Not by Twins Alone: Using the Extended Family Design to Investigate Genetic Influence on Political Beliefs Author(s): Peter

K. Hatemi, John R. Hibbing, Sarah E. Medland, Matthew C. Keller, John R. Alford, Kevin B. Smith, Nicholas G. Martin, Lindon J. Eaves Reviewed work(s): Source: American Journal of Political Science, Vol. 54, No. 3 (July 2010), pp. 798-814 Published by: Midwest Political Science Association Stable URL: http://www.jstor.org/stable/27821953 . Accessed: 13/03/2012 09:37
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Midwest Political Science Association is collaborating with JSTOR to digitize, preserve and extend access to American Journal of Political Science.

http://www.jstor.org

Not by Twins Alone: Using the Extended Family on Political to Investigate Genetic Influence Design

of Peter K. Hatemi University Iowa of John R. Hibbing University Nebraska-Lincoln of E. Medland Queensland InstituteMedical Research Sarah of Matthew C. Keller University Colorado John R. Alford Rice University of Kevin B. Smith University Nebraska-Lincoln of Nicholas G. Martin Queensland InstituteMedical Research Commonwealth Lindon J. Eaves Virginia University
but the Variance components estimates ofpolitical and social attitudes suggesta substantial level ofgenetic influence, we includeresponses results have been challengedbecause theyrelyon data from twinsonly.In thisanalysis, fromparents and nontwinfull siblings of twins,account for measurement errorby using a panel design, and estimategenetic and we variance by environmental maximum-likelihood structural equationmodeling.By doing so, address thecentralconcerns or no design offers verificationof either the equal environments randommating of critics, including that the twin-only genetic for design leads to theconclusion that mostpolitical and social attitudes, assumptions. Moving beyond thetwin-only than reportedby studies usingmore limited account for an even greaterproportion of individual differences influences data and more elementaryestimation techniques.These findings make it increasingly difficultto deny that?however formation of political and social attitudes. indirectly?genetics plays a role in the

derives from the existence of two fundamentally different types of twin pairs, each with a known level of genetic (MZ; frequently but erroneously similarity.Monozygotic called identical) twins, for several hours and often days, are the same zygote. The subsequent splitting into two zygotes means
. Hatemi

sical twin design (CTD) to estimate the genetic and environmental components of physical, psychologi cal, behavioral, and clinical traits.The value of this design

Genetic epidemiology, psychiatric genetics, and behavior genetics have long relied on the clas

essentially the same. Dizygotic (DZ; frequently called fra ternal) twins result from separate fertilization of one egg a by one sperm and a second distinct egg by second dis tinct sperm. In accord with normal meiotic cell division and subsequent fertilization, the resultant twin pair shares same as roughly 50% of the variable genetic makeup, the all nontwin

full sibling pairs. It is this fixed ratio (two to one) of shared genetic similarity between MZ and DZ twins that provides most of the leverage of twin studies. The use of this powerful design allows geneticists to estimate the roles of genes, common environmental
Iowa City, Iowa 52242 (phatemi@gmail.com). of Nebraska-Lincoln, Lincoln, NE

that the genetic heritage of each twin is

is Assistant Professor of Political

Peter

R. Hibbing Research,

of Queensland Institute Medical Department ofGenetic Epidemiology, (jhibbing@unl.edu). Sarah E.Medland isa ResearchOfficer in the
300 Herston and Fellow Rd., Brisbane, Queensland, of the Institute of Behavior Professor Science, Genetics, is Associate of Political of Genetic of Political C. Keller is Assistant Professor Matthew (Sarah.Medland@qimr.edu.au). of Colorado, Boulder, CO 80309 (matthew.c.keller@gmail.com). University Kevin TX 77251 6100 Main, Houston, Science, Rice University, (jra@rice.edu). G. Martin Nicholas of Nebraska-Lincoln, Lincoln, NE 68588 (ksmithl@unl.edu). Australia

is Foundation

Regents

Professor

of Political

of Iowa, Science, University Science and Psychology,

University

John 68588 of

Psychology John R. Alford Smith Laboratory

B. is

is Professor Head

(Nick.Martin@qimr.edu.au). and Behavioral Genetics, American ?2010, Journal

University 300 Herston Rd., Brisbane, Institute of Medical Research, Queensland Epidemiology, and Psychiatry, Virginia Professor of Human Genetics Lindon J.Eaves isDistinguished VA 23284 (ljeaves@vcu.edu). University, Richmond, Virginia Commonwealth Science, Vol. 54, No. 3, July 2010, Pp. 798-814 0092-5853

Australia Queensland, Institute for Psychiatric

of Political

Midwest

Political

Science

Association

ISSN

798

AND POLITICALBELIEFS 799 GENETICS influences (those likely to be shared by familymembers), and unique environmental influences (those not neces in traits as varied as sarily shared by family members) jority of the genetic influence on vote choice appears to be accounted for not directly but indirectly through the heritability of political attitudes. Fowler, Baker, and Dawes the important discovery that ac (2008) made fluenced tual voter turnout (not self-reported turnout) also is in substantially by genes. Further studies found

breast cancer, depression, schizophrenia, attention deficit hyperactivity disorder (ADHD), alcoholism, autism, obe sity, and personality. It has also been used to estimate sources of individual differences for political beliefs.

Genetic Influences on Political Beliefs

The most extensive early studies on attitudes were Martin et al. (1986) and Eaves, Eysenck, and Martin (1989). Us ing large data sets drawn from twins inAustralia and the

only a modest effect on direction of party identification, mostly an indirect effect on vote choice, and a stronger effect on strength of group affiliation could make sense

that strength of affiliation with a party (regardless of the particular party involved) is strongly heritable (Hatemi, Alford et al. 2009). The findings that genes appear to have

given that party identification and voting for specific can didates are time-bound phenomena whereas the tendency toward group attachment (regardless of the nature of the group) may run deeper. perceive political attitudes to be entirely Many "learned" and therefore just as time and culture bound as

United States, respectively, these scholars reported results that "undermine the na?ve assumption that the resem

blance of family members can be interpreted in purely social terms" (Martin et al. 1986, 4368). Even though later analyses, often with different data and in different (see Bouchard 1999; Olson et al. 2001), produced cal scientists took virtually no notice of these provocative as findings, perhaps because political scientists typically sume attitudes are entirely the product of environmental countries and McGue 2003; Eaves et al. similar results, politi

for attitudes toward nuclear power, property taxes, and busing, and these issues have only been present for a gen eration or two, far too short a time for specific genes

party identification (see Eagly and Chaiken 1993; Perloff 2003). After all, twin studies have reported heritability

forces such as parental socialization and do not take se riously the possibility that genes could be involved (for 1971; Peterson 1983; Segal and exceptions, seeMerelman Spaeth 1993, 234; Zaller 1992, 23). Then in 2005, Alford, Funk, and Hibbing performed additional analyses on the same combined data set collected and employed by Eaves et al. (1999) and presented the results to the political sci ence community. Similar to earlier results, the findings

may be that pertinent to such concerns to evolve. But, it while not relating directly to ephemeral issues of genes, the day, work on deeper principles of group life that in turn are relevant to specific issues depending upon how they are framed in a particular culture at a particular time.

There is likely no direct genetic basis forwhether or not to build a wall on theMexican border, but theremight be genes that indirectly shape perception of outgroups,

sensitivity to external threat, and preference for ingroup cohesion (for evidence that this maybe the case, see Oxley et al. 2008). Due and data in part to potentially demanding assumptions limitations, the classic twin design (hereafter

are shaped by a combination of environmental and ge netic influences rather than just the environment may be difficult for some political scientists to accept in that incorporating genetic influences necessitates a fairlydra matic rethinking of the nature of political attitudes. One critic of Alford, Funk, and Hibbing asserted that "if true, itwould require nothing less than a revision of our un derstanding of all of human history,much, ifnot most of

a suggested surprising degree of genetic influence for po litical attitudes but suggested that genetics may play less of a role in the direction of party identification. The claim that differences in ones political beliefs

is primarily utilized as only an initial indication of CTD) themethods by which the trait of interest has been ac quired. In particular, three shortcomings are of relevance here, and the first two are specific to the CTD; namely, univariate twin-only analyses provide little opportunity

political science, sociology, anthropology, and psychol ogy, as well as, perhaps, our understanding of what it means to be human" (Charney 2008, 300). Regardless, since 2005 interest in the heritability of political variables has increased. Hatemi et al. (2007) found that vote choice is heritable but that the ma

error creates an upward bias on estimates of the impact of unshared environment (Fisher inmodels that correct formate assortation, 1918)?and as we do here, this concern is compounded (Eaves and Hatemi 2008). In this article, we address these shortcom because measurement ings by employing a nuclear family design, which includes

shortcoming is the problem of estimating and correcting formeasurement error, a perennial concern for all em pirical investigations but a particular concern in theCTD

to detect violations of either the equal environments as sumption or the random mating assumptions. The third

800

PETER

. HATEMI

ET AL.

mea parents and nontwin siblings, as well as test-retest sures of each of the traitsof interest.By applying improved we pro procedures to a valuable data set, methodological vide amore accurate estimate of the influence on political attitudes of genetics, shared environment, and unshared
environment.

studies do not assume that the environments ofMZ are no more MZ similar than the environments of DZ

twins twins.

twins are indeed more likely than DZ twins to share certain environmental experiences, such as sleeping in the same bedroom and having the same friends (Kendler et al. 1992; Scarr and Carter-Saltzman 1979). The key question iswhether violations of the equal environments assump tion (EEA) occur with regard to political attitudes. Shar

The data we utilize were also the basis for the Eaves et al. (1999), Alford, Funk, and Hibbing (2005), and and Eaves (2009) twin-only studies, Hatemi, Medland, a data set originally known as the "Virginia 30,000" or "VA30K" for short (for information on the structure of the sample and ascertainment procedures, see Lake et al. 2000; Maes, Neale, and Eaves 1997; Truett et al. 1994). The approximately 30,000 adult subjects (aged 18-84 years) were twins (N = 14,781), spouses (N = 4,391), parents (N = = = 4,800), and 2,360), relatives (N 195), offspring (N nontwin siblings of twins (N = 3,184). The inclusion of nontwin relatives is especially helpful in identifying the multiple sources of biological and cultural inheritance et al. 1985). (Heath The social and political attitude measures were in cluded in a 28-item contemporary attitude battery gath ered as part of a larger "Health and Life Styles" inven tory conducted

a shared ing the same bedroom is one thing; expecting to lead to greater similarity bedroom during childhood in political attitudes and behaviors is another. It certainly seems unlikely that the parents ofDZ twins are less eager than the parents ofMZ twins for their children to hold

traits (for a review, seeMedland and Hatemi 2009). These methods include comparing the twin trait similarity for zygosity and for family-perceived zy those twins forwhom genetic zygosity is gosity among misperceived by familymembers (blood-determined zy blood-determined

ify thatMZ and DZ pairs are not unequally influenced a by different environments for wide array of behavioral

lations of the EEA artificially inflateheritability estimates. Over the last 30 years, a variety of methods in psy ver chology, psychiatry, and genetics have been used to

the same political beliefs. Nevertheless, theCTD by itself is incapable of empirically alleviating suspicions that vio

in 1986. Item formatwas the same as the

Wilson-Patterson

Attitude Index (Wilson and Patterson is simplified by pre 1968), where attitude measurement each item in a one- or two-word format. Respon senting dents are instructed to answer with the first reaction that
comes to mind: "agree," "uncertain," or "disagree." Data

gosity is consistently found to be the better predictor?see 1976; Plomin, Willerman, Matheny, Wilson, and Dolan and Loehlin

were collected by mail, with mail and telephone follow up of nonrespondents when needed. Approximately two years later, the same attitude items were included in a follow-up questionnaire mailed to twins aged 50+ years,

1976; Scarr and Carter-Saltzman 1979); ob serving twin treatment by family members and others to examine differences in behaviors toward the different

mental

providing measures of attitude stability for 1,019men and 2,912 women. In the remainder of this article, we ply the two-wave extended family portions of these data in order to better explore the nature and transmission of political
attitudes.

in environment for the trait of interestwhile controlling for actual zygosity (Kendler et al. 1987; Heath, Jardine, and Martin 1989); extending theCTD by partitioning the shared environment into the overall common environ ment, Qesiduab which is completely correlated for all twin pairs, and thatwhich is influenced by the perceived zygos

twin types (Lytton 1977); measuring specific environ indicators for each twin and modeling differences

The Equal Environments Assumption

An important drawback with data restricted to twin-only analyses is that questions inevitably arise over the as

MZ and forDZ sumed similarity in the environments for twin pairs. In fact, no feature of the CTD has generated more attention and concern. If the environments ofMZ twin pairs aremore similar than the environments of DZ

method, Visscher et al. (2006) obtained exactmeasures of genetic sharing of sibling pairs, and excluded MZ twins, thus removing any equal environmental concerns, and found that the heritability estimate for height was very similar to that derived from traditional CTD analyses.

and Kendler 1995; Kendler ity, Cspecifio (Hettema, Neale, et al. 1993; Xian et al. 2000); and utilizing actual genetic similarity, known as identity by descent (IBD), rather than assuming thatDZ twins or full siblings share on av erage 50% of their segregating genes. Regarding this last

to genetics when it actually belongs to environmental in is that contemporary twin fluence. Often misunderstood

twin pairs, and if this increased similarity is in any way related to the trait of interest,variance may be attributed

to assess political attitudes every two years, they found

Perhaps ofmost relevance to questions about theEEA is recent work by Hatemi, Funk et al. (2009). Utilizing a longitudinal panel study of adolescent twins (aged 8-18)

AND POLITICALBELIEFS 8oi GENETICS that therewas no difference in MZ/DZ twin pair similar adolescence but that twin pair differences ity throughout in political attitudes emerged later,when twins had de twin-specific environmental effect, themean influence of the two sources of the twin-only environmental variance (MZ and DZ) is zero.What are the possible combinations

necessary to argue that a special MZ twin environment for political attitudes exists in adolescence but remains

parted from the parental nest. Thus, in order for it to be believed that a violation of the EEA is responsible for the heritability estimates previously reported, it would be

and DZ effects thatwould produce a zero overall twin effect? In a sample with an equal or greater number ofMZ twin pairs relative to that of DZ twin pairs, the DZ twin pairs would have to have a zero environmen ofMZ tal correlation and theMZ twin pairs an environmental correlation of 1 in order for an MZ-only environment effect to exist despite the presence of a zero-combined twin impact. Since we know that DZ twin pairs share the same parents, schools, SES, and home environment, cotwin common environment

dormant until adulthood, when it is triggered by some that then shapes adult prefer unidentified mechanism
ences.

In lightof these findings, a substantial amount of evi dence runs against the existence of a special twin environ ment for political beliefs. Still, since doubts continue, we adopt an alternative strategyhere. Directly testing forpo tential differences between MZ and DZ pair environments

a scenario of a zero DZ

here is to include data on nontwin siblings, thereby allow ing themodel to partition variance separately for siblings generally and for twin siblings specifically. If the more similar treatment of MZ twins were indeed influencing relevant (i.e., political) traitvalues, then themore similar treatment of DZ

common environmental measures not typically available. MZ twins reared apart Such direct tests include analysis of and adoption studies, but these approaches have theirown problems (seeMedland and Hatemi 2009). Our approach

and for themethod by which these differences might in fluence the trait for each zygosity type requires specific

the contention of a statistically significant influence from differences in the common environment between DZ and MZ twins. Another significant advantage ofmodeling nontwin data is that it vastly increases the power to de sibling tect common environmental effects (Coventry and Keller

is not possible. Thus, we can reasonably conclude that if the twin-only environment is not significantly differ ent from zero, there is insufficient evidence to support

2005; Posthuma and Boomsma 2000; Posthuma et al. 2003). The ability to identify common environmental in fluences or genetic dominance ismaximized when there are four times asmany DZ pairs asMZ pairs and nontwin siblings effectively increase theDZ/sib toMZ ratio (for a

of twin-specific environmental effects. In simple terms, while we cannot identify specific EEA violations, we can identify the total amount of variance attributable to twin (ML) approach, the statistical significance of the twin-only environment can be estimated in twoways. The first is to examine the con fidence intervals for the twin-specific environment since ifthe bounds extend to zero, the effectof the twin-specific environment is not likely to reach statistical significance. specific environmental effects. maximum-likelihood Using a full

correcting for fixed effects (e.g., age), differences between twins and nontwin siblings provide an indirect estimate

twins relative to nontwin full siblings should also affect that trait. The degree of genetic simi larity of DZ twins and full siblings is the same, so after

more detailed conceptual and mathematical discussion, see Nance and Neale 1989 and Posthuma and Boomsma

variances provides additional information and offers re searchersmore confidence in the EEA for the trait inques tion. If themore similar treatment of MZ twins affects twins their traitvalues, themore similar treatment of DZ

2000). In sum, checking for differences between the DZ covariance and the twin-sibling and sibling-sibling co

as compared to regular siblings is likely to impact their values. The data set being employed in this research con tains information on 9,727 nontwin sibling pairs, making it extremely valuable for these purposes. The addition of parents and nontwin siblings to the analysis also allows for a more extended exploration of the extent to which some part of the genetic variation

A more robust measure is to drop the twin-only environ mental effects and compare the fit of this reduced model model. If the contribution of the twin with that of the full

only environment does not reach statistical significance, it met for the trait being studied. Given suggests the EEA is that this is an indirect test of the equal environments as sumption, it is important to consider themathematical

with more

Typically, twinmodels focus on the additive estimate be cause the combined effect of all genes can be estimated confidence than models nonadditive

is nonadditive. Nonadditive genetic influences arise from interactions either within a gene (known as dominance) or between genes (known as epistasis; Neale et al. 2003).

potential for there to be some MZ-specific environmental impact despite the overall absence of twin-specific effects. If the combination of both types of twins produces no

which partition out influences. This limitation is important when diagnostics suggest nonadditive effects are present. Diag nostics for detection of nonadditive influences are most

8 2

PETER

. HATEMI

ET AL

Table 1 Wilson-Patterson

InventoryPolychoric Correlations forRelatives

Dizygotic Twins DZM .41 .23 .35 .27 .20 .28 .25
.29

Nontwin Siblings Item Death Penalty Astrology X-rated movies Modern Women's art liberation MM .28 .25 .34 .22 .25 .20 .18
.17

Monozygotic MZM .52 .48 .58 .43 .31 .41 .28

.43

Twins M-D .35 .17 .22 .22 .29 .20 .21

.38

Parent-Offspring M-S F-D

F-S

FF .31 .25 .31 .26 .32 .23 .15

.27

MF .27 .16 .26 .20 .23 .20 .16

.24

DZF .37 .31 .40 .32 .35 .21 .25

.37

DZMF .33 .23 .31 .23 .16 .23 .16

.27

MZF .54 .47 .59 .43 .52 .46 .40

.48

.19 .18 .26 .23 .14 .22

.26

.14 .15 .14 .09 .25 .15.21

.30 .29

.15 .25 .20 .19 .25

Foreign aid Federal housing

Democrats

.21 .18 .13 .14 .15 .11 .34 .24 .33 .21 .21 .17 .19 .43 .12.18 .32 .30 .25 .16 .26 .27 .18 .37

Military drill The Abortion Property tax Gay rights Liberals Immigration Capitalism Segregation Moral Majority Censorship
Nuclear power

.18 draft .28 .42 .23 .31 .24 .26 .34 .21 .20 .29
.26

.17 .21 .46 .26 .45 .31 .24 .24 .20 .24 .17
.12

.17 .21 .41 .25 .35 .26 .23 .24 .23 .17 .15 .15
.14

.19 .29 .42 .22 .37 .32 .30 .36 .21 .19 .14 .15
.20

.24 .17 .56 .29 .49 .37 .28 .31 .26 .24 .14 .27
.26

.10 .25 .43 .27 .39 .24 .20 .23 .14 .17 .18 .15
.18

.40 .49 .55 .50 .57 .36 .45 .62 .38 .41 .36 .43
.45

.36 .36 .68 .45 .60 .47 .46 .47 .38 .43 .31 .37
.33

.20 .16 .48 .21 .48 .29 .22 .28 .19 .23 .08 .17
.22

.21 .20 .20 .19 .11 .21 .23 .23 .18 .07* .09
.18 .20

Pacifism .10 .14

.05* .12
.13

.04* .03f

Living together Republicans School prayer Socialism

.39 .26 .50 .13

.52 .22 .44 .16

.44 .21 .24 .42 .12 .19 .18

.51 .24 .32 .45 .25 .22 .36

.50 .33 .35 .46 .25 .26 .27

.36 .28 .19 .44 .14 .26 .26

.56 .47 .42 .67 .46 .38 .45

.70 .48 .49 .66 .42 .45 .42 2.39

t=

.45 .33 .25 .47 .26 .16 .21 .40

.29 .32 .36 .31 .25 .23 .12 .14 .29 .22 .20 .15 .19 .27 .24 .25 .05* .14

Divorce .27 .21 .20 Unions .18 .20 Busing .23 7.67
(p =

.52 .45 .48

(pairs)
Notes: Key: All correlations

4462
S2 17.73
are

1564

3701
13.75

610
20.13
unless

1273
15.50
otherwise

1397
28.48
marked; *=

814
7.56
.05,

1982 4802
not significant.

3233
.34

3166
.68

2315
1.35

significant

.01 or better),

MM = male siblings;FF = female siblings; = male and femalesiblings; MF DZM = male dizygotictwins; DZF = femaledizygotictwins; DZMF = mixed sexdizygotictwins; MZM = male monozygotic twins; MZF = female M-D = mother-daughter; M-S monozygotic twins;
= mother-son; F-D = father-daughter; and F-S = father-son.

fluences are likely to be important (Neale et al. 2003). The correlations presented inTable 1 show that for only three of the 28 items (for females) and seven of the 28 items total of only 10 out of a possible 56?is (formales)?a there any suggestion of nonadditive effects. For these 10, theDZ correlations are only slightlybelow half of the MZ

often performed by comparing the MZ and DZ twin cor relations. If the MZ pair correlation is significantlymore than twice as large as theDZ correlations, nonadditive in

A more developed assessment of the presence of non additive effects ismade possible by the inclusion in our data set of nontwin family members. Additive genetic twins, nontwin siblings, and parent pairs, and that, for all three of these relationship offspring categories, average at least half the size of theMZ twin effects typically produce trait correlations that are at sim ilar levels for DZ

correlations; thus preliminary analyses give little cause to suspect significant nonadditive effects.

pair correlations. When nonadditive genetic effectsdom inate, theMZ twin correlations will remain robust, but all three of the other family pairs will exhibit much re

duced similarity. This distinction gives rise to the readily

AND POLITICALBELIEFS 803 GENETICS apparent family history of traits that exhibit "narrow sense" heritability (i.e., heritability that "runs in the fam case, the additive genetic path that sets DZ = .5/MZ = 1 no longer would be accurate. Across a study population, the higher the proportion of spouses who share genes for

ilies despite the fact that theymay have equally strong, if more complex, genetic underpinnings). As it turns out, extended twin-family studies of per sonality provide clear evidence of nonadditive genetic effects (Keller et al. 2005), but our Table 1 provides no

ily" and that characterizes simple additive genetic effects) in contrast to traits that exhibit only "broad sense" heri tability (i.e., traits that show little clear clustering in fam

a trait, the closer the DZ correlation will be to theMZ correlation and themore the genetic variance of this trait will be underestimated (regardless of the genetic simi larity of parents, MZ twins share 100% of the variable genetic code). Thus, ifpeople tend to choose mates with similar positions on political issues, the CTD understates the heritability of political attitudes and inflates estimates

evidence of this pattern for political temperaments. At items examined here, trait Wilson-Patterson least for the correlations are very similar across same-sex DZ twins, nontwin siblings, and parent-offspring pairs, and for all three of these relationship categories, average correlations

are approximately half the size of the MZ twin pair corre as measured here appears lations. Political temperament to exhibit narrow sense heritability, in clear contrast to the broad sense heritability that characterizes personality
traits.

Assortative Mating
A second major assumption of the CTD is thatmating is random regarding the traitof interest. If it isnot, variance

of 5,160 spousal pairs, making it ideal for inspecting inter spousal correlations. Table 2 consists primarily of the 28 Wilson-Patterson Inventory items and also, at the top, an overall additive index of "liberal/conservative" responses to these 28 items, but for purposes of comparison we also include results from four nonpolitical variables contained in the data set: extraversion and neuroticism (asmeasured plus height by items in the Eysenck Personality Quotient), and weight. Correlations

An answer can be found by looking at the interspouse correlations for themate pairs in the VA30K study. This survey was completed by the spouses of 4,387 twins as well as by 773 mate pairs with twins as offspring?a total

of the importance of shared environment. The preliminary empirical issue thus is the extent to which assortative mating takes place with regard to et al. 1999). political issue positions (see also McCourt

components estimates will again be biased. Interestingly, however, the direction of this bias is opposite to that of the EEA. The danger here is that genetic variance will be underestimated, not overestimated. As detailed above, the assumption thatDZ twins, like any other pair of biologi cal siblings, share on average 50% of the variable genetic code, is crucial to the estimation of heritability since the contrast provides the lever 50%-for-DZ/100%-for-MZ from the shared environment. age for separating genetics

for political attitudes far outstrip those for physical and personality traits. Extraverts are as likely to marry introverts as other extraverts, and the inter

spousal correlation for neuroticism is not much larger. The correlations for height and weight of spouses are positive and statistically significant but small, suggesting that taller and heavier individuals do indeed have spouses who tend to be tall and heavy but that this pattern is often violated. In direct contrast, attitudes on political and so cial items are quite likely to be shared bymate pairs. The correlation for the overall index of attitudes is extremely high (.647) and inspection of the individual items indi

The assumption that purely genetic traits inDZ twinswill on average correlate at .50 is itselfbuilt on the assump tion that their biological parents will on average correlate

der observation, the parents do not share the same genes. mate choice This assumption, however, is violated if itself is based on the trait of interest. If, in the extreme

at .00 for the same traits, and that these traits are not on the genetically influenced. In short, the CTD is built that,with regard to the phenotypic trait un assumption

cates why. Though the correlations for some of the less salient items, such as military drill, modern art, federal housing, and censorship, aremodest, most others are sub stantial and, as was the case in Table 1, the correlations for hot-button issues such as school prayer, abortion, gay rights, and living together are very high. Spousal pairs tend to share the same political and social attitudes. Of course, some of this interspouse sim ilarity could be the result of assimilation over the course

a example, parents have identical genetic codes for traitof then the shuffling of that genetic code produced interest, by sexual reproduction will not result in any variation among DZ twins (or any other siblings) with regard to their genotype for that trait. For this particular pheno type,DZ twins of these parents will be as genetically alike as MZ twins (Eaves 1979; Heath et al. 1985). In such a

of a relationship or to social homogamy (the tendency of people tomate with those around them). However, Mar tin et al. (1986) find that the correlation between mates is due primarily to assortation and not to convergence.

8 4

. PETER HATEMIET AL

Table 2

Spousal Concordance on the 28 Individual Wilson-Patterson Items, Ranked, and Selected Nonpolitical
Traits
Pearson's Statistical

move

Corr. Nonpolitical Items EPQ neuroticism EPQ extraversion Height Weight

Liberal-Conservative Index

Significance .000 .750 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 .000 4991 4739 4964 4985 3984 5002 4968 4953 4977 4906 4902 5005 4953 4912 4895 4999 4882 4955 4980 4938 4952 4923 4978 4888 4986 4889 4978 4961 4809 4933 4977 4897 4909

.082 .005 .227 .164 .647 .647 .631 .581 .573 .527 .498 .472 .462 .451 .443 .437 .412 .410 .408 .400 .392 .381 .352 .348 .343 .336 .317 .316 .304 .303 .300 .281 .253

it appears political assortative mating does occur, more accurate estimates of the effects of heritability and the environment will be obtained if parents are included in themodel.

date each other's views with the passage of the years, or but also and perhaps primarily because?knowingly select mates in part on these views unknowingly?they in the first place. Because

(and findmates) in environments filled with people like them and not only because they grow to accommo

Individual W-P Items School prayer Abortion Gay rights Living together
Democrats

Measurement

Environment
Measurement

Error and theUnique

Republicans
X-rated movies

verse 1964; Zaller 1992). When respondents change their answers to the same item, suspicion grows that researchers are picking up noise or error. Error of this sortmay cre ate a particular problem for variance components mod

error is always a concern but especially with survey items forwhich respondents frequently provide answers that do not reflect their true feelings (see Con

Unions Liberals Capitalism Death penalty Moral Majority Divorce Women's The draft Nuclear power Property tax Busing Socialism Foreign aid Astrology Federal housing Immigration Pacifism Segregation Modern art Military drill Censorship
Source: VA30K survey data.

procedures push eling because standard methodological the error term into estimates for the unshared (unique)

liberation

mates of the importance of both the shared environment and additive genetic influences.

environment, thus inflating the apparent importance of idiosyncratic environmental events at the expense of esti

similarity, rather than any error-prone single assessment. Repeated measures offer one approach to estimating and, thus, controlling for such short-term fluctuations, thereby making itpossible to correct estimates (Eaves 1973). Not accounting for this errormay affect conclusions concern ing the relative importance of the primary shapers of attitudes, thereby leading to erroneous interpretations. lem as well. The VA30K data set provides a solution to this prob In addition to including thousands of non

To take one example, if spousal concordance exists for a "political" phenotype (see previous section), it might be that concordance is due to long-term political expected

Repeated measures offer the difference between "re liable variance" and a measure at "one point in time."

With more

regard to most attitudes, spouses do not become similar with the passage of the years. The social ho

mogamy explanation for high spousal correlations also seems to fail as these correlations persist even among spe cific demographic categories, including religious denom

twin respondents, portions of the instrument were ad ministered again, approximately two years later (note the contrast with typical procedures that repeat items just

ination, frequency of church attendance, family income level, and education. All in all, it seems that spouses have similar political and social attitudes not only because they

weeks, days, or even minutes apart), to nearly 4,000 of the initial respondents. These two separate soundings make it possible to correct for response instability, thereby af fordingmore accurate estimates of the relative influence of additive genetic, as well as shared and unshared envi ronmental influences.

AND POLITICALBELIEFS 805 GENETICS Rather than present a separate table with these test retest results, they are built into our primary tabular re sult,Table 3. In the last two columns of this table, the test retest coefficients are presented, first formales and then for females. As can be seen, these numbers are quite low on salient items such as school prayer, abortion, the death penalty, and gay rights, but the "measurement error" is viations from the estimates that result in a change in the (minus twice log likelihood, -2LL) of 3.64 to 2, = .05). (equivalent The relationship between twins can be modeled ac as illustrated cording to the conventions of path analysis in Figure 1. Squares denote observed variables, circles de

fit of themodel

much higher precisely for those less salient responses for which sentiments could reasonably be expected to vary from one time to the next: property taxes, federal hous

note latent variables, upper-case letters denote variables, lower-case lettersdenote covariances or path coefficients, single-headed arrows represent hypothesized causal re arrows represent covari lationships, and double-headed ances between variables. The expected covariance is com

ing,military drill, pacifism, and censorship. Measuring and accounting for these differential levels of test-retest correlation greatly improves the accuracy of the estimates we are produced by the extended twin family analysis about to undertake.

chain and then summing over all possible chains (trace arrow, backwards, change direction at a double-headed and then trace forwards). Thus, the variance for an MZ twin is calculated as: (a * 1 * a) + (c * 1 * c) + (e * 1 * = a2 + c2 + e2. e) Extending the CTD, the "nuclear family" model was initially formulated by Truett et al. (1994) and Eaves et al.

in all puted bymultiplyingtogether the coefficients a

TheModel
Genetic models, including extended nuclear familymod els, can be estimated using a wide variety of approaches

(1999), and first applied to political traits by Eaves and model used Hatemi (2008). The maximum-likelihood here to identify the influences of genes and environment

models and Bayesian techniques including mixed effects a review, see Rabe-Hesketh, Skrondal, and Gjessing (for 2008). Such approaches have the advantage of being more

in the visual display of relationships in path estimation of sibling interaction models, ho diagrams, mogamy models, and more complex models of parent advantages child resemblance. SEM allows for the inclusion ofmean variates, provides a way to testmodel mits comparison of alternative models, effects for co

as implemented in theMx software package, is the stan dard approach in behavior genetics (Martin et al. 1986; Neale and Cardon 1992; Truett et al. 1994) and provides

familiar to political scientists and also allowing the use of standard social science statistical software, including SPSS, Stata, SAS, or S-PLUS. Our use of SEM modeling,

planatory purposes, the nuclear family model shown is specifically for use with opposite-sex DZ twins and their parents and is adjusted for use with each type of sibling in subsequent analyses. The same method of tracing rules is applied and each path calculated accordingly (for a de

on the reliable components of family resemblance for the social attitude items is presented in Figure 2. For ex

tailed explanation of themodel and path calculations, see Truett et al. 1994). The model presented in Figure 2 allows for (1) additive genetic influences formales and females (hm and hf) on the latent constructs that represent opin ions on each of the individual attitudes ( Johannsen 1911); (2) environmental effectsnot shared by twins or siblings,

ble more complicated models relations (e.g., parents, nontwin siblings, and half-sibs). SEM requires explicit delineation of hypotheses in terms of covariance/variance matrices. ML maxi is then used to

parameters, per and makes possi including a wider range of

(unique environment) em and ey; (3) environmental ef fects shared bymale and female siblings and DZ twins but not transmitted from parents (common environment) cm and cfy (4) additional environmental similarity between twins (MZ and DZ) because twin environments often correlate more highly than siblings, tm and tf\ (5) di rect social transmission ("vertical cultural inheritance"; Cavalli-Sforza and Feldman 1981) frommothers and fa thers to their sons and daughters (um, Ufy vm?and vy); and assortment between spouses m (correla (6) phenotypic tion between mates?"assortative mating").

mize the goodness-of-fit between observed and predicted covariance/variance matrices, yielding estimates of how well themodel fits the data. The optimization converges

at the solution when it locates the parameters that pro duce the largest log-likelihood. The resulting parameters are estimates of the magnitude of the latent sources of

genetic and environmental variance (ACE: A for additive en genetic; C for common environment; E for unique vironment). The reliability of these estimates is typically are the de expressed as 95% confidence intervals,which

In addition, the nuclear family model contains two parameters corresponding to the correlations between the genotypes and phenotypes of both parents individu ally (rgmand rgf).Under the assumption that themodel parameters are stable over generations, these can be expressed as functions of the other parameters

of

Extra-Shared Genotype Vertical

Genetic vironment Environment

Sibling Cultural Shared Unique Twin Additive EnEnvironment Measurement Inheritance Covariance Error
Environment

.422 .483W?? .188 WM Mi .249 W?? W?M-.109 Nuclear .155 WM .109-390 -.145 .133 ??M1 .148.272 .143.314Ml W?? power RepublicansMi?.451 -.166 W??? .475 Socialism Mi .415 .402 ?.229 .194 Abortion .251 * .307??M????M Ml .442 .126 .135 .199 .385 .430 .318WM
Busing .336.165 .166 .200 IMI Mi .141 1 W?? MM .390 MM -436

Pacifism .304.292 W?M W??W?M .337 .179 WMW?? ?428 W?? ?.141 .521

Political types Vaf Pheno Vam

W??? .170 MM W?? Mi Ml 362 Astrology .398 .387 .259 .198 -278 .476 Foreign ? ?M ?Mi -.246 aid.617 Mi -.126 .393 .392 .226

Vtsm Vef Vem Vcf Vcm

Vtsf Vcif Cgef Verf Vqm CgemVerm

Modern Mi Mi Ml.360 art.286 Ml Mi -360 .459 .266 .267 -.139

PartyID .130 .099 .357 .379

Ml Mi

.160 .138

.161 .223 .112 Mi NA N

.256 .241 Mi .446 M X-rated movies .193 .687 .619 Hi Ml H?MiMi-.262 --.180?482 .393 .149 W?? W??WB??? .456 The draft .162 -381 .245 .381 .302 .313 W?? W?? Living .085 Divorce W???.231 .357 |H| ?-.188 WM Capitalism W?? .399-.231 .428 W?M-406Ml .271 .127 .293 -342 .658 .223 .124 .330 .837 W?? 63 WM Mi Immigration .633 W?? together .132 WMW??? ? - 146 W?M.196.390 Mi Federal housing .102 Mi .293.114 MM HHW?M ? Mi Mi .383 IH .284 .338 ? Mi .226 ?Mi ? Death .167 .385 Democrats.281HH?.329 Ml 1 ????M?463 ?434 .134 Mi. Ml HH -.105 .246 .285 .222 .472 .195 .380 penalty ??Property .497 .369 ? .143 1?Segregation -.182 -.142 .384 .525 .195 .090 .537 tax .475 MiW?? -.237 W?M .203 W?? A"Mi .141 .431 School prayer .695 .202 .643 Ml Ml -141 Mi MM Mi MM W?W Mi.439 W??M 1 liberation ? .179 Mi' Ml -402 -314 Women's .279 .343 .191 .154 WM W?? Ml ?Military ?.143 drfll .464 .449 .246 .408 .315

Liberals Mi .390 .394 .154M .362 .312 .269 li

Censorship .250 .506 .123 .168 -.250 W??? 9A AS1

Liberalism-Conservatism .343 .378 .577.160 .014.341 .013 Ml Ml Mi Ml W?? WM W??

of

the

Nuclear

Family

Model

Estimates

Note: Estimates (p in better). grayscale significant not or .05 are =

AND POLITICALBELIEFS GENETICS

807

Figure 1 Classical Twin Design?ACE Labeled Paths

Path Diagram with

Notes: A =

additive

and measurement

Medland andHatemi (2009). presented in originally

= common or shared environment, E = unique environment genetic, C = trait value for Twin 1 and = trait value for Twin 2. error, P2 Diagram

Figure 2

Path Model forBiological and Cultural Inheritance in Kinships Father Phenotype Mather Genes

Notes: Paths are labeled as follows:hm = Additive genetic effectstomale phenotype;

hf = cm = Non-transmitted shared envi Additive genetic effects to female phenotype; to female siblings; ronment to male shared environment cf = Non-transmitted siblings; to pheno em = path from environment to phenotype ef = from environment (males); shared environment

= Mother-son cultural inheritance; vm = Father-son (females); urn = cul cultural inheritance; vf = Mother-son inheritance; uf Mother-daughter = the correlation between correlation tural inheritance; m = Phenotypic rgm spouses; = the correlation between between genotype and environment (male); rgf genotype and cultural

= = type (females); tm Additional twinshared environment(males); tf Additional twin

environment(female).Diagram originally presented inEaves andHatemi (2008).

8 8

PETER

. HATEMI

ET AL.

study by Hatemi, Medland, and Eaves (2009) found sig nificant quantitative (and in some cases qualitative) sex difference in the variance components analyses for po litical attitudes. Thus, in themodel used here males and females are not equated, but rather are estimated indepen dently within the same model. Due to the already complex nature of the model, we do not correct for genetic and environmental influences thatmay vary with age. To the extent there is interaction between genetic effectsand age, parent-offspring similaritymay be reduced and additive

ton 1974). The path for nontwin siblings is obtained by allowing the effects contributing to the twin-specific en vironment (T) to be uncorrelated in siblings. A recent

intergenerational transmission. Assuming that genetic ef fects are additive, the paths from parental to offspring genetic influence are fixed at 0.5 (Jencks et al. 1972;Mor

nontwin sibling environment, and in the table these terms are labeled as Vjsm and Vtsf- Total shared environment is simply the sum of the two distinct components of shared environmental effects, again formales and females, re spectively. Vctm (vertical cultural transmission) is the share of variation attributable to the direct, nongenetic so

(specifically, the inclusion of nontwin full siblings) makes itpossible to estimate the effects specifically attributable to the "twin" environment's being more similar than a

cial transmission from parents to their male offspring and Vctf is the same for female offspring. Cgem and Cgef tap passive genotype-environment covariance resulting from the fact that, for example, advantageous genetic forces will often provide advantageous environments. Finally and importantly, because many respondents completed

genetic effects could be confounded with nonadditive ge We leave this area for future study, but it is netic effects. important to stress that there are many ways to model genetic and environmental influences and many ways to

the survey two times separated by a two-year interval, it is possible to use this test-retest coefficient as an adjustment attempts at identifying the relative effectsof genetic and environmental forcesmerely push this error into the unshared environment term, but formeasurement this data set makes error.Most

correct for error, attitude stability, and the influence of mate similarity. We only offer an initialmeans to improve upon previous, more limited models and to correct for
known concerns.

mental

itpossible to separate unique environ effects from the error term. This error is labeled

The expected correlations between familymembers were derived from the path model (Cloninger, Rice, and Reich 1979; Duncan 1966; Truett et al. 1994; Wright 1921). Variance component estimates were obtained by applying strictmaximum-likelihood

Verm formales and Verf for females. With the notation established, we turn to the results themselves. The central finding of the table is that her itability estimates for political and social attitudes per sist even when extended family data rather than twin estimates only data are used, when maximum-likelihood rather than simple polychoric correlation transforma

ble, only the fullmodels and models environments are reported here.

In theory, reduced models evaluate the implications of omitting principal sources of individual differences, but due to the large number of phenotypes examined and the numerous models possi

procedures to the raw data. are typically fit in order to

tions are employed, when mate assortation is acknowl edged, and when repeated soundings are included for

dropping the twin

to .7 range. The individual attitudes showing the largest additive genetic influences appear to be those directed to ward school prayer and X-rated movies, with heritability being responsible for roughly two-thirds of the variation in these particular attitudes (formales and for females). "Living together" is also strongly heritable but illustrates

reliability. A quick scan down the two columns report ing additive genetic influences (one formales and one for females) indicates heritability consistently in the .3

The Results
The results generated by estimating thismodel with the VA30K data are presented in Table 3. Standard notation

procedures in behavioral genetics dictate that "A" refers to additive genetic effects, "E" to unshared or unique
environmental effects, and "C" to common or shared en

the fact that sometimes additive genetic forces are quite different for males and females. The additive genetic term is .40 formales but .84 for females. Attitudes toward gay rights and immigration are also among the items showing

vironmental

effects.Thus in the table, VAm refers to the additive genetic variance for males and Vaf to the additive variance for females just as Vem and Vef refer to genetic and females, respectively, and VCm and Vcf refer to the shared (or common) environmental effects formales and females, respectively. The extended family feature of the model the unique environmental effects formales

reduced from those generated without test-retest data since they are not artificially inflated by measurement error. These effects range from .10 to .28,

the very issues that tend to be strongly heritable, just as Tesser (1993) predicted. Our estimates of "unique" environmental effects are somewhat

would appear the the highest degrees of heritability, so it issues widely perceived to be hot-button social issues are

AND POLITICALBELIEFS GENETICS 809 typically smaller than additive genetic effects but larger than the shared environmental effects (usually less than .10). Of the two components of total shared environment, the extra-shared environment attributable to siblings' be are unique or shared with siblings and cotwins?but we should stress that these negative coefficients for genotype

ing twins appears to be themore important, but overall shared environmental effects areminimal. In addition, the design employed here is able to pro vide distinct estimates of vertical cultural inheritance. As heritance" can be seen, for both males and females, cultural "in isminimal, never over .11 and usually under

environment covariance are generally quite small and most have confidence intervals that approach zero. As discussed previously, the last two columns of

Table 3 provide separate estimates of "measurement er ror." Once this error is pulled out of the estimates for unshared (or unique) environment, we see that the ef fects of the unique environment are still larger than those of the shared environment, but the gap isnow quite small (in standard twin-only designs, unique environmental

.05. Consistent with earlier findings, party identification appears to be an exception. Here we find that vertical transmission influences are similar to those of the shared environment, but still less than the unique environment and cultural effects that come from siblings, especially
cotwins.

forces tend to dwarf shared environmental forces). More over, in relative terms, additive genetic effects are much more powerful when measurement error is taken into consideration (for examples of estimates that did not cor

Genotype by environment correlation (tqe) refers to the hypothesis that an individual's genes may influence his or her exposure to certain nonrandom environmental et al. 1999; Scarr and McCartney stimuli (see McCourt This correlation can be classified as active, in which 1983). the individual's own genes influence his or her exposures to certain environments, or passive, in which the envi ronment of an individual is influenced by the genes of a relative. For example, positive passive interaction occurs

rect for assortative mating and measurement error, see and Hibbing 2005; Hatemi, Medland, and Alford, Funk, Eaves 2009). When compared to previous estimates of the effectsof genes and the environment on variations in po litical attitudes, the important improvements made here in data, methodology, and measurement indicate a larger role of genetics. Table 3 also includes estimates for a composite at

when parents with liberal genetic predispositions, sim ply by following their own listening inclinations, increase the liberalism of their daughter by providing a childhood

titude index labeled Liberalism-Conservatism (made up of all the items in theWilson-Patterson Inventory) as well as for party identification. For the overall index of Liberalism-Conservatism, genetics accounts for approxi .34 of the variance in females and over half (.58) mately

environment rich in public radio exposure. In contrast, an example of positive active interaction occurs when the daughter decides on her own to buttress her genetic

would be an example of negative active gene-environment correlation. Finally, liberal parents tempering the unwel come evidence of genetic predispositions toward conser

liberal predispositions by choosing to attend Berkeley. If the same teenager opted to self-medicate her taste for liberalism by enrolling at Oral Roberts University, that

twin-specific environ and vertical cultural transmission (parental influ in females and just .03 in ence) account for less?.16 males. The shared environment is inconsequential. Turn ment ing to party identification, in previous analyses, party identification exhibited only modest or insignificant ge netic influence and notable common environmental ef fects (Hatemi, Alford et al. 2009). With parents and non twin siblings in the analysis, however, it appears the variance previously attributed to the common en

of the variance

in males, while

that

vatism in their daughter by shipping her to aMontessori school would be an example of a negative evocative gene environment correlation. Unmodeled active tqe may ei ther inflate or deflate the estimates of genetic effects. In the nuclear family analyses presented here, we estimate the global genotype by environment correlation that is an estimate of all genotype by environment variance.

vironment can be partitioned into equal amounts of ver tical cultural transmission and common environment. Furthermore, unique environmental influence provides a greater role than it previously indicated. However, is important to note that party identification was the one measure where retest data were not available and therefore, as mentioned above, for this variable we may be underestimating familial resemblance and inflating unique environmental estimates. Equally important is that the gene-environment correlation ispositive and over .1,making party identification the only traitmeasured forwhich such a claim can be made. In essence, party identification for adults is influenced by personal experi

most political attitudes, the effectsof social (environmen tal) forces tend to oppose the effects of genetic transmis

The coefficients for genotype-environment covari ance are slightly larger than those forvertical transmission but almost always negative, meaning that,with regard to

of important environmental forces whether those forces

sion. Interestingly, it would appear that genetic predispo are often pushing in the opposite direction as that sitions

ences, but initially influenced by home environment and

8lO cultural upbringing. Furthermore, there is some evidence that those raised with a certain party disposition tend to

. PETER HATEMIETAL

choose environments that continue to support their initial position. Table 4 provides themodel fits for the twin-specific environment of the 56 independent tests (28 items for males and females independently). For males only, two traits (living together and busing) are significantly differ ent from zero. For females, just four items (Democrats, nuclear power, capitalism, and party identification) have

many political scientists and has led them to question themethodologies involved. No research methodology, whether

of findings about the transmission of.. .social attitudes" (1999, 78). Given theway inwhich attitude formation is generally conceived, evidence of a strong heritable com ponent for social and political beliefs is a surprise to

model without

be dropped from the models without harming model fit (and, in fact, improving parsimony), and for females the twin-specific environment can be dropped from the affecting model fit for all variables ex nuclear power and party identification. Thus, while, cept as noted above, the inclusion of nontwin siblings does not directly test specificMZ and DZ twin environments,

twin-specific effects that are statistically significant. Fur thermore, with the exception of living together, divorce, and busing inmales, the twin-specific environment can

fect. Fortunately, suspected flaws generally can be ad dressed with additional data and techniques, and that iswhat we offer here. We do not pretend to address all limitations regarding twin and potential methodological studies. Indeed, numerous criticisms may arise, family from generalizability to twin chorionicity. However, in

it be survey re of aggregate data, search, laboratory experiments, analysis semistructured interviews, or familial modeling, is per

response to recent concerns directed at twin studies, we have added data on the parents and siblings of twins at two We find that utilizing proven sta different points of time.

the impact thatwas attributed to genetic inheritance?is disconfirmed by the results reported here.

raised by critics of earlier estimations?i.e., that EEA vi olations could in fact be responsible for most ifnot all of

only six of the 56 tests show twin-specific environmen tal effects that reach statistical significance. The charge

findings of heritability reported previously. Attempts to dismiss these consistently appearing results by attributing them to violations of the equal environments assumption (Beckwith and Morris 2008; Charney 2008; Horwitz et al. 2003) are increasingly unpersuasive. Genetics is connected the nature of this connection

tistical techniques, a wider range of kinships, and more parameters in the model only serves to strengthen the

All told, previous claims that additive genetic influ ences account for at least 40% of the variance in po

litical and social attitudes hold up even when more so phisticated modeling techniques are employed on data from familymembers other than just twin pairs. Mod eling twin-specific environments (by including nontwin

to specify the precursors of politics, whether by analyzing neurotransmitters and hormones (lohnson et al. 2006; Madsen 1986;McDermott et al. 2008), simulations of evo lutionary pressures (Axelrod and Hammond 2006; Orbell et al. 2004), social and economic experiments (Ostrom 1998; Sell et al. 2004; Wilson and Herrnstein 1995), in

we this study, have attempted to contribute to efforts made by a growing band of empirical political scientists seeking

to political attitudes, though is likely to be circuitous. In

siblings) may diminish heritability estimates a bit, but correcting for assortative mating (by including parents) increases heritability estimates. Moreover, eliminating the variation attributable tomeasurement error (by includ ing test-retest assessments of political phenotypes)
a more accurate measure of attitudes.

en

sures

Furthermore,

we

to see relatively weak contributions from nongenetic parental effects (socialization), with unique environmental effects being somewhere between genes continue and shared environment in importance.

voluntary physiological reactions (Lodge and Taber 2005; Mutz and Reeves 2005; Oxley et al. 2008), neuroscience (Marcus 2002; McDermott 2004; Schreiber 2005), or ge netics (Alford, Funk, and Hibbing 2005; Fowler, Baker, and Dawes

Discussion
As Eaves et al. noted

and Conclusion
in reflecting on the growing set of that genetic influences work with

with our life-sciences colleagues in efforts to specify the biological pathways that are politically relevant. Given

2008; Fowler and Dawes 2008; Hatemi et al. 2007). As important as environmental forces undoubt edly are, it is unscientific to assume without empirical tests that they are the only forces operating. Instead, we propose it ismore fruitful for political scientists towork

results documenting environmental forces to shape attitudes, "one of the truly remarkable findings to emerge from behavior genetics over the past 20 years is the replication and consistency

the subtleties of evolutionary pressures, the complexity of the genome, the intricacies of neuroanatomy, and the nuances of environmental forces, the task isdaunting, but such interdisciplinary collaborations offer the best hope of obtaining a more complete understanding of attitude formation and the source of preferences.

(Comparison p-Value Model)0.121 0.154 0.111 0.880 0.729 0.304 0.329 0.989 0.637 0.426 0.403

0.395 0.603 0.331 0.231 0.238 0.334 0.022 0.409 0.067 0.674 0.230 0.007 0.448 0.323 0.051 0.001

0.440

(Comparison p-Value Model)0.136 0.258

0.343

0.641

0.680

0.576 0.975

0.192 0.344 0.975 0.649 0.920 0.305 1.000 0.246 0.013

0.619

0.903 0.5950.418 0.011

0.964

0.177

0.181

0.130

0.362

0.020

0.113

Adfp

-2LLF

-2LLM

34154.82 draft 34152.22 The Abortion30502.48 34150.43 30515.48 0.11 Democrats 31888.90rights 0.04 1.792 4.392 0.312 Gay 34015.11 0.022 0.05 0.02 30503.31 30504.12 0.832 1.642 2.933 0.09 0.00 0.01 30513.18 30515.41 2.294 0.10 31889.210.08 Liberals 0.02 31888.92 0.05 34012.18 2.224

30704.35 30702.64 30705.52 1.705 Pacifism 2.874 0.06

Divorce 0.00 0.15 33602.85 33603.51 33598.12 5.395

30919.74 30924.67 30918.14 6.532 0.03 0.14 Busing 1.606

Party ID0.655 0.08 13666.36 0.16 13667.01 13678.45 12.095

Astrology 33479.99 0.06 0.02 33478.71 33481.09 1.282 2.38 Foreign 0.07 0.00 aid 35071.54 35069.72 1.821 1.818 0.255 25495.36 0.002 movies Death 25420.52 0.00 25418.30 X-rated 25422.52 0.03 2.221 0.09 4.221 0.10 penalty 25495.11 Federal Military 31513.26 drill 31517.00 30192.70 31513.47 30192.87 0.09 housing 30194.41 0.02 0.217 3.746 0.02 0.00 1.708 0.17 Confidence Fit Table Model Intervals 4Environment for the Twin-Specific and -2LL liberation Women's 0.04 33494.91 33494.01 0.00 33494.65 0.633 0.9

29079.09 0.001 29079.99 0.00 Property tax 0.00 0.901

34392.63 34392.84 34393.64 0.207 1.012 0.00 0.13 0.06Capitalism Immigration 0.06 0.05 29373.69 29376.20 29376.63 2.51 2.941 Republicans Socialism 31734.28 31732.94 0.09 0.02 31735.13 2.195 1.344 31911.84 31910.94 0.01 0.897 31918.62 7.679 23592.38 23592.62 0.247 23593.41 School 0.05 0.00 prayer 0.788 33942.02 33942.01 0.00 34107.69 0.01 Nuclear 33939.80 2.21 0.14 9.79 0.01 1.051 power 28403.03 Censorship 28404.89 0.09 34097.90 0.07 Segregation 34098.95 0.00 1.859 0 Living 29599.17 29593.04 6.13 29594.83 1.79 0.00 0.16 together

Modern 0.07 34593.51 34593.53 0.01 0.015 34595.79 2.261 art

Conservatism 0.05 0.01 38310.17 38304.49 38304.21 5.961 0.283 factor

Vtsf

Environment Extra-Shared 'tsm Note: bold inor Estimates font significant are = (p better). .05

Twin

Political Phenotypes

8i2

PETER

. HATEMI

ET AL.

References
R. R., Alford, John Carolyn L. Funk, and John Hibbing. 2005.
"Are Political Orientations Genetically Transmitted?" Amer

Fowler, 94.

Predict Voter Turnout." Journal of Politics 70(July): 579

James H.,

and Christopher

T. Dawes.

2008.

"Two Genes

Hatemi, Peter K., JohnR. Alford, JohnR. Hibbing, Nicholas

G. Martin, 600. and Lindon

icanPolitical ScienceReview 99 (May): 153-68.

Axelrod, 36.

of Ethnocentrism." Journal ofConflictResolution 50: 926

Robert,

and Ross

A. Hammond.

2006.

"The

Evolution

in Your Genes?" Political Research Quarterly 62(3): 584

J. Eaves.

2009.

"Is There

'Party'

Hatemi, PeterK., Carolyn L. Funk,Hermine Maes, Judy Silberg, 2009. "Genetic Influenceson PoliticalAttitudes over theLife Course." JournalofPolitics 71(3): 1141-56.
Hatemi, Sarah E. Medland, Nicholas G. Martin, and Lindon J.Eaves.

Beckwith, Jon,and Corey A. Morris. 2008. "Twin Studies of Political Behavior: Untenable Assumptions?" Perspectiveson Politics 6 (December): 785-92.
Bouchard, T. J., Jr., and Matt McGue. on Human ronmental Influences 2003. "Genetic and Envi Psychological W. Feldman. NJ: Differences."

Journalof Neurobiology 54(1): 4-45.

Cavalli-Sforza, Luigi tural Transmission Press. L., and and Marcus Evolution.

Politics Gender Gap?" Journalof "Do Genes Contribute to the 71(1): 262-76.
Peter K., Sarah E. Medland, Katherine 2007. I. Morley, An and Nicholas G. Martin. "The Genetics

Peter K.,

Sarah

E. Medland,

and Lindon

J.Eaves.

2009.

1981.

Cul

Hatemi,

Princeton,

Princeton

drew C. Heath, 88-112.

University

ofVoting: An Australian Twin Study." Behavior Genetics 37:

A. C, and and Eaves, Social L. J., 1985. "Resolving on Mate the Effects Selection." of Be

Charney, Evan. 2008. "The Uses and Abuses of Genetics for Political Science." Perspectiveson Politics 6: 299-320. Cloninger, C. Robert, JohnRice, and Theodore Reich. 1979.
"Multifactorial Inheritance with Cultural Transmission Journal and

Heath,

Phenotype havior Genetics Heath, A. C, R.

Background 15: 15-30. and N.

AssortativeMatting II:A General Model ofCombined Poly

Inheritance." genic and Cultural 31: 176-98. Genetics American Human of

Effects of Genotype and Social Environment on Alcohol Consumption inFemale Twins." JournalofStudies onAlcohol
60: 38-48.

Jardine,

G. Martin.

1989.

"Interactive

Mass Converse, Philip E. 1964. "The Nature of Belief Systems in Politics." In IdeologyandDiscontent y David E. Apter.New ed.
York: Free Press, William 206-61. Coventry, L., and Matthew C. Keller. 2005. "Estimat

Heath, Andrew C, Kenneth S. Kendler, Lindon J.Eaves, and David Markell. 1985. "The Resolution of Cultural and Bi
ological ships." Hettema, Inheritance: Behavior M., J. M. Informativeness 15: 439-65. and K. S. Kendler. 1995. "Phys of Different Relation Genetics C. Neale,

Human Genetics 8(3): 214-23.

Duncan, Otis D. 1966.

Estimates of Parameter from Extended sign: A Comparison Twin Designs." Twin Research and Classical and Twin-Family "Path Analysis:

ing the Extent of Parameter Bias in theClassical Twin De

ical Similarity and the Equal EnvironmentsAssumption in Twin Studies of Psychiatric Disorders." Behavior Genetics 25:
327-35.

American JournalofSociology 72: 1-16.

Attitudes. New York: Harcourt Brace.

Sociological

Examples."

Eagly, Alice H., and Shelly Chaiken. 1993. The Psychology of An Eaves, Lindon J.1973. "Assortative Mating and Intelligence: Analysis of Pedigree Data." Heredity 30: 199-210. Eaves, Lindon J. 1979. "The Use of Twins in theAnalysis of
Assortative Mating." H. J., Heredity 43: 399-409. Eaves, Lindon J.Eysenck, and N. G. Martin. 1989. Genes,

Horwitz, Allan, Tami Videon, Mark Schmitz, and Diane Davis. Twins and Environments: Possible Social 2003. "Rethinking Sources for Assumed Genetic Influences inTwin Research." Health and Social Behavior 44: 111-29. Journalof Jencks, Christopher,Marshall Smith,Henry Acland, Mary Jo StephenMichelson. 1972. Inequality:A Reassessment of the EffectofFamily and Schooling inAmerica. New York: Basic
Books. Bane, David Cohen, Herbert Gintis, Barbara Heyns, and

Academic Eaves,

An Culture, and Personality: EmpiricalApproach. San Diego:

Press. Lindon J., and Peter K. Hatemi. 2008. "Transmission

Wilhelm. Johannsen,
Heredity." Johnson, D. American D.

Naturalist

1911. "The Genotype Conception of

45: 129-59. E. S. Barrett, and S. P. Rosen. R. J.Cosden, 2006. "Over

P., R. McDermott, M. H. Mclntyre,

Genetics Eaves, L.

of Attitudes toward Abortion and Gay Rights." Behavior

38: 247-56. J., A. K. C. Heath, S. Kendler, N. . M. G. Martin, Kirk, H. H. Maes, 1999. M. "Com C.

Wrangham,

confidence in Wargames: Experimental Evidence on Expec tations."Proceedings of the Royal Society273: 2513-20.
G. Martin. 2005. "Widespread Evidence for Non

Neale,

paring the Biological and Cultural Inheritance of Person ality and Social Attitudes in theVirginia 30,000 Study of Twins and Their Relatives." Twin Research 2 (June): 62
80. Fisher, R. A. 1918. "The Correlations between Relatives Transactions on the

and L. Corey.

Andrew C. Heath, and Keller,Matthew C, William L. Coventry,

Nicholas

additiveVariation inCloninger's and Eysenck s Personality Dimensions Using a Twin Plus Sibling Design." Behavior Genetics 35 (November): 707-21.
Kendler, Lindon G. Martin, and S., Andrew C. Heath, Nicholas and Symp 1987. "Symptoms of Anxiety toms of Depression: Same Genes, Different Environments?" Kenneth J. Eaves. Kenneth and S., Michael L. J. Eaves. C. Neale, 1992. R. C. Kessler, Andrew Twin

Royal Society ofEdinburgh 52: 399-433.

Fowler, 2008. James H., "Genetic Laura A. Variation

Supposition

of Mendelian

Inheritance."

of the

can Political ScienceReview 102 (May): 233-48.

T. Dawes. Baker, and Christopher in Political Ameri Participation."

Archives ofGeneral Psychiatry44 (May): 451-57.

Kendler, C. Heath,

"A Population-Based

GENETICSAND POLITICALBELIEFS 813 Women: The Impact ofVary Study ofMajor Depression in Definitions of Illness."Archives ofGeneral Psychiatry49: ing
257-66. Kendler, Heath, Kenneth S., Michael C. Neale, "Major R. C. Kessler, Depression Andrew and Phobias: C. Mutz, Diana C, and Byron Reeves. 2005. "The New Video

malaise: Effects of Televised Incivility on Political Trust." American Political Science Review 99: 1-16.
Walter

Nance,

The Genetic and Environmental Sources of Comorbidity." Medicine 23 (May): 361-71. Psychological
A. C. Heath, and N. G. I. E., L. J.Eaves, H. H. M. Maes, 2000. "Further Evidence against the Environmental

and L. J.Eaves.

1993.

Analysis: A Power Study." Behavior Genetics 19: 143-50.

M.

E., and Michael

C. Neale.

1989.

"Partitioned

Twin

Neale,

tical Modeling. 5th ed. Richmond: Virginia Commonwealth

C,

S. Boker,

G. Xie,

and H. Maes.

2003. Mx:

Statis

Lake, R. Martin.

University.

Neuro ticismfrom Transmission of IndividualDifferences in a Collaborative Study of 45,850 Twins and Relatives on Two
Continents." Behavior Genetics 30: 223-33.

Neale, M. C, and L. R. Cardon. 1992.Methodology for Ge neticStudies ofTwins and Families. Dordrecht, Netherlands:
Kluwer Academic.

Lodge, Milton, and Charles S. Taber. 2005. "The Automaticity ofAffect for Political Leaders, Groups, and Issues: An Ex perimental Test of theHot Cognition Hypothesis." Political Psychology26: 455-82. inTwins?" Developmental Psychology 13: 456-59. Madsen, Douglas. 1986. "Power SeekersAre Different:Further Biochemical Evidence." American Political Science Review
80: 261-69. Maes, Hermine H., Michael C. Neale, and Lindon J. Eaves. Lytton, H. 1977. "Do Parents Create, or Respond to, Differences

M? Philip V. Vernon, JulieAitken, and Kerry Olson, James L. Jang. 2001. "The Heritability of Attitudes: A Study of Twins." Journalof Personality and Social Psychology 80: 845
60.

T. Orbell, J., Morikawa, J. Hartwig, J. Hanley, andN. Allen. 2004. "Machiavellian Intelligence as a Basis for the Evolution of Cooperative Dispositions." American Political ScienceReview
98: 1-16.

Ostrom, Elinor. 1998. "A Behavioral Approach to the Ratio nal Choice Theory of Collective Action." American Political
Science Review 92: 1-22.

1997. "Genetic and Environmental Factors in Relative BodyWeight and Human Adiposity." Behavior Genetics 27:
325-51.

Marcus, George E. 2002. The Sentimental Citizen: Emotion in Democratic Politics.University Park: The Pennsylvania State
University Martin, Press. Nicholas G., Lindon J. Eaves, A. C. Heath, R. Jardine,

R. Oxley, Douglas R., Kevin B. Smith, John Alford,Matthew V. L. Hibbing, Jennifer Miller, Mario Scalora, PeterK. Hatemi, and JohnR. Hibbing. 2008. "Political Attitudes Vary with Physiological Traits." Science 321 (19 September): 114-18. R. Persuasion: Communication Perloff, M. 2003. TheDynamics of and Attitudes in the21stCentury.Mahwah, NJ: Erlbaum. Peterson, StevenA. 1983. "Biology and Political Socialization: A Cognitive Developmental Link?" Political Psychology 4:
265-88. Plomin, R., L.Willerman, and C. Loehlin. J. 1976. "Resemblance

L. M. Feingold, and H. J.Eysenck. 1986. "Transmission of Social Attitudes." Proceedings of theNational Academy of Sciences 15 (June): 4364-68.
A.

Matheny,

6: 343-51. McCourt, Keyes.

tions between Twins' Similarity ofAppearance and Behav ioral Similarity:Testing an Assumption." Behavior Genetics
K., T. 1999. D. T. Lykken, A. Tellegen, J.Bouchard, Revisited: Genetic "Authoritarianism and M. and En

P., R.

S. Wilson,

and A.

B. Dolan.

1976.

"Rela

inAppearance and the Equal Environments Assumption in Twin Studies of Personality Traits." Behavior Genetics 6(1):
43-52. Danielle 30: A., and Dorret I. Boomsma. 2000. "A Note

Posthuma,

vironmental Influences Examined in Twins Reared Apart and Together." Personality and Individual Differences 27:
985-1014.

Genetics

on theStatisticalPower inExtended Twin Designs." Behavior

147-58.

McDermott, Rose. 2004. "The Feeling ofRationality: TheMean ing ofNeuroscientific Advances forPolitical Science." Per spectiveson Politics 2(4): 707-24. Cowden, Giovanni McDermott, Rose, Dustin Tingley, Jonathan Oxidase A Gene (MAOA) Predicts Behavioral Aggression National Academy Following Provocation." Proceedingsof the ofSciences 106(7): 2118-23.
Medland, Sarah Frazzetto, and Dominic D. P. Johnson. 2008. "Monoamine

Danielle C. de Geus, G. Caroline A., Leo Beem, Eco J. Posthuma, M. van Baal, Jacob B. von Hjelmborg, Ivan Iachine, and Dor in Quantitative ret I. Boomsma. 2003. "Theory and Practice

Genetics." Twin Research 6(5): 361-76.

S., A. Skrondal, and .K.

Rabe-Hesketh,

metrical Modeling ofTwin and FamilyData Using Standard Mixed Model Software."Biometrics 64: 280-88. Defense of a Critical Assumption." Behavior Genetics 9:
527-42. Sandra, and K. McCartney. 1983. Child "How Peo Sandra, and L. Carter-Saltzman. 1979. "Twin Method:

Gjessing.

2008.

"Bio

Scarr,

ical Science, Biometrie Theory, and Twin Studies: A Methodological Introduction."PoliticalAnalysis 17(2): 191
214.

E.,

and

Peter

K.

Hatemi.

2009.

"Polit

Scarr,

ple Make

Their Own
Environment

Environments: A

Genotype? 424-35.

Effects."

Theory

of
54:

Development

Merelman, Richard M. 1971. "The Development of Policy Adolescence." American Political ScienceReview Thinking in 65(4): 1033-47.
N. Morton, duction." 30. of Family Resemblance: "Analysis Genetics American Journal of Human 1974. Intro 26: 318

Schreiber,Darren. 2005. "Evaluating Politics: A Search for the Neural Substrates of Political Thought." PhD dissertation, UCLA. A., Spaeth. 1993. The SupremeCourt Segal, Jeffrey and Harold J. Attitudinal Model. New York: Cambridge University and the
Press.

8l4 Michael J. Sell, Jane, Lovaglia, Elizabeth A. Mannix, Charles D. Samuelson, and RickWilson. 2004. "Investigating Conflict, Power, and Statuswithin and among Groups." Small Group Research 35(1): 44-72. Tesser, A. 1993. "The Importance ofHeritability in Psycholog ical Research: The Case of Attitudes." PsychologicalReview 100 (January): 129-42.
Truett,

. PETER HATEMIETAL

Sharing between Full Siblings." PLoS Genetics 2(3): 316?

24. Wilson, Glen D., and John R. Patterson. 1968. "A New Mea

sure ofConservatism." BritishJournalofSocial and Clinical Psychology7: 264-69.

James Q., Nature. Sewall. and New Richard York: Herrnstein. Simon 1995. Crime and and Schuster. and Causation." Journal of

Wilson, Human Wright,

C. Heath, JohnK. Hewitt, JoanneM. Meyer, JudyL. Sil berg,Michael C. Neale, Nicholas G. Martin, and Kenneth S. Kendler. 1994. "AModel System for theAnalysis of Fam ilyResemblance in Extended Kinships of Twins." Behavior
Genetics 24: 35-49. M., Peter Sarah K. E. Medland, Cornes, 2006. Grant Katherine W. I. Mor

Kimberly

R., Lindon

J.Eaves,

Ellen

E. Walters,

Andrew

1921. Research

"Correlation 20: 557-85.

Agricultural Xian, H.,

J. F. Sherrer, S. A. Eisen, W. R. True, A. C. Heath, 2000. "Self-Reported J.Lyons, and M.T. Tsuang. Golberg, M.

J.

Visscher,

Zygosity and theEqual EnvironmentsAssumption forPsy chiatricDisorders in theVietnam Era Twin Registry."Be
havior Genetics 30: 303-10.

ley, Gu Zhu, and Nicholas

Belinda G. Martin.

tionofHeritability from Genome-Wide Identity-by-Descent

Montgomery, Estima "Assumption-Free

R. Mass Opinion. Zaller, John 1992. The Nature and Origins of New York: Cambridge University Press.