducing loneliness through attach-     However, aging (particularly if accompanied by declining                    ment to God. To the extent that intelligent people have less health) is likely to increase awareness of mortality. Religious                 need for these functions, they are less likely to be religious. beliefs can help manage the terror of one’s impending death                     Obviously, these conclusions are a topic for future research.                                               Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 24-----------------------

Zuckerman et al.                                                                                                                                                 23     One last limitation of the present work is the lack of evi-                     might also be expanded to explain lower religiosity of other dence supporting our explanations for the intelligence–reli-                         distinct groups who are in less need of the functions that reli- giosity association. Except for the extreme case of religious                        gion provides. Finally, functional equivalence might be com- fundamentalism (Sherkat, 2010), we clearly posited a causal                         plemented by a concept of functional deficiency. Inasmuch as relation from intelligence to religion and identified specific                      people   possessing   the   functions   that   religion   provides   are mechanisms to account for it. As described below, the edifice                        likely to adopt atheism, people lacking these very functions we built is in need of empirical testing.                                            (e.g., the poor, the helpless) are likely to adopt theism. Conclusion                                                                          Acknowledgments                                                                                     We   thank   the   investigators   who   provided   additional   information The present work comprises two parts. The first part was a                           about their studies at our request. We are particularly grateful to meta-analysis of the relation between intelligence and religi-                      Margie E. Lachman who, in response to our inquiry, sent us a com- osity.   The   second   part   examined   possible   explanations   for             pilation of data from her published articles on the relation between the relation that was observed.                                                     intelligence and personal control beliefs; and to Satoshi Kanazawa     Results of the meta-analysis established a reliable nega-                        for   performing   a   number   of   statistical   analyses   on   his   data,   and tive relation between intelligence and religiosity. It was also                     invariably sending us the results on the same day he received our shown that this relation is weaker in precollege populations                        query. We thank Adam Broitman, Minsi Lai, and Yaritza Perez for                                                                                     their invaluable assistance in the preparation of this article. relative to college and non-college populations. Additional analyses   demonstrated   that   the   relation   is   more   negative                                                                                      Declaration of Conflicting Interests when      religiosity    measures       assessed     religious    beliefs    as opposed   to   religious   behaviors.   It   was   proposed   that   reli-          The author(s) declared no potential conflicts of interest with respect gious beliefs are more likely to represent intrinsic religiosity                    to the research, authorship, and/or publication of this article. (and perhaps “truer” religion), at least for the samples exam- ined   herein. At   the   precollege   level,   the   mean   correlation             Funding (unweighted and weighted) between intelligence and beliefs-                         The author(s) received no financial support for the research, author- based measures of religiosity was −.08; at college and non-                          ship, and/or publication of this article. college levels, the corresponding unweighted and weighted mean correlations ranged from −.20 to −.25.                                          Notes     We reviewed a number of explanations for the negative                            1.   Kanazawa conducted these analyses in response to our request relation between intelligence and religiosity, as well as the                             (S. Kanazawa, personal communication, April 2012). reasons that this association changes with age. All of the pro-                     2.    The formula for correcting r for range restriction is (Sackett & posed explanations involve mediators that are linked to intel-                            Yang, 2000): ligence and religiosity. For example, one of the functional                                                ˆ             (Sx / sx ) rxy                                                                                                            ρ  =        2     2   2      1/ 2 , interpretations was that intelligence and religiosity allow the                                             xy    [1+ r   (Sx  / sx − 1)] individual to exercise better self-regulation, and that intelli-                                                       xy gence leads to lower religiosity because it obviates the need                             where S  and s  are standard deviations of the unrestricted and                                                                                                     x      x for   the   self-regulatory   function   of   religion.   However,   with                 restricted  x   distributions,   respectively;  r    is   the   correlation                                                                                                                                            xy the exception of Shenhav et al. (2011) and Pennycook et al.                               between x and y for the restricted x distribution (see Sackett & (2012), the meta-analyzed studies did not measure the pro-                                Yang, 2000, for a general discussion of range restriction and a posed mediators, thus precluding the possibility of mediation                             classification scheme of range-restriction scenarios). analyses. In addition, we found no longitudinal research that                        3.   In the meta-analysis, we used the raw “uncorrected” correla- examined the relation between intelligence and religiosity at                             tion that Bertsch and Pesta (2009) reported.                                                                                     4.    The   formula   for   converting  r   to   Cohen’s   d   is   (Rosenthal, several     time    points.   These     limitations     can   be   overcome                                                                                            1991): through   future   research   that   utilizes   a   longitudinal   design                                                                                                                             2r and assesses intelligence, religiosity, and the proposed medi-                                                       d =          . ators. Such research might shed light on the causal direction                                                               1− r2 of the intelligence–religiosity relation and on our proposed explanations for this relation.                                                      5.   The trim and fill method identified r       −.23 for the college group,                                                                                           which becomes r         −.33, after correction for range restriction.     On a more general level, the functional approach to reli-                                                                                           This latter value is higher than weighted and unweighted mean gion (Sedikides, 2010) is in its infancy. In future, the list of                                                                                           correlations in the non-college group. However, this value is functions is likely to be expanded and the relations among                                hypothetical and should be treated with caution. functions  are  likely  to  be  elaborated.  It  remains  to  be  seen               6. Means on the 1-to-3 scale were rescaled to means on the 0-to-4 whether   higher   intelligence   confers   not   only   the   functions                  scale by subtracting one point from each mean and multiplying discussed in this paper but also functions that are yet to be                             the difference by 2 (e.g., a mean of 3 on the 1-to-3 scale will discovered. In addition, the concept of functional equivalence                            become a mean of 4 on the 0-to-4 scale); standard deviations                                             Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 25-----------------------

24                                                                                                          Personality and Social Psychology Review XX(X)       of scores on the 1-to-3 scale were multiplied by 2 to accom-                     Beckwith, B. P. (1986). The effect of intelligence on religious faith.       plish the same rescaling procedure.                                                   Free Inquiry, 6, 46-53. 7.   Note   also   the   paradox   that   our   emphasis   on   intrinsic   religios-  Beit-Hallahmi, B., & Argyle, M. (1997). The psychology of religious       ity creates. On one hand, we suggest that it is intrinsic reli-                       behavior, belief, and experience. New York, NY: Routledge.       giosity    (aka   religious   beliefs)   that  provides    the   functions       Bell, P. (2002, February). Would you believe it? Mensa Magazine ,       common to religiosity and intelligence. On the other hand, any                         12-13. Retrieved from       discussion   of   the   functions   that   religion   may   provide,   treats    *Bender, I. E. (1968). A longitudinal study of church attenders and       religiosity   as   extrinsic   rather   than   intrinsic. Additional   com-           nonattenders. Journal   for   the   Scientific   Study   of   Religion ,  7,       plications arise because of the proposed distinction between                          230-237. doi:10.2307/1384630       two forms of extrinsic religiosity—social extrinsic orientation                  Bering,   J.   M.   (2006).   The   folk   psychology   of   souls.  Behavioral       (attainment of social benefits) and personal extrinsic orienta-                       &     Brain   Sciences,   29,    453-462.     doi:10.1017/S0140525X0       tion (overcoming personal problems; Gorsuch &McPherson,                               6009101       1989). Interestingly, Flere and Lavric (2008) showed that in                     *Bertsch,   S.,   &   Pesta,   B.   J.   (2009).   The   wonderlic   personnel   test       religious groups other than American Protestant sample, per-                          and   elementary   cognitive   tasks   as   predictors   of   religious   sec-       sonal extrinsic and intrinsic religious orientations form a sin-                      tarianism,     scriptural   acceptance     and    religious   questioning.       gle dimension that is distinct from social extrinsic orientation.                     Intelligence, 37, 231-237. doi:10.1016/j.intell.2008.10.003       Clearly, intrinsic and extrinsic religious orientations are not as               *Blanchard-Fields,   F.,   Hertzog,   C.,   Stein,   R.,   &   Pak,   R.   (2001).       distinct as they appeared to be in Allport and Ross’s (1967)                          Beyond a stereotyped view of older adults’ traditional family       original conceptualization.                                                           values. Psychology and Aging, 16, 483-496. doi:10.1037/0882-                                                                                             7974.16.3.483 References                                                                             Blazys, G. (2009). Personality, intelligence, and marital outcomes                                                                                             (Working Paper). University of Washington. Ackerman,   P.,   Beier,   M.   E.,   &   Boyle,   M.   O.   (2005).   Working         Block, J. H., & Block, J. (1980). The role of ego-control and ego-      memory   and   intelligence:   The   same   or   different   constructs?               resiliency   in   the   organization   of   behavior.   In   W.   A.   Collins      Psychological Bulletin, 131, 30-60.                                                    (Ed.), Minnesota symposium on child psychology (Vol. 13, pp. Allport, G. W., & Ross, J. M. (1967). Personal religious orientation                        39-101). Hillsdale, NJ: Lawrence Erlbaum.      and prejudice. Journal of Personality and Social Psychology ,                     *Bloodgood,   J.   M.,   Turnley,   W.   H.,   &   Mudrack,   P.   (2008).   The      5, 432-433. doi:10.1037/h0021212                                                       influence of ethics instruction, religiosity, and intelligence on Argue,   A.,   Johnson,   D.   R.,   &   White,   L.   K.   (1999).   Age   and             cheating   behavior. Journal   of   Business   Ethics ,  82,   557-571.      religiosity:     Evidence     from    a   three-wave     panel    analysis.            doi:10.1007/s10551-007-9576-0      Journal   for   the   Scientific   Study   of   Religion ,  38,   423-435.        Bloom, P. (2007). Religion is natural. Developmental Science, 10,      doi:10.2307/1387762                                                                                              147-151. doi:10.1111/j.1467-7687.2007.00577.x Argyle,     M.    (1958).   Religious      behaviour .     London,     England:                                                                                        Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R.      Routledge.                                                                                             (2005). Comprehensive meta-analysis, Version 2. Englewood, Argyle, M., & Beit-Hallahmi, B. (1975). The social psychology of                            NJ: Biostat.      religion. London, England: Routledge.                                                                                        Bowlby, J. (1980). Attachment and loss: Vol. 3. Loss . New York, Arnett, J. J. (1997). Young people’s conceptions of the transition                          NY: Basic Books.      to   adulthood.  Youth   &   Society, 29,   3-23.   doi:10.1177/00441                                                                                        Boyer,   P.   (2001). Religion   explained:   The   evolutionary   origin   of       18X97029001001                                                                                             religious thought. New York, NY: Basic Books. Arnett,   J.   J.   (1998).   Learning   to   stand   alone:   The   contemporary                                                                                        Bremner, R. H., Koole, S. L., & Bushman, B. J. (2011). “Pray for      American transition to adulthood in cultural and historical con-                                                                                             those who mistreat you”: Effects of prayer on anger and aggres-      text. Human Development, 41, 295-315. doi:10.1159/000022591                                                                                             sion. Personality and Social Psychology Bulletin, 37, 830-837. Arnett, J. J., & Jensen, L. A. (2002). A congregation of one: Individualized                                                                                             doi:10.1177/0146167211402215      religious beliefs among emerging adults. Journal of Adolescent                                                                                        *Brown, D. G., & Lowe, W. L. (1951). Religious beliefs and per-      Research, 17, 451-467. doi:10.1177/0743558402175002                                                                                             sonality    characteristics    of  college   students.   The    Journal    of Atran, S., & Norenzayan, A. (2004). Religion’s evolutionary land-                                                                                             Social   Psychology, 33,   103-129.   doi:10.1080/00224545.1951      scape:     Counterintuition,     commitment,      compassion,      commu-                                                                                             .9921803      nion. Behavioral & Brain Science, 27, 713-770. doi:10.1017/                                                                                        Brown,   S.   L.,   Nesse,   R.   M.,   House,   J.   S.,   &   Utz,   R.   L.   (2004).       S0140525X04000172                                                                                             Religion and emotional compensation: Results from a prospec- Bandura,   A.   (1997).  Self-efficacy:   The   exercise   of   control.   New                                                                                             tive study of widowhood. Personality and Social Psychology      York, NY: Freeman.                                                                                             Bulletin, 30, 1165-1174. doi:10.1177/0146167204263752 Barratt,    E.  S.   (1985).   Impulsiveness       substraints:   Arousal     and                                                                                        Campitelli,     G.,  &   Labollita,   M.   (2010).    Correlations    of  cogni-      information processing. In J. T. Spence, & C. E. Izard (Eds.),                                                                                             tive   reflection   with   judgements   and   choices.  Judgment   and      Motivations,   emotions   and   personality   (pp.   137-146).   North                                                                                             Decision Making, 5, 182-191.      Holland, Netherlands: Elsevier.                                                                                        *Carlson, H. B. (1934). Attitudes of undergraduate students. The Barrett, J. L. (2004).  Why would anyone believe in God? Walnut                                                                                             Journal of Social Psychology , 5, 202-213. doi:10.1080/00224      Creek, CA: Alta Mira Press.                                                                                             545.1934.9919448 Batson, C. D., Schoenrade, P., & Ventis, W. L. (1993). Religion                                                                                        *Carothers, S. S., Borkowski, J. G., Burke Lefever, J., & Whitman,      and   the   individual:   A   social-psychological   perspective.   New                                                                                             T.   L.   (2005).   Religiosity   and   the   socioemotional   adjustment      York, NY: Oxford University Press.                                                   Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 26-----------------------

Zuckerman et al.                                                                                                                                                       25       of   adolescent   mothers   and   their   children. Journal   of   Family         D’Onofrio, B. M., Eaves, L. J., Murrelle, L., Maes, H. H., & Spilka,      Psychology, 19, 263-275. doi:10.1037/0893-3200.19.2.263                                 B.   (1999).    Understanding       biological   and    social  influences Carroll, J. B. (1993). Human cognitive abilities: A survey of fac-                           on    religious   affiliation,  attitudes,   and   behaviors:    A   behav-       tor   analytic   studies.   Cambridge,   UK:   Cambridge   University                  ior   genetic   perspective. Journal   of   Personality ,  67,   953-984.       Press.                                                                                 doi:10.1111/1467-6494.00079 Cavalli-Sforza, L. L., Feldman, M. W., Chen, K. H., & Dornbusch,                        *Dreger, R. M. (1952). Some personality correlates of religious atti-       S. M. (1982). Theory and observation in cultural transmission.                         tudes, as determined by projective techniques. Psychological       Science, 218, 19-27. doi:10.1126/science.7123211                                       Monographs:        General    and   Applied ,  66,    1-18.   doi:10.1037/ *Ciesielski-Kaiser,   H.   A.   (2005). Religious   identity   development                   h0093604       and personality (Unpublished doctoral dissertation). Southern                     Dronkers, T. (2002). Bestaat er en samenhang tussen echtscheiding       Illinois University Carbondale, Carbondale.                                            en intelligentie? [Is there a relation between divorce and intel- Cohen,     A.   B.,  Hall,   D.  E.,  Koenig,    H.   G.,  &   Meador,     K.   G.           ligence?]. Mens en Maatschappij , 77, 25-42.       (2005). Social versus individual motivation: Implications for                     Duval, S., & Tweedie, R. (2000). Trim and fill: A simple funnel-       normative      definitions    of  religious   orientation.   Personality               plot-based method of testing and adjusting for publication bias       and     Social    Psychology      Review,    9,    48-61.    doi:10.1207/              in meta-analysis. Biometrics, 56, 455-463. doi:10.1111/j.0006-       s15327957pspr0901_4                                                                    341X.2000.00455.x Cokely,   E.   T.,   &   Kelley,   C.   M.   (2009).   Cognitive   abilities   and      Epley, N., Akalis, S., Waytz, A., & Cacioppo, J. T. (2008). Creating       superior decision making under risk: A protocol analysis and                           social connection through inferential reproduction: Loneliness      process model evaluation. Judgment and Decision Making , 4,                             and    perceived     agency    in   gadgets,   gods,    and   greyhounds.       20-33.                                                                                 Psychological       Science,    19,     114-120.     doi:10.1111/j.1467- *Corey, S. M. (1940). Changes in the opinions of female students                             9280.2008.02056.x       after one year at a university. The Journal of Social Psychology,                 Epstein,    S.  (1994).   Integration    of  the  cognitive    and   psychody-       11, 341-351. doi:10.1080/00224545.1940.9918754                                         namic     unconscious.     American       Psychologist ,   49,      709-724. *Cottone, J., Drucker, P., & Javier, R. A. (2007). Predictors of moral                       doi:10.1037/0003-066X.49.8.709       reasoning: Components of executive functioning and aspects                        Exline,    J.  J.  (2002).   Stumbling     blocks    on   the  religious    road:       of religiosity. Journal for the Scientific Study of Religion , 46,                     Fractured relationships, nagging vices, and the inner struggle       37-53. doi:10.1111/j.1468-5906.2007.00339.x                                            to   believe. Psychological   Inquiry,  13,   182-189.   doi:10.1207/ Coyle, T. R., & Pillow, D. R. (2008). SAT and ACT predict college                            S15327965PLI1303_03       GPA after removing g. Intelligence , 36, 719-729. doi:10.1016/j.                  *Feather, N. T. (1964). Acceptance and rejection of arguments in       intell.2008.05.001                                                                     relation to attitude strength, critical ability, and intolerance of *Crossman,   A.   M.   (2001). Predicting   suggestability:   The   role   of                inconsistency. Journal   of   Abnormal   and   Social   Psychology ,       individual differences and socialization (Unpublished doctoral                         69, 127-136. doi:10.1037/h0046290       dissertation). Cornell University, Ithaca, NY.                                    *Feather, N. T. (1967). Evaluation of religious and neutral arguments Deary,     I.  J.,  Strand,   S.,  Smith,    P.,  &   Fernandes,     C.   (2007).            in religious and atheist student groups. Australian Journal of       Intelligence and educational achievement. Intelligence , 35, 13-                       Psychology, 19, 3-12. doi:10.1080/00049536708255556       21. doi:10.1016/j.intell.2006.02.001                                              Feingold, A. (1983). The validity of the information and vocabu- De   Haan,   L.   G.,   &   Schulenberg,   J.   (1997).   The   covariation   of             lary subtests of the WAIS for predicting college achievement.       religion    and   politics  during    the  transition   to  young    adult-            Educational and Psychological Measurement, 43, 1127-1131.       hood:    Challenging      global   identity   assumptions.    Journal     of           doi:10.1177/001316448304300422      Adolescence , 20, 527-552. doi:10.1006/jado.1997.0108                              Flere, S., & Lavric, M. (2008). Is intrinsic religious orientation a *Deptula,   D.   P.,   Henry,   D.   B.,   Shoeny,   M.   E.,   &   Slavick,   J.   T.       culturally specific American Protestant concept? The fusion of       (2006). Adolescent sexual behavior and attitudes: A costs and                          intrinsic and extrinsic religious orientation among non-Protes-      benefits   approach.  Journal   of   Adolescent   Health ,  38,   35-43.                tants.  European   Journal   of   Social   Psychology,  38,   521-530.       doi:10.1016/j.jadohealth.2004.08.026                                                   doi:10.1002/ejsp.437 de Wit, H., Flory, J. D., Acheson, A., McCloskey, M., & Manuck,                         *Foy, D. W. (1975). A study of the relationship between intelli -       S.   B.   (2007).   IQ   and   non-planning   impulsivity   are   indepen-             gence, religiosity and locus of control (Unpublished doctoral       dently    associated     with   delay    discounting     in  middle-aged               dissertation). University of Southern Mississippi, Hattiesburg.       adults. Personality   and   Individual   Differences,  42,   111-121.             *Francis, L. J. (1979). School influence and pupil attitude towards       doi:10.1016/j.paid.2006.06.026                                                         religion. British Journal of Educational Psychology, 49, 107- *Dodrill, C. B. (1976). Brain functions of Christians and non-Chris-                         123. doi:10.1111/j.2044-8279.1979.tb02405.x       tians. Journal of Psychology & Theology , 4, 280-285.                             *Francis, L. J. (1997). Personal and social correlates of the “closed Dolan,   M.,   &   Fullam,   R.   (2004).   Behavioural   and   psychometric                 mind” among 16 year old adolescents in England. Educational       measures of impulsivity in a personality disordered population.                        Studies, 23, 429-437. doi:10.1080/0305569970230308       The   Journal   of   Forensic   Psychiatry   &   Psychology,  15,   426-          *Francis,   L.   J.   (1998).   The   relationship   between   intelligence   and       450. doi:10.1080/14789940410001721048                                                  religiosity among 15-16 year olds. Mental Health, Religion and Dom,      G.,  De    Wilde,    B.,  Hulstijn,    W.,   &   Sabbe,    B.   (2007).            Culture, 1, 185-196. doi:10.1080/13674679808406508       Dimensions       of   impulsive     behavior     in  abstinent     alcohol-       *Francis, L. J., Pearson, P. R., & Stubbs, M. T. (1985). Personality       ics. Personality      and    Individual    Differences,    42,      465-476.           and religion among low ability children in residential special       doi:10.1016/j.paid.2006.08.007                                                         schools. The British Journal of Mental Subnormality, 31, 41-45.                                              Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 27-----------------------

26                                                                                                           Personality and Social Psychology Review XX(X) *Franzblau,   A.   N.   (1934). Religious   belief   and   character   among            *Gragg, D. B. (1942). Religious attitudes of denominational college      Jewish      adolescents .     New     York,    NY:    Teachers      College,            students. The Journal of Social Psychology, 15, 245-254. doi:1      Columbia University Bureau of Publications.                                             0.1080/00224545.1942.9921534 Frederick,     S.   (2005).    Cognitive     reflection   and   decision     mak-       Graham, J., & Haidt, J. (2010). Beyond beliefs: Religions bind individ-      ing.    Journal       of    Economic       Perspectives ,     19,      25-42.           uals into moral communities. Personality and Social Psychology      doi:10.1257/089533005775196732                                                          Review, 14, 140-150. doi:10.1177/1088868309353415 Frey, M. C., & Detterman, D. K. (2004). Scholastic assessment or                        Granqvist,   P.,   Mikulincer,   M.,   &   Shaver,   P.   R.   (2010).   Religion      g?   The   relationship   between   the   Scholastic   Assessment   Test                as    attachment:     Normative     processes     and   individual    differ-      and general cognitive ability. Psychological Science, 15, 373-                          ences. Personality and Social Psychology Review, 14, 49-59.      378. doi:10.1111/j.0956-7976.2004.00687.x                                               doi:10.1177/1088868309348618 Funk, R. B., & Willits, F. K. (1987). College attendance and atti-                      Greene,   A.   L.,   Wheatley,   S.   M.,   &   Aldava,   J.   F.   I.   V.   (1992).      tude change: A panel study, 1970-81. Sociology of Education,                             Stages   on   life’s   way:   Adolescents’   implicit   theories   of   the       60, 224-231. doi:10.2307/2112558                                                       life   course.   Journal     of  Adolescent      Research ,   7,     364-381. Gabriel, M. T., Critelli, J. W., & Ee, J. S. (1994). Narcissistic illusions                  doi:10.1177/074355489273006      in self-evaluations of intelligence and attractiveness. Journal                    Grover, D. R., & Hertzog, C. (1991). Relationships between intel-      of   Personality,    62,   143-155.     doi:10.1111/j.1467-6494.1994.                   lectual control beliefs and psychometric intelligence in adult-      tb00798.x                                                                               hood. Journal of Gerontology , 46, 109-115. *Gallup   Organization.   (1965).  Gallup/Catholic   digest   poll   #671:              Gudjonsson,   G.   H.   (1987).   A   parallel   form   of   the   Gudjonsson      Religion   in   America   [Data   file].   Available   from   http://www.                Suggestibility   Scale. British   Journal   of   Clinical   Psychology,                                                                  26, 215-221. *Gallup Organization. (1986). Gallup poll #1986-848: Elections                          Gudjonsson,   G.   H.,   &   Clare,   I.   C.   H.   (1995).   The   relationship       [Data    file].  Available     from   http://www.ropercenter.uconn.                    between confabulation and intellectual ability, memory, inter-      edu/                                                                                    rogative     suggestibility     and   acquiescence.     Personality      and Gervais,   W.   M.,   &   Henrich,   J.   (2010).   The   Zeus   problem:   Why              Individual Differences , 19, 333-338.      representational   content   biases   cannot   explain   faith   in   gods.        Guthrie, S. (1993). Faces in the crowd: A new theory of religion .      Journal of Cognition and Culture , 10, 383-389. doi:10.1163/                            New York, NY: Oxford University Press.       156853710X531249                                                                  *Hadden,   J.   K.   (1963).   An   analysis   of   some   factors   associated Gervais, W. M., & Norenzayan, A. (2012a). Analytic thinking pro-                             with    religion    and   political  affiliation   in  a  college   popula-      motes religious disbelief. Science, 336, 493-496. doi:10.1126/                          tion. Journal for the Scientific Study of Religion , 2, 209-216.       science.1215647                                                                        doi:10.2307/1385072 Gervais, W. M., & Norenzayan, A. (2012b). Reminders of secular                          Hall, D. L., Matz, D. C., & Wood, W. (2010). Why don’t we prac-      authority reduce believers’ distrust of atheists. Psychological                         tice what we preach? A meta-analytic review of religious rac-      Science. Advance online publication.                                                    ism. Personality and Social Psychology Review, 14, 126-139. Gervais, W. M., Shariff, A. F., & Norenzayan, A. (2011). Do you                              doi:10.1177/1088868309352179      believe in atheists? Distrust is central to anti-atheist prejudice.                Harris,    J.  A.  (2004).   Measured     intelligence,    achievement,     open-      Journal     of  Personality     and   Social   Psychology ,    101,     1189-           ness   to   experience,   and   creativity. Personality   and   Individual       1206. doi:10.1037/a0025882                                                             Differences, 36, 913-929. doi:10.1016/S0191-8869(03)00161-2 Gervais, W. M., Willard, A., Norenzayan, A., & Henrich, J. (2011).                      Hedges,   L.   V.,   &   Vevea,   J.   L.   (1998).   Fixed-and   random-effects      The     cultural   transmission     of  faith:   Why    natural   intuitions            models in meta-analysis. Psychological Methods, 3, 486-504.      and memory biases are necessary, but insufficient, to explain                           doi:10.1037/1082-989X.3.4.486      religious     belief. Religion,    41,    389-410.     doi:10.1080/00487           Herrnstein, R. J., & Murray, C. (1994). The bell curve: Intelligence      21X.2011.604510                                                                         and   class   structure   in   American   life.   New   York,   NY:   Free Gibson,     H.   M.,   Francis,   L.   J.,  &  Pearson,    P.  R.   (1990).   The            Press.      relationship       between      social   class    and    attitude    toward        *Hoge, D. R. (1969). College student’s religion: A study of trends      Christianity      among     fourteen    and    fifteen-year    old   adoles-            in attitudes and behavior (Unpublished doctoral dissertation).      cents.  Personality   and   Individual   Differences,  11,   631-635.                   Harvard University, Cambridge, MA.      doi:10.1016/0191-8869(90)90046-T                                                   Hoge, D. R. (1974). Commitment on campus: Changes in religion *Gilliland,   A.   R.   (1940).   The   attitude   of   college   students   toward          and values over five decades. Philadelphia, PA: Westminster      God and the church. The Journal of Social Psychology, 11, 11-                           Press.       18. doi:10.1080/00224545.1940.9918729                                             Hoge, D. R., Johnson, B., & Luidens, D. A. (1993). Determinants Gorsuch,       R.    L.   (1988).     Psychology       of    religion.   Annual              of    church    involvement      of  young     adults   who    grew    up   in      Review      of  Psychology,     39,    201-221.     doi:10.1146/annurev.                Presbyterian      churches.    Journal     for  the   Scientific   Study   of      ps.39.020188.001221                                                                     Religion, 32, 242-255. doi:10.2307/1386663 Gorsuch,   R.   L.,   &   McPherson,   S.   E.   (1989).   Intrinsic/Extrinsic          Hoge, D. R., Petrillo, G. H., & Smith, E. I. (1982). Transmission      Measurement: I/E-Revised and Single-Item Scales. Journal for                            of    religious    and    social   values    from    parents    to   teenage      the Scientific Study of Religion, 28, 348-354.                                          children.    Journal     of   Marriage     and    Family ,  44,      569-580. Gottfredson,      L.  S.  (1997).    Mainstream      science   on   intelligence:            doi:10.2307/351580      An   editorial   with   52   signatories,   history,   and   bibliography.         Holahan, C. K., & Sears, R. R. (1995).  The gifted group in later      Intelligence , 24, 13-23. doi:10.1016/S0160-2896(97)90011-8                             maturity. Stanford, CA: Stanford University Press.                                                    Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 28-----------------------

Zuckerman et al.                                                                                                                                                        27 Holley,     P.,  Yabiku,    S.,  &   Benin,    M.   (2006).   The    relationship       Kelemen,   D.   (2004).   Are   children   “intuitive   theists?”   Reasoning       between intelligence and divorce. Journal of Family Issues , 27,                        about purpose and design in nature. Psychological Science, 15,       1723-1748. doi:10.1177/0192513X06291887                                                295-301. doi:10.1111/j.0956-7976.2004.00672.x *Horowitz, J. L., & Garber, J. (2003). Relation of intelligence and                     Kelemen, D., & Rosset, E. (2009). The human function compunc-       religiosity to depressive disorders in offspring of depressed and                      tion: Teleological explanations in adults. Cognition, 111, 138-       nondepressed   mothers. Journal   of   the   American   Academy   of                    143. doi:10.1016/j.cognition.2009.01.001       Child & Adolescent Psychiatry, 42, 578-586. doi:10.1097/01.                       Kirkpatrick, L. A. (1998). God as a substitute attachment figure:       CHI.0000046831.09750.03                                                                A   longitudinal   study   of   adult   attachment   style   and   religious *Howells, T. H. (1928). A comparative study of those who accept                               changes in college students. Personality and Social Psychology       as   against   those   who   reject   religious   authority.  University   of          Bulletin, 24, 961-973. doi:10.1177/0146167298249004      Iowa Studies in Character , 2, 1-80.                                               Kirkpatrick, L. A. (2005). Attachment, evolution and the psychol- Hunsberger, B. (1978). The religiosity of college students: Stability                        ogy of religion. New York, NY: Guilford.       and change over years at university. Journal for the Scientific                   Kirkpatrick,      L.  A.,   &   Shaver,    P.  R.   (1992).    An   attachment-       Study of Religion, 17, 159-164. doi:10.2307/1386159                                    theoretical     approach     to  romantic    love   and   religious   belief. Hunt, E. (2011). Human intelligence . New York, NY: Cambridge                                Personality      and    Social   Psychology      Bulletin,  18,      266-275.       University Press.                                                                       doi:10.1177/0146167292183002 Ingersoll-Dayton, B., Krause, N., & Morgan, D. (2002). Religious                        Kirkpatrick,      L.   A.,  Shillito,   D.   J.,  &   Kellas,    S.  L.   (1999).       trajectories and transitions over the life course. International                       Loneliness,   social   support,   and   perceived   relationships   with      Journal      of   Aging     &    Human      Development ,      55,     51-70.            God. Journal of Social and Personal Relationships , 16, 513-       doi:10.2190/297Q-MRMV-27TE-VLFK                                                         522. doi:10.1177/0265407599164006 *Inzlicht, M., McGregor, I., Hirsh, J. B., & Nash, K. (2009). Neural                    Koenig,   K.   A.,   Frey,   M.   C.,   &   Detterman,   D.   K.   (2008).   ACT       markers   of   religious   conviction.  Psychological   Science,  20,                   and    general    cognitive     ability. Intelligence ,    36,      153-160.       385-392. doi:10.1111/j.1467-9280.2009.02305.x                                           doi:10.1016/j.intell.2007.03.005 Jensen, A. R. (1998).  The g factor: The science of mental ability.                     Koenig,   L.   B.,   McGue,   M.,   &   Iacono,   W.   G.   (2008).   Stability       Westport, CT: Praeger.                                                                  and    change     in  religiousness     during    emerging      adulthood. Jones, H. E., & Bayley, N. (1941). The Berkeley Growth Study.                                Developmental   Psychology,  44,   532-543.   doi:10.1037/0012-       Child Development, 12, 167-173. doi:10.2307/1125347                                     1649.44.2.532 *Jones,   V.   (1938).   Attitudes   of   college   students   and   the   changes      Koole, S. L., McCullough, M. E., Kuhl, J., & Roelofsma, P. H. M.       in   such   attitudes   during   four   years   in  college.  Journal     of           P. (2010). Why religion’s burdens are light: From religiosity      Educational Psychology, 29, 14-25. doi:10.1037/h0055012                                 to implicit self-regulation. Personality and Social Psychology Judge, T. A., Hurst, C., & Simon, L. S. (2009). Does it pay to be                            Review, 14, 95-107. doi:10.1177/1088868309351109       smart,    attractive,  or   confident    (or  all  three)?  Relationships         *Kosa,   J.,   &   Schommer,   C.   O.   (1961).   Religious   participation,       among      general   mental    ability,  physical    attractiveness,    core           religious     knowledge,     and   scholastic    aptitude:   An   empirical       self-evaluations, and income. Journal of Applied Psychology ,                           study. Journal   for   the   Scientific   Study   of   Religion ,  1,   88-97.       94, 742-755. doi:10.1037/a0015497                                                       doi:10.2307/1385179 Judge,   T.   A.,   Jackson,   C.   L.,   Shaw,   J.   C.,   Scott,   B.   A.,   &   Rich, Lachman, M. E. (1983). Perceptions of intellectual aging? Antecedents       B.   L.  (2007).    Self-efficacy    and   work-related      performance:               or   consequence      of   intellectual   functioning?    Developmental       The integral role of individual differences. Journal of Applied                        Psychology, 19, 482-498. doi:10.1037/0012-1649.19.4.482      Psychology, 92, 107-127. doi:10.1037/0021-9010.92.1.107                            Lachman,   M.   E.,   Baltes,   P.,   Nesselroade,   J.   R.,   &   Willis,   S.   L. Kahneman,   D.   (2003).   A   perspective   on   judgment   and   choice:                    (1982).    Examination       of  personality-ability     relationships     in       Mapping bounded rationality. American Psychologist , 58, 697-                          the    elderly:   The   role   of  the   contextual    (interface)   assess-       720. doi:10.1037/0003-066X.58.9.697                                                    ment mode. Journal of Research in Personality , 16, 485-501. Kanazawa, S. (2010a). Why liberals and atheists are more intelligent.                         doi:10.1016/0092-6566(82)90007-1       Social Psychology Quarterly,  73, 33-57. doi:10.1177/019027                       Larsen,   L.,   Hartmann,   P.,   &   Nyborg,   H.   (2008).   The   stability   of       2510361602                                                                              general    intelligence     from    early   childhood     to  middle-age. Kanazawa,   S.   (2010b).   Evolutionary   psychology   and   intelligence                   Intelligence , 36, 29-34. doi:10.1016/j.intell.2007.01.001       research.  American       Psychologist ,   65,   279-289.     doi:10.1037/        Larson, E. J., & Witham, L. (1998). Leading scientists still reject       a0019378                                                                                God. Nature, 394, 313. doi:10.1038/28478 Kay,     A.  C.,   Gaucher,     D.,  McGregor,      I.,  &   Nash,   K.   (2010).       Laurin,   K.,   Kay,   A.   C.,   &   Moscovitch,   D.   A.   (2008).   On   the       Religious     belief   as  compensatory       control.  Personality     and            belief in god: Towards an understanding of the emotional       Social   Psychology   Review,  14,   37-48.   doi:10.1177/10888683                      substrates of compensatory control. Journal of Experimental       09353750                                                                               Social      Psychology,      44,     1559-1562.       doi:10.1016/j.jesp. Kay,   A.   C.,   Gaucher,   D.,   Napier,   J.   L.,   Callan,   M.   J.,   &   Laurin,     2008.07.007       K.   (2008).   God   and   the   government:   Testing   a   compensa-            Lefkowitz,        E.   S.    (2005).     “Things      have     gotten     better”:       tory   control   mechanism   for   the   support   of   external   systems.            Developmental changes among emerging adults after the tran-      Journal      of  Personality    and    Social   Psychology ,    95,    18-35.            sition to university. Journal of Adolescent Research , 20, 40-63.       doi:10.1037/0022-3514.95.1.18                                                           doi:10.1177/0743558404271236 Kay, A. C., Moscovitch, D. A., & Laurin, K. (2010). Randomness,                         Leuba, J. H. (1916). The belief in God and immortality: A psycho-       attributions    of   arousal,   and    belief   in  God.   Psychological                logical,   anthropological,   and   statistical   study.   Boston,   MA:       Science, 21, 216-218. doi:10.1177/0956797609357750                                      Sherman, French.                                               Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 29-----------------------

28                                                                                                          Personality and Social Psychology Review XX(X) *Lewis, G. J., Ritchie, S. J., & Bates, T. C. (2011). The relationship                      findings and theoretical developments. American Psychologist ,      between intelligence and multiple domains of religious belief:                         67, 130-159. doi:10.1037/a0026699      Evidence from a large adult US sample. Intelligence , 39, 468-                    *Nokelainen, P., & Tirri, K. (2010). Role of motivation in the moral      472. doi:10.1016/j.intell.2011.08.002                                                  and religious judgment of mathematically gifted adolescents. Long, H. B. (1972). Differences between content conformity and                              High Ability Studies, 21, 101-116. doi:10.1080/13598139.201      visual   conformity.  Adult   Education   Quarterly ,  12,   282-290.                  0.525343      doi:10.1177/074171367202200403                                                    Norenzayan, A., & Gervais, W. M. (2012). The cultural evolution Lynch,   A.   D.,   &   Clark,   P.   (1985).   Relationship   of   self-esteem,            of religion. In E. Slingerland, & M. Collard (Eds.),  Creating      IQ, and task performance for a sample of USA undergradu-                               consilience: Integrating the sciences and the humanities (pp.       ates.  Psychological        Reports ,   56,    955-962.     doi:10.2466/              243-265). Oxford, UK: Oxford University Press.      pr0.1985.56.3.955                                                                 Norenzayan, A., Gervais, W. M., & Trzesniewski, K. H. (2012). Lynn, R., Harvey, J., & Nyborg, H. (2009). Average intelligence                             Mentalizing deficits constrain belief in a personal God. PLoS      predicts atheism rates across 137 nations. Intelligence , 37, 11-                      ONE, 7(5), e36880.       15. doi:10.1016/j.intell.2008.03.004                                             Nyborg, H. (2009). The intelligence-religiosity nexus: A represen- Madsen, G. E., & Vernon, G. M. (1983). Maintaining the faith during                         tative study of white adolescent Americans. Intelligence , 37,      college: A study of campus religious group participation. Review                       81-93. doi:10.1016/j.intell.2008.08.003      of Religious Research, 25, 127-141. doi:10.2307/3511490                           Obrecht, N. A., Chapman, G. B., & Gelman, R. (2009). An encoun- Martel,     J.,  McKelvie,     S.  J.,  &   Standing,    L.  (1987).    Validity            ter   frequency   account   of   how   experience   affects   likelihood      of    an   intuitive   personality     scale:   Personal    responsibility             estimation. Memory   &   Cognition , 37,   632-643.   doi:10.3758/       as   a  predictor   of  academic     achievement.  Educational         and            MC.37.5.632      Psychological        Measurement,      47,    1153-1163.      doi:10.1177/        O’Connor, T. P., Hoge, D. R., & Alexander, E. (2002). The relative       0013164487474033                                                                      influence of youth and adult experiences on personal spiritual- Mayer, J. D., Caruso, D. R., Panter, A. T., & Salovey, P. (2012).                           ity and church involvement. Journal for the Scientific Study of      The     growing      significance    of   hot   intelligence.   American               Religion, 41, 723-732. doi:10.1111/1468-5906.00157      Psychologist, 67, 502-503. doi:10.1037/a0029456                                   Osborn, D. P. (2005).A correlational study of conformity and intelli- Mayer, J. D., Caruso, D. R., & Salovey, P. (1999). Emotional intelli-                       gence (master’s thesis). Retrieved from ProQuest Dissertations      gence meets traditional standards for an intelligence. Intelligence,                   and Theses Full Text Database. (Order No. 1427613)      27, 267-298. doi:10.1016/S0160-2896(99)00016-1                                    Pathare, S. S., & Kanekar, S. (1990). Self-esteem, intelligence and *McCullough,   M.   E.,   Enders,   C.   K.,   Brion,   S.   L.,   &   Jain,   A.   R.      academic achievement of Indian college students in relation to       (2005).   The   varieties   of   religious   development   in   adulthood:            their sex roles. The Irish Journal of Psychology, 11, 31-40.      A   longitudinal   investigation   of   religion   and   rational   choice.       Patton,   J.   H.,   Stanford,   M.   S.,   &   Barratt,   E.   S.   (1995).   Factor      Journal     of  Personality     and   Social   Psychology ,    89,    78-89.           structure     of   the  Barratt    Impulsiveness      Scale.  Journal     of      doi:10.1037/0022-3514.89.1.78                                                          Clinical       Psychology,      51,     768-774.       doi:10.1002/1097- McCullough, M. E., & Willoughby, B. L. B. (2009). Religion, self-                           4679(199511)51:6<768::AID-JCLP2270510607>3.0.CO;2-1      regulation,   and   self-control:   Associations,   explanations,   and           *Pennycook, G., Cheyne, J. A., Seli, P., Koehler, D. J., & Fugelsang,      implications. Psychological Bulletin, 135, 69-93. doi:10.1037/                         J.   A.   (2012).   Analytic   cognitive   style   predicts   religious   and       a0014213                                                                              paranormal       belief. Cognition,     123,    335-346.    doi:10.1016/j. McGregor, I., Haji, R., Nash, K. A., & Teper, R. (2008). Religious                          cognition.2012.03.003      zeal     and   the   uncertain     self. Basic     and    Applied    Social       Pesta,   B.   J.,   McDaniel,   M.   A.,   &   Bertsch,   S.   (2010).   Toward   an      Psychology, 30, 183-188. doi:10.1080/01973530802209251                                 index of well-being for the fifty U.S. states. Intelligence , 38, McGregor, I., Nash, K. A., & Prentice, M. (2009). Religious convic-                          160-163. doi:10.1016/j.intell.2009.09.006       tion after uncertainty threat as reactive approach motivation:                   Pesta, B. J., & Poznanski, P. J. (2008). Black-white differences on IQ       The eager chase of a transcendent tale? Unpublished manu-                             and grades: The mediating role of elementary cognitive tasks.       script, York University, Toronto, Ontario, Canada.                                    Intelligence , 36, 323-329. doi:10.1016/j.intell.2007.07.004 Meng, X.-L., Rosenthal, R., & Rubin, D. B. (1992). Comparing cor-                      *Poythress, N. G. (1975). Literal, antiliteral, and mythological reli-      related   correlation   coefficients.  Psychological   Bulletin,  111,                 gious orientations. Journal for the Scientific Study of Religion ,       172-175. doi:10.1037/0033-2909.111.1.172                                              14, 271-284. doi:10.2307/1384909 Miller, L. M. S., & Lachman, M. E. (2000). Cognitive performance                       Prenda, K. M., & Lachman, M. E. (2001). Planning for the future:       and the role of control beliefs in midlife.Aging, Neuropsychology,                    A life management strategy for increasing control and life sat-      and Cognition, 7, 69-85. doi:10.1076/1382-5585(200006)7:2;1–                           isfaction   in   adulthood.  Psychology   and   Aging,  16,   206-216.      U;FT069                                                                                doi:10.1037/0882-7974.16.2.206 Millet, K., & Dewitte, S. (2007). Non-conformity may be hidden                         Quensel, C. T. E. (1958). The interrelations of marital status, fertil-      driver behind relation. British Medical Journal, 334, 327-328.                         ity, family size and intelligence test scores. Population Studies:      doi:10.1136/bmj.39121.870139.1F                                                        A Journal of Demography , 11, 234-250. doi:10.2307/2172529 Nisbett,    R.  E.,  Aronson,     J.,  Blair,  C.,  Dickens,    W.,   Flynn,   J.,     Randolph-Seng,        B.,   &   Nielsen,    M.   E.  (2007).    Honesty:     One      Halpern, D. F., & Turkheimer, E. (2012a). Group differences                            effect     of   primed     religious    representations.    International      in IQ are best understood as environmental in origin. American                         Journal      for   the   Psychology      of   Religion ,   17,      303-315.      Psychologist, 67, 503-504. doi:10.1037/a0029772                                        doi:10.1080/10508610701572812 Nisbett,    R.  E.,  Aronson,     J.,  Blair,  C.,  Dickens,    W.,   Flynn,   J.,     *Räsänen, A., Tirri, K., & Nokelainen, P. (2006). The moral and      Halpern, D. F., & Turkheimer, E. (2012b). Intelligence: New                            religious   reasoning   of   gifted   adolescence.   In   K.   Tirri   (Ed.),                                                   Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 30-----------------------

Zuckerman et al.                                                                                                                                                        29      Nordic perspectives on religion, spirituality and identity (pp.                    Sedikides, C., & Gebauer, J. E. (2010). Religiosity as self-enhance-       97-111). Helsinki, Finland: University of Helsinki.                                    ment: A meta-analysis of the relation between socially desirable Reeve,      C.   L.   (2009).    Expanding      the   g-nexus:     Further    evi-           responding and religiosity. Personality and Social Psychology       dence    regarding     the  relations   among     national   IQ,   religios-           Review, 14, 17-36. doi:10.1177/1088868309351002       ity   and   national   health   outcomes. Intelligence ,  37,   495-505.          Shamosh,   N.   A.,   &   Gray,   J.   R.   (2008).   Delay   discounting   and       doi:10.1016/j.intell.2009.06.003                                                       intelligence:      A   meta-analysis.     Intelligence ,    36,      289-305. Reeve,   C.   L.,   &   Basalik,   D.   (2011).   A   state   level   investigation           doi:10.1016/j.intell.2007.09.004       of the associations among intellectual capital, religiosity and                   *Shenhav, A., Rand, D. G., & Greene, J. D. (2012). Divine intu-       reproductive      health.  Intelligence ,   39,   64-73.    doi:10.1016/j.             ition:   Cognitive     style  influences    belief   in  God.  Journal     of       intell.2010.09.007                                                                     Experimental Psychology: General, 141, 423-428. doi:10.1037/ Reiss, S. (2004). The sixteen strivings for God. Zygon, 39, 303-320.                          a0025391       doi:10.1111/j.1467-9744.2004.00575                                                Sherkat,   D.   E.   (1998).   Counterculture   or   continuity:   Competing Rhodes,   N.,   &   Wood,   W.   (1992).   Self-esteem   and   intelligence                  influences   on   baby   boomers’   religious   orientation   and   par-       affect influenceability: The mediating role of message recep-                          ticipation.     Social     Forces,    76,     1087-1114.      doi:10.2307/       tion. Psychological Bulletin, 111, 156-171. doi:10.1037/0033-                           3005704       2909.111.1.156                                                                    *Sherkat, D. E. (2010). Religion and verbal ability. Social Science Richards, M. (2002). IQ and the wealth of nations: Richard Lynn                              Research, 39, 2-13. doi:10.1016/j.ssresearch.2009.05.007       and Tatu Vanhanen. Intelligence , 30, 574-576.                                    *Sherkat, D. E. (2011). Religion and scientific literacy in the United Ridgell,   S.   D.,   &   Lounsbury,   J.   W.   (2004).   Predicting   academic              States. Social Science Quarterly, 92, 1134-1150. doi:10.1111/       success: General intelligence, “Big Five” personality traits, and                      j.1540-6237.2011.00811.x       work drive. College Student Journal, 38, 607-618.                                 Sherkat,     D.   E.,  &   Wilson,    J.  (1995).   Preferences,     constraints, Rosenthal, R. (1991). Meta-analytic procedures for social research                            and    choices   in  religious    markets:    An   examination      of  reli-       (Rev. ed.). Newbury Park, CA: SAGE.                                                     gious   switching   and   apostasy.  Social   Forces,  73,   993-1026. Rosenthal, R. (1995). Writing meta-analytic reviews. Psychological                            doi:10.2307/2580555      Bulletin, 118, 183-192. doi:10.1037/0033-2909.118.2.183                            *Sinclair, R. D. (1928). A comparative study of those who report Rounding, K., Lee, A., Jacobson, J. A., & Ji, L. (2012). Religion                            the experience of the divine presence and those who do not.       replenishes   self-control. Psychological   Science, 23,   635-642.                     University of Iowa Studies in Character, 2, 1-62.       doi:10.1177/0956797611431987                                                      Smith, S., Murphy, D. B., & Wheeler, L. S. (1964). Relations of Sackett,   P.   R.,   Kuncel,   N.   R.,   Arneson,   J.   J.,   Cooper,   S.   R.,   &      intelligence and authoritarianism to behavioral contagion and       Waters, S. D. (2009). Socioeconomic status and the relationship                         conformity.     Psychological       Reports,   14,    248.   doi:10.2466/       between SAT and freshman GPA: An analysis of data from 41                              pr0.1964.14.1.248       colleges and universities (College Board Research Report No.                      *Southern, M. L., & Plant, W. T. (1968). Personality characteristics       2009-1). New York, NY: The College Board.                                               of   very   bright   adults.  The   Journal   of   Social   Psychology,  75, Sackett, P. R., & Yang, H. (2000). Correction for range restriction:                          119-126.       An   expanded   typology.  Journal   of   Applied   Psychology ,  85,             Spearman, C. (1904). General intelligence, objectively determined       112-118.                                                                                and measured. American Journal of Psychology , 15, 201-292. *Salter, C. A., & Routledge, L. M. (1974). Intelligence and belief                            doi:10.2307/1412107       in   the   supernatural.   Psychological        Reports,   34,      299-302.      Stanovich,      K.   E.,   &   West,    R.   F.   (2000).    Individual    differ-       doi:10.2466/pr0.1974.34.1.299                                                           ences   in   reasoning:   Implications   for   the   rationality   debate? *Saroglou,   V.,   &   Fiasse,   L.   (2003).   Birth   order,   personality,   and          Behavioral       &   Brain    Sciences,    23,    645-655.    doi:10.1017/       religion:   A   study   among   young   adults   from   a   three-sibling               S0140525X00003435       family.  Personality       and   Individual    Differences,    35,    19-29.      *Stanovich, K. E., & West, R. F. (2007). Natural myside bias is       doi:10.1016/S0191-8869(02)00137-X                                                      independent   of   cognitive   ability.  Thinking   &   Reasoning,  13, *Saroglou, V., & Scariot, C. (2002). Humor Styles Questionnaire:                             225-247. doi:10.1080/13546780600780796       Personality and educational correlates in Belgian high school                     Stanovich, K. E., & West, R. F. (2008). On the relative independence       and college students. European Journal of Personality, 16, 43-                          of thinking biases and cognitive ability. Journal of Personality       54. doi:10.1002/per.430                                                                and     Social    Psychology,      94,    672-695.     doi:10.1037/0022- Schermer, J. A., & Vernon, P. A. (2010). The correlation between                              3514.94.4.672       general intelligence (g), a general factor of personality (GFP),                  Stark, R. (2002). Physiology and faith: Addressing the “universal”       and social desirability. Personality and Individual Differences,                        gender   difference   in   religious   commitment.  Journal   for   the       48, 187-189. doi:10.1016/j.paid.2009.10.003                                            Scientific Study of Religion, 41, 495-507. Schmidt, F. L., Oh, I.-S., & Hayes, T. L. (2009). Fixed-versus ran-                     Sternberg,     R.   J.  (1999).   The   theory    of  successful    intelligence.       dom-effects models in meta-analysis: Model properties and an                           Review of General Psychology, 3, 292-316. doi:10.1037/1089-       empirical comparison of differences in results. British Journal                        2680.3.4.292       of Mathematical and Statistical Psychology, 62, 97-128. doi:1                     Sternberg, R. J. (2006). The rainbow project: Enhancing the SAT       0.1348/000711007X255327                                                                through assessment of analytical, practical, and creative skills. Sedikides,     C.   (2010).   Why     does   religiosity  persist?  Personality              Intelligence , 34, 321-350. doi:10.1016/j.intell.2006.01.002       and     Social    Psychology      Review,     14,    3-16.    doi:10.1177/        Stolzenberg, R. M., Blair-Loy, M., & Waite, L. J. (1995). Religious       1088868309352323                                                                       participation   in   early   adulthood:   Age   and   family   life   cycle                                               Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013
----------------------- Page 31-----------------------

30                                                                                                           Personality and Social Psychology Review XX(X)       effects on church membership. American Sociological Review ,                      Volken,   T.   (2003).   IQ   and   the   wealth   of   nations:   A   critique   of       60, 84-103. doi:10.2307/2096347                                                         Richard   Lynn   and   Tatu   Vanhanen’s   recent   book.  European Subotnik, R. F., Karp, D. E., & Morgan, E. M. (1989). High IQ                                 Sociological Review, 19, 411-412. doi:10.1093/esr/19.4.411       children at midlife: An investigation into the generalizability                    *Wahlig, E. L. (2005). Health outcomes in multiple sclerosis: The       of Terman’s genetic studies of genius. Roeper Review, 11, 139-                          role   of   beliefs   and   personality   in   cognitive   function,   depres-       144. doi:10.1080/02783198909553190                                                      sion, fatigue, and quality of life (Unpublished doctoral disserta- *Symington,   T.   A.   (1935). Religious   liberals   and   conservatives .                  tion). State University of New York at Buffalo.      New      York,    NY:     Teachers     College,    Columbia      University        Willits,   F.   K.,   &   Crider,   D.   M.   (1989).   Church   attendance   and      Bureau of Publications.                                                                  traditional   religious   beliefs   in   adolescence   and   young   adult- *Szobot,   C.   M.,   Rohde,   L.   A.,   Bukstein,   O.,   Molina,   B.   S.   G.,           hood: A panel study. Review of Religious Research, 31, 68-81.      Martins, C., Ruaro, P., & Pechansky, F. (2007). Is attention-                            doi:10.2307/3511025       deficit/hyperactivity disorder associated with illicit substance                  Win-Gallup   International.   (2012).        Global   index   of   religion   and      use disorders in male adolescents? A community-based case-                               atheism. Available from       control study. Addiction , 102, 1122-1130. doi:10.1111/j.1360-                    Wood, P., & Englert, P. (2009). Intelligence compensation theory:       0443.2007.01850.x                                                                       A   critical   examination   of   the   negative   relationship   between Terman, L. M. (1925-1959). Genetic studies of genius: Vols. 1-5.                              conscientiousness        and    fluid   and    crystallised    intelligence.       Stanford, CA: Stanford University Press.                                                The   Australian   and   New   Zealand   Journal   of   Organizational Terman, L. M., & Oden, M. H. (1947). Genetic studies of genius:                               Psychology, 92, 19-29.       Vol.   4.  The   gifted   child  grows    up.   Stanford,   CA:    Stanford       Wood,       W.   (1982).    The    retrieval   of  attitude-relevant     informa-      University Press.                                                                        tion    from    memory:      Effects    on    susceptibility    to   persua- Terman, L. M., & Oden, M. H. (1959). Genetic studies of genius:                               sion   and   on   intrinsic   motivation. Journal   of   Personality   and       Vol. 5. The gifted group at mid-life: 35 years’ follow-up of the                        Social     Psychology,     42,     798-810.     doi:10.1037/0022-3514.      superior child. Stanford, CA: Stanford University Press.                                 42.5.798 Thorndike, R. L. (1949). Personnel selection: Test and measure-                         Wood,   W.,   Kallgren,   C.   A.,   &   Preisler,   R.   M.   (1985).   Access   to      ment techniques. Oxford, UK: Wiley.                                                      attitude-relevant      information     in  memory       as  a  determinant Tolor, A., & Reznikoff, M. (1967). Relation between insight, repression-                      of   persuasion:     The   role   of  message     attributes.  Journal     of       sensitization, internal-external control, and death anxiety. Journal                    Experimental Social Psychology, 21, 73-85. doi:10.1016/0022-       of Abnormal Psychology, 72, 426-430. doi:10.1037/h0025084                               1031(85)90007-1 Toplak,     M.   E.,  West,    R.   F.,  &  Stanovich,     K.   E.  (2011).    The       *Yankelovich, Skelly, & White, Inc. (1978). Time soundings poll       Cognitive   Reflection   Test   as   a   predictor   of   performance   on              #8140   [Data      file].  Available    from    http://www.ropercenter.      heuristics-and-biases   tasks. Memory   &   Cognition ,  39,   1275-                  1289. doi:10.3758/s13421-011-0104-1                                                *Yankelovich        Clancy    Shulman,      Inc.  (1991).   Yankelovich/Time Trimble,       D.    E.   (1997).     The     religious     orientation     scale:            Magazine/CNN poll: October omnibus [Data file]. Available      Review       and    meta-analysis      of   social   desirability    effects.            from      Educational   and   Psychological   Measurement,  57,   970-986.                    *Young, R. K., Dustin, D. S., & Holtzman, W. H. (1966). Change in       doi:10.1177/0013164497057006007                                                         attitude toward religion in a southern university. Psychological *Turner, E. B. (1980). General cognitive ability and religious atti-                          Reports, 18, 39-46. doi:10.2466/pr0.1966.18.1.39      tudes in two school systems. The British Journal of Religious                      Ysseldyk, R., Matheson, K., & Anisman, H. (2010). Religiosity as      Education, 2, 136-141. doi:10.1080/0141620800020404                                      identity:   Toward   an   understanding   of   religion   from   a   social Vail,   K.   E.,   III,   Rothschild,   A.   K.,   Weise,   D.   R.,   Solomon,   S.,         identity     perspective.    Personality      and    Social     Psychology      Pyszczynski,       T.,  &    Greenberg,     J.  (2010).    A   terror   man-             Review, 14, 60-71. doi:10.1177/1088868309349693       agement      analysis   of   the   psychyological      functions    of  reli-     Zuckerman,        P.   (2007).    Atheism:     Contemporary        numbers     and       gion.  Personality   and   Social   Psychology   Review,  14,   84-94.                  patterns.    In   M.   Martin    (Ed.),  The     Cambridge      companion       doi:10.1177/1088868309351165                                                            to   atheism.    Cambridge,       UK:    Cambridge      University     Press. *Verhage, F. (1964). Intelligence and religious persuasion.Nederlands                         doi:10.1017/CCOL0521842700.004       Tijdschrift   voor   de   Psychologie   en   Haar   Grensgebieden,  19,           Zuckerman, P. (2008). Society without God. New York: New York      247-254.                                                                                 University Press.                                                    Downloaded from at FUND COOR DE APRFO PESSL NIVE on August 12, 2013