High Quality
Open the downloaded document, and select print from the file menu (PDF reader required).
Conservatism and cognitive ability
Lazar Stankov
National Institute of Education (NIE), 1 Nanyang Walk, Singapore
a r t i c l e i n f o a b s t r a c t
Article history:
Received 17 July 2008Received in revised form 7 December 2008Accepted 8 December 2008Available online 3 February 2009
Conservatism and cognitive ability are negatively correlated. The evidence is based on 1254community college students and 1600 foreign students seeking entry to United States'universities. At the individual level of analysis, conservatism scores correlate negatively withSAT, Vocabulary, and Analogy test scores. At the national level of analysis, conservatism scorescorrelate negatively with measures of education (e.g., gross enrollment at primary, secondary,and tertiary levels) and performance on mathematics and reading assessments from the PISA(Programme for International Student Assessment) project. They also correlate withcomponents of the Failed States Index and several other measures of economic and politicaldevelopment of nations. Conservatism scores have higher correlations with economic andpolitical measures than estimated IQ scores.© 2009 Elsevier Inc. All rights reserved.
Keywords:
ConservatismIntelligenceMulti-level
1. Introduction
There has been an increased interest in the construct of conservatism. Recent evidence indicates that some existingstereotypes are not supported by the available data. Forexample,Brooks (2006, 2008)reports that conservativesengage more than liberals in charitable activities and peopleonthepoliticalrightarenearlytwiceashappyasthoseontheleft. The work of Napier and Jost (2008)shows that con-servatives tend to be happier than liberals because of theirtendencytojustifythecurrentstateofaffairsandbecausetheyare less bothered by inequalities in the society. The focus of these investigators is onpolitical conservatism
—
tendency toattach high importance to topics that arehigh on the agendasof right-wing political parties within a given society and,consequently, endorse these parties' candidates in elections.For example, a version of the USA Wilson
–
Patterson Con-servatism Scale (WPC; seeWilson, 1973) used in a studyreported byBouchard et al. (2003)contained 28 items thatasked participants to state how important topics such asabortion, property tax, gay rights, liberals and immigrationare.
1
In the studies reported in the main body of this paper,political conservatism was not examined directly. However, Jost,Glaser,Kruglanski,andSulloway(2003)suggest thatit istime to re-examine the links between political conservatismandahostofindividualdifferencevariables.Aconstellationof these individual difference variables may be called
Conserva-tive syndrome
. Although an alternative label, psychologicalconservatism, may be more appropriate if one's aim is tocontrast politics and psychology, the term syndrome appearsto be adequate for a discourse within the
fi
fi
rms that severalpsychologicalvariablespredict politicalconservatism.The listincludes death anxiety; system instability; dogmatism;intolerance of ambiguity, low openness to experience, anduncertainty; need for order, closure, and negative integrativecomplexity; and fear of threat and loss of self-esteem. Thetheory of Jost et al. (2003)treats political conservatism as
Intelligence 37 (2009) 294
–
304
E-mail addresses:
1
The remaining 23 topics from the list are: death penalty, astrology,x-rated movies, modern art, women's liberation, foreign aid, federalhousing, democrats, military drill, the military draft, capitalism, segrega-tion, moral majority, paci
fi
sm, censorship, nuclear power, living together,republicans, divorce, school prayer, unions, socialism, and busing (Bou-chard et al., 2003).0160-2896/$
–
Contents lists available atScienceDirect
Intelligence
motivated cognition and builds on a large body of researchaccumulated sincethe end of World War II. One antecedent isthe approach advocated byWilson's (1973)dynamic theorythat also saw conservatism as a motivated response to un-certainty. The threat or uncertainty may derive from fear of death, anarchy, foreigners, dissent, complexity, novelty,ambiguity, and social change. Responses to these sources of uncertainty include superstition, religious dogmatism, eth-nocentrism, militarism, authoritarianism, punitiveness, con-ventionality, and rigid morality. Wilson postulated thatpolitical conservatism derives from genetic sources (anxietyproneness, stimulus aversion, low intelligence, and physicalunattractiveness) as well as environmental in
fl
uences (par-ental coldness, punitiveness, rigidity, inconsistency, and lowsocial class).Jost et al. (2003)summarize their own positionin the following way:
“
The core ideology of conservatismstresses resistance to change and justi
fi
cation of inequalityand is motivated by needs that vary situationally and dis-positionally to manage uncertainty and threat.
”
(p. 339).In this paper, I examine the hypothesis that low cognitiveability may be related to conservative syndrome (or con-servatism, for short) which, in turn, is de
fi
ned in terms of measures of personality, social attitudes, values, and socialnorms. There are two ways to arrive at this assumption. First,we can assume that cognitive ability affects conservatismdirectly. Thus, the perceived threat may vary depending oncognitivelevel
—
sourcesofthreatsuchascomplexity,novelty,and ambiguity may be more threatening to those who scorelow as opposed to those who score high on cognitive tests.Second, we can postulate that there exists an independentprocess that in
fl
uences both conservatism and cognitivefunctioning. A candidate for this role may be mental rigidity.My primary aim in this paper is to present evidence of correlation, not to test these two causal models.A recent paper byDeary, Batty, and Gale (2008)providesdevelopmental evidence for a link between intelligenceassessed at the age of 10 and anti-traditional and liberalsocial attitudes (i.e., the opposite of conservatism) at age 30.They report the results of a structural equation modellinganalysis that shows a signi
fi
cant direct path coef
fi
cient of .46betweenageneral cognitivefactor
g
anda latentattitudetraitthey label as Liberal Non-traditional Social Attitudes.
1.1. Conservatism across the domains of Personality, Social Attitudes, Values and Social Norms
Our approach differs from previous work in the way wede
fi
ne and measure the construct of conservatism. This con-struct emerged, somewhat unexpectedly, in three studies (seeMethodsectionforfurtherdetail).The
fi
rststudywasdesignedto assess cross-cultural differences on a set of measures fromthe domains of Personality, Social Attitudes, Values and SocialNorms. Measures from these domains have been used inprevious studies of others and cross-cultural differences havebeen reported but no single study covered all four domains.Most of the information to be reported here derives fromthis
fi
rst study (seeStankov & Lee, 2008). The second and thethird study (seeStankov, 2007) were based on the US samplesonly. Structural (i.e., factor-analytic) results of these latterstudies proved to be in agreement with the results of the
fi
rststudy.Inourwork,conservatismiscapturedbyascore
—
usuallyafactor score
—
obtained from several scales that were notdeveloped speci
fi
cally for the measurement of conservatism.Thus, it incorporates measures of Personality (Big Five fromIPIP), Social Attitudes (Saucier, 2000; Stankov & Kne
,2005), Values (Schwartz & Bardi, 2001), and Social Norms
(GLOBE;House, Hanges, Javidan, Dorfman, & Gupta, 2004)
—
atotalof43differentsubscalescores.Nevertheless,ouranalysesshowthepresenceof a factorof
Conservatism
that has loadingsfrom subscales from all these domains and captures manyconstructsthatareincludedinthenomologicalnetof Jostetal.(2003)andWilson (1973). This factor is expected to correlate
with cognitive ability for reasons outlined above.What are the other factors that emerge from the analysisof 43 subscales? Are they also expected to correlate withcognitiveability?Stankov(2007)foundthreedomain-relatedfactors. They are quite different from the Conservatism factorin that they show very little overlap between the domains.These are:
∘
Personality/Social Attitudes
. Thisisusuallyabipolarfactorcontrasting Personality traits on the negative side andSocial Attitudes on the positive side. Loadings of Person-ality traits on this factoraretypicallylower than loadingsfrom the Social Attitudes measures. In some of ouranalyses, this factor splits into a separate Personalityfactor representing
“
good
”
evaluative processes (orperhaps social desirability) and a Social Attitudes factorrepresenting anti- or amoral attitudes towards socialobjects (Stankov & Kne
, 2005).
∘
Values
. See Method section for the interpretation of thisfactor.
∘
Social Norms.
Several Social Norms scales from GLOBEstudy (House et al., 2004) load on this factor.In this paper I report the analyses based on a smaller (22)number of variables that correspond quite closely to thesolution obtained with the full set of 43 measures. Smallernumber of variables is employed in order to carry out simul-taneous (i.e., multilevel) structural equation modelling of individual- and country-level data that has not been reportedin the past.Thereis noempiricalevidenceortheoreticalargumentsinthe literature that suggest a relationship between cognitiveability and Values or Social Norms.
2
Thus, it is reasonable toassume that these two constructs do not correlate with cog-nitive measures. The situation is different with the Person-ality/Social Attitudes dimension.Jost (2006)reports thatConscientiousness (positively) and Openness to Experience(negatively) correlate with Democrat/Republican voting pre-ferencesofthestateswithintheU.S.,interpretedasre
fl
ectionsof liberal/conservative tendencies. Openness to Experience isalso known to correlate about .30 with measures of intelli-gence (Stankov, 2005; Stankovand Lee, 2008). The other side
2
An unknown reviewer pointed to the fact that Values and Social Normsmay be related to
“
moral
”
behavior and that neo-Piagetian theories argue forthe link between such behavior and cognitive ability. This link is tenuous
—
measures of both Values and Social Norms are relatively new and theirrelationship to moral behavior is unknown at present.295
L. Stankov / Intelligence 37 (2009) 294
–
304
of this bipolar factor, Social Attitudes, captured by Toughness,Maliciousness, and Betaism (i.e., non-PC motives for beha-vior), have qualities reminiscent of Dogmatism and Author-itarian personalities that are often seen as components of conservatism (seeJost et al., 2003). Since in our work theyde
fi
ne a factor that is separate from conservatism, it isreasonable to assume that there is a separation betweenthuggish and rough Social Attitudes trait and Conservativesyndrome that captures not only social attitudes but alsoValues,SocialNorms,andPersonalitytraits.Theseroughsocialattitudesarealsolikelytoberelatedtocognitiveability
—
theyoftenre
fl
ectdif
fi
cultiesordisinclinationtomake
fi
1.1.1. Individual-level and country-level conservatism
Our work that led to the
fi
nding of the above four factorswas motivated in part by interest in cross-cultural compar-isons. In one of our studies, the participants came from boththe U.S. and foreign countries. Within the tradition of cross-culturalpsychology,totalvariancesonmeasuresofinterestaresplitintotwocomponents
—
withinlevel(orindividuallevel)and between level (or country level). The country levelvariance
–
covariance matrix can be arrived at by calculatingan aggregate measure such as arithmetic mean for allparticipants from a given country (seeHofstede, 2001).Thus, each of the 35 countries in our cross-cultural studywill have a score oneach of the 22measures employed inthisstudy, and a data reduction procedure like factor analysis canbe applied to this 35 by 22 matrix. One issue of interest iswhetherthestructuresatindividualandcountrylevelsarethesameordifferent.Iftheyarethesame,itcanbeconcludedthatthe same in
fl
uences operate at both levels. If different, theassumptionhastobethatin
fl
uencesarenotthesameandtheargument may be that the country level, not individual level,structure re
fl
ects true cultural differences.Since our interest is in the relationship between con-servatism and cognitive ability, the between-countries scoresprovide an opportunity to examine the same question fromthe cross-cultural perspective. Thus, if countries differ interms of conservatism, how are these differences related tomeasures of countries' cognitive performance and educa-tional achievements? What may be the cause(s) of country-level differences in conservatism?Whiletheindividuallevelofanalysisfocusesonimportantpsychological issues, the country-level analysis brings intofocus important social policy issues. Together, they point to alink between psychological and political processes that hasbeen neglected since the 1970s ( Jost, 2006). The evidence forthe existence of such a link at the country level is importantsince it may guide decisions related to the deployment of resources.Apart from showing the link between cognitive perfor-manceandconservatism,country-levelanalysesallowfortheexamination of broader issues. For example, they provide foran opportunity to examine the relationship of these two con-structs with other country-level measures, including variouseconomic and social indicators.The link between IQ, economic measures of wealth and ahost of other variables has been explored extensively. Forexample, according toKanazawa (2006; p. 593)the meanPearson's product-moment correlation between national IQ and various measures of Gross Domestic Product (GDP) acrossnumerousyearsamongthe185nationsthatisreportedbyLynnand Vanhanen (2002, pp.110
116)is .577.Lynnand Vanhanen(2006)interpret these correlations as showing that IQ is animportant factor contributing to the differences in nationalwealth and rates of economic growth.Rindermann (2007;Table 4, p. 686) reported correlations between the sum of several cognitive ability measures and GDP in year 1998 to be.63. In his subsequent papers (Rindermann, 2008a,b) he alsoreports correlations between IQ estimates and country-levelmeasures of education, democracy, the rule of law and manyother economic and social indicators.Similar country-level correlations between Conservatismand economic and social indicators do not exist in the litera-ture. If they turn out to be of the same order of magnitude,shall we assume that Conservatism is another
“
importantdeterminant
”
like IQ? Our data will allow us to address thisissue.I employ structural equation modelling and multilevelprocedures (Muthén,1994) and regression analysis to exam-ine the nature of conservatism at both the individual andcountry levels.
1.1.2. Aims
My aims in this paper are threefold. First, I present struc-tural evidence for the existence of stable factors at both theindividual- and country-levels of analysis. Although the over-all structure at the individual and country levels may differ, aconservatism factor is expected to emerge at both levels.Second,correlationsbetweenfactorscoresfrombothlevelsof analysis with individual- and country-level cognitive mea-sures are presented. Individual cognitive measures are scoreson typical aptitude tests. Country-level proxies for cognitivemeasures are both statistics regarding educational enrolmentand scores from the objective achievement tests. Myexpecta-tion is that the strength of the individual- and country-levelconservatism will be negatively correlated with cognitiveability scores. Third, I report on the relationship betweenConservatismandahostofcountry-leveleconomicandsocio-logicalvariables.Theaimistocomparepredictivevaliditiesof IQ and Conservatism scores.
2. Method
2.1. Participants
The
fi
ndingstobe reported inthis paperderive fromthreestudies, all of which employed the same set of measures fromthe four domains of Personality, Social Attitudes, Values, andSocial Norms. The
fi
rst study (
N
=1600) was a cross-culturalstudy with participants from 73 countries. These were thepeople who took the Test of English as a Foreign Language(TOEFL iBT) as a prerequisite for enrolment, mostly ingraduate schools, at U.S. universities. The samples of partici-pantsfromdifferentcountriesarenotrepresentativeandmaydiffer from the parent population in many ways. After takingTOEFL,theywereaskedtoparticipateinaseparatesurveyfora$20 payment. The second (
N
=430) and third (
N
=824)studies employed students from 22 community colleges fromacross the U.S. The data presented inTable 1are based on the
fi
rstandsecondstudies.Iaddedthesecondsampletothe
fi
rst
296
L. Stankov / Intelligence 37 (2009) 294
–
304
Add a Comment
James Gatcombleft a comment