Copyright 1998 by the American Psychological Association, Inc.
0022-3514/98/S3.00
Journal of Personality and Social Psychology
1998, Vol. 74, No. 3, 818-827
Genetic and Environmental Influences on MMPI Factor Scales:
Joint Model Fitting to Twin and Adoption Data
Jeremy M. Beer, Richard D. Arnold, and John C. Loehlin
University of Texas at Austin
In general, the shared family environment appears to play a negligible role in determining individual
differences in personality and interests. Nevertheless, scattered reports of significant shared environmental influence on such variables appear in the literature. Using data from die Texas Adoption
Project (TAP), the current study attempted to replicate twin study findings of significant shared
environmental variance on four of nine Minnesota Multiphasic Personality Inventory (MMPI) factor
scales (Rose, 1988). Conventional behavioral genetic analyses of the adoption data agreed in affirming a significant shared environmental influence on individual differences in Religious Orthodoxy
only. Subsequent simultaneous modeling of Rose's twin data and TAP adoption data resulted in three
scales (Extraversion, Inadequacy, and Religious Ortfiodoxy) showing significant shared environmental influence. Again, effects were most substantial for Religious Orthodoxy, where the shared environment accounted for nearly 50% of the variance. It is argued that assortative mating cannot explain
this finding.
(1993) found significant shared environmental effects on a measure of agreeableness. Loehlin and Gough (1990) came to the
same conclusion regarding Norm-Favoring, a scale derived from
the California Psychological Inventory. Complementing these
personality studies is one showing that, once contrast effects in
parental ratings are taken into account, the shared environment
emerges as a significant source of variance in temperament as
well (Saudino, McGuire, Reiss, Hetherington, & Plomin, 1995).
A twin study by Rose (1988) is particularly noteworthy because it included measures of beliefs and interests as well as
personality. Rose used nine MMPI scales derived via principalcomponents analysis by Costa, Zonderman, McCrae, and Williams (1985). Shared environmental influence proved significant on four: Masculinity-Rsmininity (64%), Extraversion
(24%), Religious Orthodoxy (61%), and Intellectual Interests
(40%). Rose's sample was comparatively large (228 monozygotic [MZ1 and 182 dizygotic [DZ] pairs); at the time it was
the largest twin study ever conducted using data from the MMPI.
Of course, among the large number of behavioral-genetic
studies that exist it may be expected that findings like those
noted earlier will sometimes result from chance alone; particular
results must be replicated before they are taken seriously, preferably via multiple methods. With respect to Rose's findings, data
from the Texas Adoption Project (TAP; Horn, Loehlin, & Willerman, 1979) provide such an opportunity. The adoption design
yields direct measures of both shared environmental and genetic
variation, rather than relying on differences between correlations, as does the twin study. In addition, it is less vulnerable
to certain twin-study criticisms, such as the assumptions of
equal environments for identical and fraternal twins and random
mating with respect to the trait in question. TAP data allow
for the assessment of similarity among genetically unrelated
individuals reared together and genetically related persons
reared apart (Loehlin, Horn, & Willerman, 1981). For these
reasons and others (twin and adoption methods are differently
affected by age differences, and by the presence of nonadditive
Numerous behavioral genetic studies have concluded that
shared familial environmental variation is an unimportant
source of individual differences in personality and interests
(Betsworth et al., 1994; Loehlin & Nichols, 1976; Loehlin,
Willerman, & Horn, 1987; Pederson, Plomin, McClearn, & Friberg, 1988; Plomin & Daniels, 1987; Rowe & Plomin, 1981;
Scarr, 1992). Although originally counterintuitive (Bouchard,
Lykken, McGue, Segal, & Tellegen, 1990; Costa & McCrae,
1987; Tellegen et al., 1988), this claim is rarely greeted with
surprise today, at least among behavioral geneticists. In fact,
recent reports have indicated that shared environmental variance
may be negligible in determining individual differences in social, political, and religious attitudes and beliefs as well (Waller,
Kojetin, Bouchard, Lykken, & Tellegen, 1990), especially when
assortative mating for these traits is taken into account (Eaves,
Eysenck, & Martin, 1989; Martin et al., 1986).
However, scattered throughout the literature there remain a
small number of studies drawing the opposite conclusion regarding a variety of personality, temperament, and attitude measures.
Tellegen et al. (1988) reported significant shared environmental
influence on measures of two extraversion-related traits, Positive
Emotionality and Social Closeness, in a twin study using their
Multidimensional Personality Questionnaire. Bergeman et al.
Jeremy M. Beer, Richard D. Arnold, and John C. Loehlin, Department
of Psychology, University of Texas at Austin.
This material is based on work supported under a National Science
Foundation graduate fellowship. The research on which this article is
based was supported by Grant MH-24280 from the National Institute
of Mental Health and Grants BNS-7902918 and BNS-8209882 from
the National Science Foundation. We thank Joe Horn and the late Lee
Willerman for their helpful comments in reviewing the manuscript and
Paul Costa for providing the factor scale item key.
Correspondence concerning this article should be addressed to Jeremy
M. Beer, Department of Psychology, University of Texas at Austin, Mezes
Hall 330, Austin, Texas 78712. Electronic mail may be sent to
beer @ mail.utexas.edu.
818
819
GENETIC AND ENVIRONMENTAL VARIANCE IN THE MMPI
genetic variance), the adoption design can produce different
results than the twin design for the same traits (Tellegen et al.,
1988), and thus provides a means of testing the validity and
generalizability of Rose's conclusions.
The adoption paradigm also allows for a direct test of the
hypothesis that assortative mating can account for variation that
might otherwise be ascribed to the family environment. This
hypothesis has recently been offered as an alternative explanation for what appear to be substantial effects of the shared
environment on measures of social and religious attitudes
(Eaves, Eysenck, & Martin, 1989). Phenotypic assortment for
a trait results in greater genetic similarity between first-degree
relatives than the assumed 50%, spuriously inflating correlations
between dizygotic twins and confounding conventional twin
analyses (Rowe, 1994). When only twin data are available,
models may be created that attempt to account for this confound,
but they must make many assumptions and thus may not adequately reflect reality. Adoption data can avoid this difficulty
(Eaves, Martin, & Heath, 1990; Truett, Eaves, Meyer, Heath, &
Martin, 1992). Assuming there has been no selective placement,
any similarity between genetically unrelated persons who have
shared the same environment, that is, adoptees and their adoptive
parents and siblings, must be due to that environment, not assortative mating.
To augment conventional heredity-environmental analyses of
the adoption data, the present study uses model fitting to simultaneously examine both the twin and adoption data. Structural
equation modeling is widely used within behavior genetics because, unlike traditional behavioral genetic analyses, it allows
the investigator to arrive at unique estimates of genetic and
environmental variance using data from multiple genetically informative relationships. Fitting models to twin and adoption
data simultaneously also helps circumvent the unique problems
associated with each design.
Finally, we interpret our modeling results for these nine factor
scales in light of the "Big Five" personality dimensions of
Neuroticism, Extraversion, Openness, Agreeableness, and Conscientiousness. Although there are several recent twin studies
that employ personality measures explicitly measuring these
dimensions (Angleitner, Riemann, & Strelau, 1995; Beer &
Rose, 1995; Bergeman et al., 1993; Goldsmith, Losoya, Bradshaw, & Campos, 1994), there are no comparable adoption data
available. Even though our scales are not comprehensive in their
coverage of the Big Five (Costa, Busch, Zonderman, & McCrae,
1986), the correlations of the MMPI factor scales with Big
Five dimensions have been reported (Costa et al., 1986), and
therefore these data can provide a partial check on conclusions
derived from twin studies.
Method
Sample
In the Texas Adoption Project's first study, 300 families who had
adopted at least one child through a Texas church-related home for
unwed mothers were given a battery of psychological tests. Because the
home routinely administered personality, intelligence, and vocational
interests tests as part of the counseling services provided to residents,
psychological data were available for the unwed mothers as well. Details
on this sample are available elsewhere (Horn, Loehlin, & Willerman,
1979).
At the time of the first study, most of the adoptees were too young to
be administered the MMPI and other self-report personality inventories.
However, approximately 10 years after the original study a follow-up
study was conducted, at which time the MMPI was administered. Of
the 300 original families, 181 (60%) agreed to participate in the followup, including 258 (55%) of the 469 adopted children and 95 (57%) of
the 167 biological children of the adoptive parents. Thirty-four children
(17 adoptees and 17 biological children) mat did not participate in the
first study were included in the follow-up. Further details are available
in Loehlin et al. (1987).
In the present study, MMPI data were analyzed only for tiiose adopted
and biological children whose data, were judged valid and who were at
least 14 years old when they took the MMPI (Loehlin et al., 1987).
This resulted in the inclusion of 186 adoptees and 80 biological children,
with mean ages of 17,7 and 20.2, respectively. For these children, the
MMPIs of 150 adoptive fathers, 148 adoptive mothers, and 130 biological
mothers were available from the first study; the mean ages of these
groups at the time of testing were 40.4, 37.8, and 19.0, respectively.
Measures
The MMPIs for all participants were scored for nine factor scales
identified by Costa et al. (1985) and labeled by them as Neuroticism,
Psychoticism, Masculinity-femininity, Extraversion, Religious Orthodoxy, Somatic Complaints, Inadequacy, Cynicism, and Intellectual Interests. Their principal-components analysis was the first ever of the entire
MMPI item pool based on a psychiatrically normal sample. Although
not believed by Costa et al. to represent adequately the entire range of
normal personality, these scales have desirable statistical properties and
may be more useful than the MMPI clinical scales in research on normal
personality variation.
In the Texas Adoption Project, only the adopted and biological children
received the full MMPI. Because birth mothers had received only a
shortened version (373 items covering the clinical and validity scales),
the adoptive parents were administered this version as well. For consistency, therefore, scale scores were calculated for all groups using only
those items that appeared on die shortened MMPI. (The Appendix lists
the items comprising each scale. See Costa et al., 1985, for descriptions
of the scales.) An examination of the correlations between the short and
full versions of the scales in Table 1 shows that this strategy appears to
have been satisfactory; only for Religious Orthodoxy does the correlation between the two versions drop below .9, and that is above .8.
Given these correlations, it seemed reasonable to conclude that we
Table 1
Correlations Between the Shortened and Full Forms
of the MMPI Factor Scales
Number of items
Factor scale
Short form
Full scale
Correlation"
Neuroticism
Psychoticism
Masculinity -Femininity
Extraversion
Religious Orthodoxy
Somatic Complaints
Inadequacy
Cynicism
Intellectual Interests
42
96
20
11
11
39
18
22
7
65
120
39
23
26
44
30
37
11
.977
.937
.924
.906
.830
.988
.942
.947
.913
Note. MMPI = Minnesota Multiphasic Personality Inventory. Correlations are based on adoptive and biological children data, for which full
MMPIs were available.
*N = 266.
820
BEER, ARNOLD, AND LOEHLIN
could make valid comparisons between our results and those reported
by Rose (1988).
Table 3
Correlations Between Adoptive Parents
on MMPI Factor Scales
Analyses
Factor scale
Conventional
With adoption data, estimates of the relative contributions of genetic, shared environmental, and unshared environmental variance to individual differences on a given trait are relatively straightforward. Those individuals who share no genes, but do share an environment, are similar only because of that environment. Therefore,
correlations between nonbiological siblings are direct measures of
shared environmental influence. Those individuals who do not share an
environment but do share genes are similar only because of their shared
genes. Twice the correlation between biological mothers and their
adopted-away offspring is a direct estimate of genetic variance (the
correlation must be doubled because they share only half their genes).
Other genetic variance estimates depend on differences between correlations. Adoptive parents' correlations with their biological children minus the same correlations with their adoptive children is an estimate of
half the genetic variance and must be doubled. Twice the difference
between biological sibling and adoptive sibling correlations provides
another estimate of genetic variance.
The correlation between adoptive parents and their adopted children
is also a direct measure of environmental effects. It is not, strictly
speaking, a measure of "shared" environment, because it may reflect
other environmentally mediated processes, such as direct influence of
the parent on the child, but for simplicity it will be treated as shared
environment in most of our analyses.
For the preceding relationships to hold, certain assumptions must be
met. The genetic effects are assumed to be additive and stable across
the age ranges involved. In adoption studies, another important assumption is that adoptees have been placed in homes randomly with respect
to the trait(s) in question; that is, there has been no selective placement
of children into adoptive families that are similar to their biological
families (again, with respect to the traits being analyzed). If selective
placement is absent, biological parents' and adoptive parents' scores
should be uncorrelated, and this appears to be nearly so for our data.
Table 2 presents correlations between biological mothers and adoptive
parents for each of the nine scales. None is large enough to suggest a
substantial bias from selective placement, and most are consistent with
random fluctuation around zero.
Table 2
Correlations Between Birth Mothers and Adoptive
Parents on MMPI Factor Scales
Factor scale
Birth motheradoptive
mother"
Birth motheradoptive
father"
Average
Neuroticism
Psychoticism
Masculinity-Femininity
Extraversion
Religious Orthodoxy
Somatic Complaints
Inadequacy
Cynicism
Intellectual Interests
.147
.131
-.120
.134
-.018
-.057
.226
.073
-.056
.008
.177
.115
.014
-.121
.140
-.081
.066
-.073
.077
.154
-.003
.074
-.069
.042
.073
.069
-.065
.051
.073
.027
.014
.039
.069
M
Mdn
Note. Correlations are based on data corrected for the effects of age,
sex, and birth-mother-child status. MMPI = Minnesota Multiphasic
Personality Inventory.
'N = 127. *N= 129.
Neuroticism
Psychoticism
Masculinity—Femininity
Extraversion
Religious Orthodoxy
Somatic Complaints
Inadequacy
Cynicism
Intellectual Interests
.157
.215
.036
.003
.425
.123
.076
.132
.018
M
Mdn
.128
.123
Note. Correlations are based on data corrected for the effects of age
and sex. MMPI = Minnesota Multiphasic Personality Inventory.
"iV= 148.
Another assumption, which is relevant only to those pairings in which
individuals are biologically related, is that there has been no assortative
mating between spouses. Because evidence suggests that spouses do not
become more similar during their marriage (Buss, 1984; Feng & Baker,
1994; Mascie-Taylor, 1989), we can estimate the validity of this assumption by simply looking at spousal correlations for the factor scales (see
Table 3).
The assumption of no assortative mating is clearly violated in the
case of Religious Orthodoxy. For the other traits, assortative mating
appears to be fairly modest.
Model fitting. Although they have the benefit of simplicity, traditional behavioral genetic analyses do not have the potential to examine
simultaneously several subsamples. Model fitting allows the researcher
to do just that. In order to maximize the power of the analyses, we fit
models jointly to our adoption data and Rose's (1988) twin data. Modeling was carried out using Mx, a modeling program developed by Neale
(1994). This procedure also allows for explicit tests of the presence of
effects usually assumed either absent or present in conventional analyses,
such as assortative mating, nonadditive genetic influences, and special
common environmental effects. If any of these are significant, they can
then be included in the model.
For all scales, our full model included parameters for additive genes,
dominance, emergenesis (in which a particular configuration of alleles
interact at different loci to create trait variance; see Lykken, 1982; Lykken, McGue, Tellegen, & Bouchard, 1992), the shared environment, the
unshared environment, scale reliability, and measurement error. Figure
1 depicts the model in the form of a path diagram.
The reliability and measurement error paths were set equal to the
square root of the scale reliability and the square root of one minus the
reliability, respectively. Reliability was estimated using coefficient alpha;
these values are reported in Table 4 for each scale.
Short-form alphas were calculated using TAP data, whereas alphas
for the long form were estimated by calculating the unweighted mean
of the values reported in the literature for three samples (Costa et al.,
1985; Popham & Holden, 1991). All variances were fixed at 1.0, and
all other parameters were estimated by the model.
Results
Data were first adjusted for age and sex in two separate
regressions, one based on the adoptive parents, and one on the
children and birth mothers. In the first, the adoptive parents'
scores were regressed on age and sex. In the second, the scores
GENETIC AND ENVIRONMENTAL VARIANCE IN THE MMPI
821
MZ = 1.0; DZ, bio. sib, bio. p-c «.50; others * 0
MZ - 1.0; DZ, Wo. sib * .25; others - 0
Figure I. Path model of sources of variation on Minnesota Multiphasic Personality Inventory factor scales.
MZ = monozygotic twins; DZ = dizygotic twins; bio. sib = biological siblings; bio. p-c - biological
parent and child; b.m,-child = birth mother and adopted-away child; a = additive genetic effects; d =
dominance; i - eroergenesis; c = common environment; \ — latent trait; o - observed trait; e = unshared
environment; r - scale reliability; m = measurement error.
of the birth mothers and children were regressed on age, sex,
and group (birth mother or child). Separate regressions were
done because there was almost no overlap in the age distribution
of the adoptive parent group and the children/birth mother
group- In general, these adjustments made little difference. Excepting Masculinity-Femininity and correlations between biological siblings, the maximum difference between correlations
calculated using adjusted and unadjusted scores was .04. Because of smaller sample sizes, the correlations between biological siblings were somewhat more variable. Among the scales,
Table 4
Alpha Coefficients for the Full and Shortened Forms
of the MMPI Factor Scales
Factor scale
Short form2
Full form
Neuroticism
Psychoticism
Masculinity -Femininity
Extraversion
Religious Orthodoxy
Somatic Complaints
Inadequacy
Cynicism
Intellectual Interests
.878
.901
.818
.678
.607
.805
.814
.842
.528
.923
.897
.843
.687
.800
,887
.847
.827
.663
Note. Short-form alpha coefficients were calculated using Texas Adoption Project data. Full-form alpha coefficients were calculated by taking
a simple average of alpha coefficients reported in the literature for these
scales (see text). MMPI = Minnesota Multiphasic Personality Inventory.
11
#=985.
scores on Masculinity -t^rnininity changed the most. In fact,
mean differences between males and females on this trait were
so large that it is questionable whether the residuals carry much
information. Together age and sex explained nearly 80% of the
variance in Masculinity -femininity among adoptive parents,
and age, sex, and group status explained 64% of the trait's
variance among children and birth mothers.
Conventional Analyses
Using residual scores throughout, Pearson product-moment
correlations were calculated between adoptees and their biological mothers and between adoptive parents and their adopted and
biological children (see Table 5 ) .
Because many families adopted more than one child or had
more than one biological child, many adoptive parents were
included more than once in calculating these correlations. The
degrees of freedom, therefore, reflect the number of separate
families included in the calculations, rather than the number of
pairs. Double-entry matrices were used to calculate intraclass
correlations between siblings, which are also reported in Table
5. When there are more than two siblings in a family, those
individuals were included in more than one sibling pair. Degrees
of freedom were adjusted to allow for double-entry and multiple
sibships.
Inspection of Table 5 suggests that, despite some inconsistency within scales, genetically related individuals, whether they
have lived together or not, tend to resemble each other more
than do individuals who are not biologically related. Furthermore, with the exception of Religious Orthodoxy, the effects of
822
BEER, ARNOLD, AND LOEHLIN
Table 5
Correlations Among Adoptive Siblings, Biological Siblings, Adoptees With Birth Mothers,
and Adoptive Parents With Their Adopted and Natural Children on MMPI Factor Scales
Factor scale
Neuroticism
Psychoticism
Masculinity Femininity
Extraversion
Religious
Orthodoxy
Somatic
Complaints
Inadequacy
Cynicism
Intellectual
Interests
M
Mdn
Natural
childrenadopted
mothersg
Biological
siblings*1
Adoptees adopted
fathers"
.197
.169
.242
-.045
.133
.248
-.011
.013
.052
.066
.194
.033
.180
-.010
.020
.051
.220
.288
.073
.125
.137
.074
-.074
.048
-.024
.179
-.040
.162
.357
.531
.034
.215
.207
.348
.430
.130
.164
.026
-.224
.368
.123
.080
.269
.314
.019
.074
-.010
-.103
.190
-.095
.056
.047
.233
.155
.243
-.003
-.037
.000
.140
-.042
-.087
.010
.171
.120
.130
.167
.220
.157
.133
.052
.019
.023
.048
.119
.056
.143
.162
Adoptive
siblings"
Adopteesadopted
mothers6
Natural
childrenadopted
fathers'
Adoptees birth
mothers"
Note. Correlations are based on data corrected for age and sex. Birth mother-adopted-away offspring
correlations also corrected for status (birth mother or not). MMPI = Minnesota Multiphasic Inventory.
a
t f = 2 3 2 ( 8 3 4 / ) . bN = 48 (20 df). 'N = 131 (130 df), d N = 182 (138 df). eN= 179 (136 df).
f
N = 79 (57 df).
S
N = 79 (57 df).
the shared environment appear to be weak. Average heritability
and shared environmental estimates for each scale, and their
ranges, are reported in Table 6.
These are simple genetic and environmental estimates, calculated as described earlier. The genetic estimates, h2, were obtained as twice the biological mother-child correlation, twice
the difference in the correlations of parents with their biological
and adoptive children, and twice the difference in the correlations between biological and adoptive siblings. The shared environmental estimates, c2, were obtained from adoptive sibling
correlations and the correlations of each adoptive parent with
his or her adopted child.
Relatively consistent, although moderate, genetic effects are
shown for Extraversion and Neuroticism, as well as Cynicism,
Religious Orthodoxy, and perhaps Inadequacy and Intellectual
Interests. Religious Orthodoxy shows a consistent effect of
shared environment as well, as does, to a lesser extent, Inadequacy. Little consistency for either genetic or environmental
effects is shown by Psychoticism, Masculinity-Femininity, or
Somatic Complaints.
Model Fitting
Using Mx (Neale, 1994), structural equation models were fit
simultaneously to the adoptive family correlations and the twin
Table 6
Averages and Ranges of Genetic and Shared Environmental
Variance Estimates for MMPI Factor Scales
h2
c2
Factor scale
Average
Low
High
Average
Low
High
Neuroticism
Psychoticism
Masculinity-Femininity
Extraversion
Religious Orthodoxy
Somatic Complaints
Inadequacy
Cynicism
Intellectual Interests
.256
-.011
.073
.291
.282
.011
.250
.373
.207
.090
-.428
-.322
.210
.068
-.708
-.054
.184
-.074
.410
.496
.400
.474
.446
.516
.538
.628
.516
.079
.083
.028
.058
.260
.015
.143
-.026
-.055
-.011
.013
-.074
.048
.207
-.103
.074
-.095
-.087
.197
.169
.137
.074
.357
.130
.190
.026
-.037
Note. Averages are unweighted means of four genetic variance estimates and three shared environmental
variance estimates calculated from the correlations in Table 5 as described in the text. MMPI = Minnesota
Multiphasic Personality Inventory, h2 = genetic variance, c2 — shared environmental variance.
GENETIC AND ENVIRONMENTAL VARIANCE IN THE MMPI
Table 7
Intraclass Correlations Between MX and
DZ Twins on MMPI Factor Scales
Factor scale
MZ a
DZ b
Neuroticism
Psychoticism
Masculinity -Femininity
Extraversion
Religious Orthodoxy
Somatic Complaints
Inadequacy
Cynicism
Intellectual Interests
.406
.722
.546
.602
.722
.431
.539
.507
.563
.208
.388
.394
.413
.668
.231
.202
.339
.483
M
Mdn
.560
.546
.370
.388
Note. Correlations are based on those reported by Rose (1988). Here
they are corrected for the effects of sex. MZ = monozygotic; DZ =
dizygotic; MMPI = Minnesota Multiphasic Personality Inventory.
a
N = 228. bN= 182.
correlations reported by Rose (1988; Rose's correlations were
adjusted for sex but not age. See Table 7 ) .
The model was thus fit to data from nine subsamples: adoptive
fathers and their adopted children, adoptive fathers and their
natural children, adoptive mothers and their adopted children,
adoptive mothers and their natural children, birth mothers and
their adopted-away offspring, adoptive siblings, biological siblings, dizygotic twins, and monozygotic twins. Chi-squares and
parameter estimates were calculated for the full models for each
factor scale and are reported in Table 8. For Masculinity-Femininity, the model was also fit separately for males and females.
We then attempted to reduce each model by testing parameters
in the following stages: (a) shared environment, (b) dominance
and emergenesis (if the deletion of the dominance and
emergenesis parameters resulted in a significantly poorer fit,
each was tested for significance individually), and (c) additive
genetic effects. If, after dropping a parameter, the result was a
significant increase in the model's chi-square (a = .05), that
parameter was retained in the model. Furthermore, if the shared
environmental parameter was retained, we tested whether the
model could be significantly improved by allowing the shared
environmental parameters to differ for sibling and parent-child
relationships. Finally, the reduced model had to make sense;
for example, the most parsimonious model for MasculinityFemininity was one including dominance but not additive genetic variance. However, we report the model including both
additive genetic effects and dominance as being biologically
more plausible.1 The reduced models for each scale are reported
in Table 9. Degrees of freedom were adjusted for correlation
matrices throughout according to the method described in Neale
and Cardon (1992).
Note that for some scales, Psychoticism, Masculinity-Femininity, and Intellectual Interests, the fit of the reduced models
did not meet the rule-of-thumb criterion that the chi-square per
degrees of freedom should be less than 2 (Loehlin, 1987).
However, even for these factors the fits were not egregiously
poor, with the possible exception of the fit to the male Masculinity-Femininity correlations.
Additive genetic effects are substantial for each factor, with
823
variance estimates ranging from 22% to 85%. These are somewhat higher than those reported in Table 6, but it must be remembered that error variance has now been explicitly modeled. Others have suggested that genetic effects may account for twothirds of reliable personality variance (Bouchard, 1994), and
these results are generally consistent with that position. Somewhat surprisingly, dominance emerged as a significant source
of variance for Psychoticism and Masculinity-Femininity, although here at least two caveats must be mentioned: For neither
scale does the model fit particularly well, and dominance within
our data is confounded with the possibility that biological siblings share more similar environments than do adoptive siblings
or parents and their children. The shared environment parameter
is significant for three scales: Extraversion, Religious Orthodoxy, and Inadequacy. For the first and last of these, the amount
of variance accounted for is small—roughly 10%. For Religious
Orthodoxy, though, the shared environmental variance is nearly
50%. Letting the shared environment parameter differ by relationship (siblings versus adoptive parents and their children)
did not result in significantly better fit for any of these three
variables. For no scale did the emergenesis parameter emerge
as significant, although again it should be noted that our power
to discriminate these effects from those of dominance or additive
genetic effects was undoubtedly low.
Finally, for Religious Orthodoxy, on which adoptive parents
were correlated substantially, we conducted a rough assessment
of assortative mating's importance by setting the genetic correlation between DZ twins, biological nontwin siblings, and parents
and their biological children to values higher than .50. Setting
these correlations to .55 and .60 resulted in only minor decreases
in chi-square, .274 and 1.114 respectively, and parameter estimates were left basically unchanged. We concluded that it was
probably not necessary to account for the effects of assortative
mating with more elaborate models.
Discussion
In his twin study, Rose (1988) found significant shared environmental variance on four of the nine MMPI factor scales:
Masculinity-Femininity, Extraversion, Religious Orthodoxy,
and Cynicism. Using data from the Texas Adoption Project,
we were able to unambiguously replicate Rose's findings for
Religious Orthodoxy, with a hint of a slight amount of shared
environmental influence on Extraversion. Masculinity—Femininity, after adjustment for age and sex, demonstrated no shared
environmental influence, but due to the large amount of variance
accounted for by sex, it is questionable how meaningful this
result is. For this trait, within-sex correlations would probably
be most appropriate; unfortunately, our adoption sample was
too small to allow this. Finally, there was no evidence in the
adoption data of shared environmental influence on Intellectual
Interests. Differences between our findings and Rose's may be
due to real population differences, slight differences in the scales
used, or sampling error.
A more powerful method of examining the data from both
samples is to fit models to the twin and adoption data simultaneously. Our full model-fitting results suggested substantial genetic
1
We thank an anonymous reviewer for pointing this out.
824
BEER, ARNOLD, AND LOEHLIN
Table 8
Fit Indexes and Parameter Estimates for Models Fit Jointly to Twin
and Adoption Data, AH Model Parameters Included
Parameter estimates
Factor scale
x2a
Neuroticism
Psychoticism
Masculinity -Femininity
Males only
Females only
Extraversion
Religious Orthodoxy
Somatic Complaints
Inadequacy
Cynicism
Intellectual Interests
3.22
14.08
15.35
12.ll b
2.44C
4.96
13.23
8.51
8.63
10.02
18.59
df
X2ldf
A
C
D
I
E
5
5
5
2
3
5
5
5
5
5
0.64
2.82
3.07
6.06
0.81
0.99
2.65
1.70
1.73
2.00
3.72
.542
.517
.376
.000
.617
.665
.641
.491
.562
.740
.863
.265
.218
.239
.396
.197
.345
.705
.097
.350
.000
.000
.234
,702
.682
.619
,559
.572
.000
.488
.000
.274
.349
.147
.000
.000
.000
.000
.000
.000
.000
.445
,000
.000
.749
.438
.578
.677
.516
.332
.288
.715
.603
.614
.355
5
Note. A = additive genetic effects; C = common environment; D = dominance; I = epistasis; E =
unshared environment.
a
N = 1,200. bN= 341. <=W = 436.
ableness, and Conscientiousness (Bergeman et a l , 1993; Loehlin, 1992; Rose, 1995). Costa et al. (1986) reported correlations
between the nine MMPI factors and the five factors of Costa
and McCrae's NEO Personality Inventory (NEO-PI; Costa &
McCrae, 1985), as well as the empirical relations between the
MMPI factor scales and the NEO-PI facets of the neuroticism,
extraversion, and openness dimensions. Their findings revealed
important within-dimension differences in how the MMPI factor
scales relate to the Big Five, which may be of some value in
terms of clarifying the incongruous pronouncements of behavior
geneticists concerning the role of shared environment in personality development.
Typically, twin studies of the Big Five personality dimensions
indicate moderate heritabilities, generally in the range of .3 to
.6, and minimal influence of the shared environment, generally
in the range of .00 to .10 (Angleitner et al., 1995; Bouchard &
influence on every trait. The reduced models for each scale
indicate total genetic variance estimates that range from 4 1 %
(Religious Orthodoxy) to 85% (Intellectual Interests). In this
respect, the results only substantiate what dozens of other studies have shown, namely that the genes are a significant source
of individual differences in personality, attitudes, and interests
(Bouchard, 1994; Loehlin, 1992; Plomin & Daniels, 1987).
However, shared environmental influence was also significant
for three scales, accounting for about 10% of the variance in
Extraversion and Inadequacy, and around 50% of the variance
in Religious Orthodoxy.
Because interest in the five-factor model of personality continues to grow (Digman, 1990; Goldberg, 1990; Wiggins & Pincus,
1992), it is worth considering our results within that taxonomy,
especially because there has been relatively little behavior genetic research examining three of the factors, Openness, Agree-
Table 9
Fit Indexes and Parameter Estimates for Models Fit Jointly to Twin
and Adoption Data, Reduced Models
Parameter estimates
Factor scale
x22
df
Neuroticism
Psychoticism
Masculinity-Femininity
Males only
Females only
Extraversion
Religious Orthodoxy
Somatic Complaints
Inadequacy
Cynicism
Intellectual Interests
5.60
15.15
19.42
16.83h
4.71C
9.38
13.23
11,18
11.60
10.35
18.99
8
7
8
5
6
7
7
8
7
8
8
0.70
2.16
2.43
3.37
0.78
1.34
1.89
1.40
1.66
1.29
2,37
A
C
D
I
.658
.580
.469
.710
.844
.864
.641
.667
.706
.781
.924
—
—
—
—
.318
.705
—_
.314
—
—
.686
.667
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
—
E
.754
.437
.577
.704
.540
.395
.288
.745
.636
.625
.372
Note. A = additive genetic effects; C = common environment; D = dominance; I = epistasis; E = unshared
environment. Missing value indicates that dropping the parameter from the model did not significantly worsen
the fit.
*N = 1,200. hN = 341. CN = 436.
GENETIC AND ENVIRONMENTAL VARIANCE IN THE MMPI
McGue, 1990; Goldsmith et al., 1994). But they also include
notable exceptions, such as Goldsmith et al.'s (1994) heritability
estimate of .02 and shared environmental estimate of .32 for
Intellect/Sophistication, their analogue of Openness, and the
Bergeman et al. (1993) finding that the shared environment
accounted for 21% of the variance in Agreeableness. Reconciling these discrepant findings has proved difficult, but given the
fact that researchers have used different measures of the five
factors, and that these factors are multidimensional, we might
expect to find that, if the various facets within dimensions are
found to be differentially heritable, the heritability and shared
environmental estimates of the Big Five dimensions will vary
between studies.
In the present study, the Religious Orthodoxy factor appears
to exemplify the potential influence of particular personality
facets on behavior genetic parameter estimates. The NEO-PI
dimension of Openness is comprised of the six facets of Fantasy,
Aesthetics, Feelings, Actions, Ideas, and Values. Costa et al.
(1986) showed that Religious Orthodoxy was correlated - . 0 6 ,
.08, .05, - . 0 3 , and .04 with the first five of these facets. However,
it correlated —.56 with the sixth facet, Values (Costa et al.,
1986). Additionally, Religious Orthodoxy correlated .29 with
NEO-PI Agreeableness, specific facets of which had not yet
been enumerated. It is not implausible that the inclusion or
omission of items measuring openness to values or particular
facets of Agreeableness, or both, might thus substantially affect
the degree to which these dimensions are estimated to be influenced by the shared environment.
Besides Religious Orthodoxy, the MMPI scales of Inadequacy
and Extraversion also showed some shared environmental influence. The shared environment accounted for about 10% of
the variance on these scales, and this is fairly consistent with
previous behavior genetic findings. Costa et al. (1986) reported
that Inadequacy is correlated moderately with all facets of Neuroticism, but is most highly correlated (negatively) with Assertiveness, a facet of Extraversion. The MMPI Extraversion scale
is moderately correlated with all NEO-PI Extraversion facets
except Activity. The other six MMPI factor scales, in which
variation is primarily genetically influenced, are more diffuse
in their relations to the NEO-PI, with the NEO-PI dimension of
Neuroticism best represented. The MMPI scales of Neuroticism,
Psychoticism, Somatic Complaints, and Masculinity-Femininity all correlate moderately with the six Neuroticism facets.
Intellectual Interests correlates moderately with Openness facets, while Cynicism seems to reflect mainly the lack of Agreeableness. As with the results for Religious Orthodoxy, this latter
result is of some interest because it also indicates that there
may be differences in genetic and environmental influences on
Agreeableness at the facet level; those correlated with Cynicism
may be relatively free of shared environmental effects. (But see
a recent twin study by Jang, Livesley, and Vernon, 1996, which
shows more cross-facet consistency in shared environmental
effects than does our analysis.)
Aside from their implications for genetic-environmental
analyses of the Big Five personality dimensions, the importance
of our results lies mostly in the substantial effect of the shared
environment on variation in Religious Orthodoxy. Both the
adoption data alone and the joint twin-adoption model indicate
that the family environment is a major contributor to individual
differences in religiosity. Moreover, phenotypic assortment—
825
which spuriously inflates shared environmental variance estimates—cannot account for this finding. The adoption data show
that among biologically unrelated relatives, the shared environment accounts for a substantial amount of variation; although
the spousal correlations indicate that assortative mating for religiosity has undoubtedly occurred, in these correlations it is
simply an irrelevant consideration. In addition, setting the genetic correlations between biological relatives to values higher
than .50 in our model-fitting analyses did not significantly improve the fit of the model to the Religious Orthodoxy data or
change its parameter estimates, suggesting that more elaborate
models explicitly accounting for assortative mating were unlikely to lead to a different conclusion.
Although religious affiliation has been found to be influenced
by the family environment (Eaves et al., 1990), this finding
seems to contradict reports (Martin et al., 1986; Waller et al.,
1990) and reviews (Eaves et al., 1989; Rowe, 1994) suggesting
that moral, religious, and social attitudes and beliefs are little
influenced by the shared environment once individuals reach
adulthood (for analyses suggesting that both genetic and shared
environmental factors may be important for social attitudes, see
Loehlin, 1993; Truett et al., 1992). It should be noted that most
of the adoptees in our sample had not yet reached adulthood by
the time they took the MMPI (their mean age was just under
18), and previous work suggests that the influence of the family
environment may decline as children leave the home (Plomin &
Daniels, 1987). Nonetheless, the correlations of the adoptee's
age with absolute pair differences between the adoptee and his
or her adoptive father and mother are only .044 and -.002, not
indicative of a trend toward a decreasing influence of the family
environment as children age.
Despite the finding of substantial shared environmental effects, our data also demonstrate genetic effects on religiosity.
Although once regarded as a trait on which variation was not
influenced by the genes (Plomin, 1989), the present result lends
credence to earlier twin study findings of genetic influence (Martin et al., 1986; Waller et al., 1990). To be sure, the scales used
in the various studies have been different. The Martin et al.
study used a measure designed to assess social attitudes, not
religiosity per se, although it included many items of a religious
character. In addition, their model, which assumed phenotypic
assortment, may be questionable. Reynolds, Baker, and Pederson
(1996) have recently pointed out that if assortment is assumed
to occur by phenotype when it is in fact due to environmental
similarities or propinquity, the influence of the shared environment is likely to be underestimated. The study by Waller et al.
(1990) included five different measures of religiosity, including
one, Wiggins's Religious Fundamentalism Scale, that has six
items in common with our Religious Orthodoxy scale, since
they are both derived from the MMPI item pool. Also relevant
are results from an earlier twin study by Pogue-Geile and Rose
(1985) demonstrating, on two separate occasions in the same
sample, what appeared to be substantial shared environmental
effects on the Wiggins scale.
It is also important to remember that our sample of adoptive
families, because selected by way of a particular church-related
home for unwed mothers, is not representative of the general
U.S. population (Horn et al., 1979). Because this may have
acted to decrease the variance of religiosity in our data compared with that in the population, it could be argued that shared
826
BEER, ARNOLD, AND LOEHLIN
environmental effects on religiosity in the general population
may be even larger than those reported here. In contrast, if one
argues that parents are most likely to influence their children
on the traits that are most important to them (Buss, 1987;
McGue, 1994), it is possible that among less religiously oriented families, shared environmental effects are smaller because
there is little trait-relevant between-family variance to begin
with.
Although it is difficult to resolve entirely the discrepancies
between the Waller et al. (1990) and Pogue-Geile and Rose
(1985) studies and ours, it is possible that the Religious Orthodoxy scale used here taps more than one component of religiosity. It seems plausible to us that it reflects both a personality
component of traditionalism, which may be more genetically
influenced, and an orthodox Christian belief or behavior component, which may be more influenced by one's rearing environment. Whether this distinction is defensible or not must await
future analyses of this and related scales. In any case, behavior
genetic research should attempt to more precisely delineate the
components of moral and religious attitudes, beliefs, and behaviors and their sources of variation. If future investigations, using
adult samples, replicate the substantial shared environmental
effects on religiosity found here, then strong evidence will have
been adduced to counter the emerging belief that the family
environment does not create enduring similarity in attitudes
among those who share it."
References
Angleitner, A., Riemann, R., & Strelau, J. (1995, July). A study of twins
using the self-report and peer-report NEO-FF1 scales. Paper presented
at the eighth meeting of the International Society for the Study of
Individual Differences, Warsaw, Poland.
Beer, J. M , & Rose, R. J. (1995). The five factors of personality: A
study of twins [Abstract]. Behavior Genetics, 25, 254.
Bergeman, C. S., Chipuer, H. M , Plomin, R., Pederson, N. L., McClearn,
G. E., Nesselroade, J. R., Costa, P. T, Jr., & McCrae, R. R. (1993).
Genetic and environmental effects on openness to experience, agreeableness, and conscientiousness: An adoption/twin study. Journal of
Personality, 61, 159-179.
Betsworth, D. G., Bouchard, X J., Jr., Cooper, C. R., Grotevant, H. D.,
Hansen, J. C , Scan, S., & Weinberg, R. A. (1994). Genetic and environmental influences on vocational interests using adoptive and biological families and twins reared apart and together. Journal of Vocational Behavior, 44, 263-278.
Bouchard, T. J., Jr. (1994). Genes, environment, and personality. Science, 264, 1700-1701.
Bouchard, T. J., Lykken, D. T, McGue, M , Segal, N. L., & Tellegen, A.
(1990). Sources of human psychological differences: The Minnesota
Study of Twins Reared Apart, Science, 250, 223-228.
Bouchard, T J., Jr., & McGue, M. (1990). Genetic and environmental
influences on adult personality: An analysis of adopted twins reared
apart. Journal of Personality, 58, 263-292,
Buss, D. M. (1984). Marital assortment for personality dispositions:
Assessment with three different data sources. Behavior Genetics, 14,
111-123.
Buss, D. M. (1987). Evolutionary hypotheses and behavioral genetic
methods: Hopes for a union of two disparate disciplines. Behavioral
and Brain Sciences, 10, 20.
Costa, P. T., Jr., Busch, C. M., Zonderman, A. B., & McCrae, R. R.
(1986). Correlations of MMPI factor scales with measures of the five
factor model of personality. Journal of Personality Assessment, 50,
640-650.
Costa, P. T., Jr., & McCrae, R. R. (1985). The NEO Personality Inventory manual. Odessa, FL: Psychological Assessment Resources.
Costa, P. T., Jr., & McCrae, R. R. (1987). On the need for longitudinal
evidence and multiple measures in behavioral-genetic studies of adult
personality. Behavioral and Brain Sciences, 10, 22-23.
Costa, P. T., Jr., Zonderman, A. B., McCrae, R. R., & Williams, R. B.,
Jr. (1985). Content and comprehensiveness in the MMPI: An item
factor analysis in a normal adult sample. Journal of Personality and
Social Psychology, 48, 925-933.
Digman, J. M. (1990). Personality structure: Emergence of the fivefactor model. Annual Review of Psychology, 41, 417-440.
Eaves, L. J., Eysenck, H.J., & Martin, N. G. (1989). Genes, culture
and personality. San Diego, CA: Academic Press.
Eaves, L. J., Martin, N. G., & Heath, A. C. (1990). Religious affiliation
in twins and their parents: Testing a model of cultural inheritance.
Behavior Genetics, 20, 1-22.
Feng, D., & Baker, L. (1994).Spouse similarity in attitudes, personality,
and psychological well-being. Behavior Genetics, 24, 357-364.
Goldberg, L.R. (1990). An alternative "description of personality":
The Big-Five factor structure. Journal of Personality and Social Psychology, 59, 1216-1229.
Goldsmith, H. H., Losoya, S. H., Bradshaw, D. L., & Campos, J. J.
(1994). Genetics of personality: A twin study of the five-factor model
and parent-offspring analyses. In C. F. Halverson, Jr., G. A. Kohnstamm, & R. P. Martin (Eds.), The developing structure of temperament and personality from infancy to adulthood (pp. 241-265). Hillsdale, NJ: Erlbaum.
Horn, J. M., Loehlin, J. C , & Willerman, L. (1979). Intellectual resemblance among adoptive and biological relatives: The Texas Adoption
Project. Behavior Genetics, 9, 177-207.
Jang, K. L., Livesley, W. J., & Vemon, P. A. (1996). Heritability of the
big five personality dimensions and their facets: A twin study. Journal
of Personality, 64, 577-591.
Loehlin, J. C. (1987). Latent variable models: An introduction to factor,
path, and structural analysis. Hillsdale, NJ: Erlbaum.
Loehlin, J. C. (1992). Genes and environment in personality development. Newbury Park, CA: Sage.
Loehlin, J. C. (1993). Nature, nurture, and conservatism in the Australian Twin Study. Behavior Genetics, 23, 287-290.
Loehlin, J. C , & Gough, H. G. (1990). Genetic and environmental variation on the California Psychological Inventory vector scales. Journal
of Personality Assessment, 54, 463-468.
Loehlin, J. C , Horn, J. M., & Willerman, L. (1981). Personality resemblance in adoptive families. Behavior Genetics, 11, 309-330.
Loehlin, J. C , & Nichols, R. C. (1976). Heredity, environment, and
personality. Austin, TX: University of Texas Press.
Loehlin, J. C , Willerman, L., & Horn, J. M. (1987). Personality resemblance in adoptive families: A 10-year follow-up. Journal of Personality and Social Psychology, 53, 961-969.
Lykken,D. T. (1982). Research with twins: The concept of emergenesis.
Psychophysiology, 19, 361-373.
Lykken, D. T., McGue, M., Tellegen, A., & Bouchard, T. J., Jr. (1992).
Emergenesis: Genetic traits that may not run in families. American
Psychologist, 47, 1565-1577.
Martin, N. G., Eaves, L. J., Heath, A. C , Jardine, R., Feingoid, L. M., &
Eysenck, H. J. (1986). Transmission of social attitudes. Proceedings
of the National Academy of Sciences, USA, 83, 4364-4368.
Mascie-Taylor, C. G. (1989). Spouse similarity for IQ and personality
and convergence. Behavior Genetics, 19, 223-227.
McGue, M. (1994). Why developmental psychology should find room
for behavioral genetics. In C. A. Nelson (Ed.), Threats to optimal
development: Integrating biological, psychological, and social risk
factors. The Minnesota symposia on child psychology. Vol. 27 (pp.
105-119). Hillsdale, NJ: Erlbaum.
Neale, M. C. (1994). Mx: Statistical modeling (2nded.). Box710MCV,
Richmond, VA 23298: Department of Psychiatry.
GENETIC AND ENVIRONMENTAL VARIANCE IN THE MMPI
Neale, M. C , & Cardon, L. R. (1992). Methodology for genetic studies
of twins and families. Dordrecht, The Netherlands: Kluwer.
Pederson, N. L., Plomin, R., McClearn, G. E., & Friberg, L. (1988).
Neuroticism, extraversion, and related traits in adult twins reared apart
and reared together. Journal of Personality and Social Psychology,
55, 960-967.
Plomin, R. (1989). Environment and genes: Determinants of behavior.
American Psychologist, 44, 105-111.
Plomin, R., & Daniels, D. (1987). Why are children in the same family
so different from each other? Behavioral and Brain Sciences, 10, 1 16.
Pogue-Geile, M. F , & Rose, R. J. (1985). Developmental genetic studies
of adult personality. Developmental Psychology, 21, 547-557.
Popham, S. M., & Holden, R. R. (1991). Psychometric properties of
MMPI factor scales. Personality and Individual Differences, 12, 5 1 3 517.
Reynolds, C. A., Baker, L. A., & Pederson, N.L. (1996). Models of
spouse similarity: Applications to fluid ability measured in twins and
their spouses. Behavior Genetics, 26, 73-88.
Rose, R. J. (1988). Genetic and environmental variance in content dimensions of the MMPI. Journal of Personality and Social Psychology,
55, 302-311.
Rose, R. J. (1995). Genes and human behavior. Annual Review of Psychology, 46, 625-654.
827
Rowe, D. C. (1994). The limits of family influence. New "ibrk: Guilford
Press.
Rowe, D. C , & Plomin, R. (1981). The importance of nonshared ( E 0
environmental influence in behavioral development. Developmental
Psychology, 17, 517-531.
Saudino, K. J., McGuire, S., Reiss, D., Hetherington, E. M., & Plomin,
R. (1995). Parent ratings of EAS temperaments in twins, full siblings,
half siblings, & step siblings. Journal of Personality and Social Psychology, 68, 723-733.
Scan, S. (1992). Developmental theories for the 1990s: Development
and individual differences. Child Development, 63, 1-19.
Tellegen, A., Lykken, D. T., Bouchard, T. J., Jr., Wilcox, K. J., Segal,
N. L., & Rich, S. (1988). Personality similarity in twins reared apart
and together. Journal of Personality and Social Psychology, 54, 10311039.
Truett, K. R., Eaves, L. J., Meyer, J. M., Heath, A. C , & Martin, N. G.
(1992). Religion and education as mediators of attitudes: A multivariate analysis. Behavior Genetics, 22, 43-62.
Waller, N. G., Kojetin, B. A., Bouchard, T. J., Jr., Lykken, D. T, & Tellegen, A. (1990). Genetic and environmental influences on religious
interests, attitudes, and values: A study of twins reared apart and
together. Psychological Science, 1, 138-142.
Wiggins, J. S., & Pincus, A. L. (1992). Personality: Structure and assessment. Annual Review of Psychology, 43, 473-504.
Appendix
Items Included on Short Form of MMPI Factor Scales
Factor scales (short form)
Neuroticism
Psychoticism
Masculinity -Femininity
Extraversion
Religious Orthodoxy
Somatic Complaints
Inadequacy
Cynicism
Intellectual Interests
Note.
Items
13, 15, 21, 39, 64, 67, 79 (R), 94, 100, 102, 129, 134, 138, 142, 145, 148,
152 (R), 162, 170 (R), 217, 226, 234, 236, 238, 242 (R), 266, 278, 282,
299, 301, 303, 322, 336, 337, 338, 340, 359, 361, 362, 366, 383, 397
16, 17 (R), 22, 24, 27, 35, 37 (R), 40, 48, 49, 50, 52, 53, 54 (R), 60 (R),
65 (R), 69, 75 (R), 76, 83 (R), 85, 88 (R), 97, 104, 106, 110, 113 (R),
121, 122 (R), 123, 133 (R), 137 (R), 139, 146, 150 (R), 151, 154 (R),
156, 157, 160, 168, 177 (R), 178 (R), 182, 184, 194, 196 (R), 197, 200,
202, 205, 209, 210, 211, 212, 216, 218, 220 (R), 224, 227, 245, 246,
247, 252, 256, 257 (R), 258 (R), 269, 275, 276 (R), 285 (R), 286, 291,
293, 306 (R), 309 (R), 312, 324, 339, 341, 344, 345, 347 (R), 349, 350,
351, 352, 354, 355, 358, 360, 363, 364, 365, 398, 502 (R)
1, 4 (R), 7, 70 (R), 74 (R), 77 (R), 81, 87 (R), 92 (R), 118, 132 (R),
149 (R), 176, 203 (R), 215, 219, 223, 261 (R), 283, 300
57, 99, 165, 171 (R), 181, 207, 208, 229, 231, 254, 292 (R)
30 (R), 45 (R), 58, 95, 98, 115, 116 (R), 135 (R), 206, 249, 277 (R)
3 (R), 9 (R), 10, 23, 29, 34, 41, 43, 44, 46 (R), 47, 51 (R), 55 (R), 62, .
63 (R), 68 (R), 72, 103 (R), 108, 114, 119 (R), 125, 130 (R), 153 (R),
161, 163 (R), 174 (R), 175 (R), 186, 189, 190 (R), 192 (R), 193 (R),
230 (R), 243 (R), 251, 272 (R), 273, 330 (R)
82, 86, 147, 159, 172, 180, 201, 259, 260, 264 (R), 267, 304, 321, 342,
343, 356, 357, 374
59, 71, 84, 89, 93, 109, 117, 124, 136, 244, 250, 265, 271, 280, 284, 298,
313, 316, 319,327,348,406
78, 126, 164, 173, 204, 221, 295
Items that are reverse-scored are denoted by (R).
Received November 7, 1996
Revision received April 28, 1997
Accepted April 28, 1997