MATE SELECTION
57
children (Plomin, DeFries, & Roberts, 1977). It is a curious fact
that by far the strongest and best replicated psychological simi-
larity between spouses is to be found in the cluster of traits that
includes religiosity (Waller, Kojetin, Bouchard, Lykken, & Tel-
legen, 1990), conservatism (Martin et al., 1986), and authoritar-
ianism (Altemeyer, personal communication, March 13,199 1);
spousal correlations on these correlated dimensions range from
.40 to nearly .70. We have replicated and extended some of this
research on assortative mating in our studies of Minnesota-
born, middle-aged twins and their families.
Method
The Minnesota Twin Registry (Lykken, Bouchard, McGue, & Telle-
gen, 1990) has collected biographic and inventory data from a large,
birth-record-based sample of middle-aged twins born from 1936
through 1955 and from many of their spouses and parents, a sample
that is demographically representative of the general population of the
north central United States. For purposes unrelated to the present
problem, five inventories, totaling 623 items concerned with personal-
ity, self-rated talents, occupational and recreational interests, and atti-
tudes on 14 controversial topics, were mailed to these twins and their
spouses. The return rate was 74% overall; completed inventories were
returned by both members of 901 pairs of married twins, by 1,052 of
their spouses, and by the spouses of both twins in 269 of these pairs.
The battery was also completed by 133 pairs of the parents of the
younger participant twins. The twins ranged in age from 29 to 50 years
=M = 37.9 years, SD = 7.8), the spouses from 22 to 68 years (Λ∕ = 38.0
years, SD = 8.2), and the parents from 45 to 85 years (Λ∕= 62.3 years,
SD= 5.4).
All participantscompleted the300-item Multidimensional Personal-
ity Questionnaire (MPQ; Tellegen, 1982; Tellegen et al., 1988; Tellegen
& Waller, in press), which yields 11 first-order personality dimensions
and the 3 higher order or superfactors of Positive and Negative Emo-
tionality and Constraint. The Minnesota Vocational Interests Test
consists of 100 items written to tap the broader factors ofoccupational
interest that are commonly identified. The Minnesota Leisure-Time
Interest Test consists of 120 items describing a wide variety of leisure
time activities such as “going fishing,” “nightlife: bars, nightclubs, dis-
cos, etc.,” “volunteer work,” “taking a college course in some subject of
interest,” “going on a camera safari in Africa, Borneo, the desert, or the
Amazon basin,” or “getting involved in controversial issues.” Most of
the itemsarestated rather generally, for example, “risky pastimes: hang
gliding, mountain climbing, surfing, etc.” The Minnesota Talent Sur-
vey is a 40-item survey of familiar talents. The four alternatives are
worded so that 1 represents outstanding or professional-level talent, 3 is
labeled average, and 4 is below average. The 40 items range widely (e.g.,
“carpentry,” ‘cooking,” “singing,” “athletics,” “buying and selling,”
“public speaking,” “physical strength,” “judgment,” “good host,” “writ-
ing ability,” and “cool headed”).
The Self-Rating Inventory consisted of 3 1 items assessing qualities
not directly addressed by the other items in the test battery. These
items asked about abstract intelligence, creativity, mental energy, physi-
cal energy, morality, empathy, irritability, self-discipline, affability,
nurturance, emotional control, and team spirit-attributes that also
have potential relevance to marital satisfaction. The 14 attitude items
were concerned with controversial topics such as welfare, gun control,
defense spending, religious belief, capital punishment, protection of
the environment, and the proper role of government. These invento-
ries are described more fully in Lykken, Bouchard, McGue, and Telle-
gen (in press).
In addition to the 14 attitude itemsand the 14 scales of the MPQ, the
inventory items generated 18 talent and self-rating factors (e.g., good
looks, mental ability, and probity), 19 recreational interest factors (e.g.,
intellectual pursuits, gambling, hunting, and religious activities), and
18 occupational interest factors (e.g., blue collar, personal service,
farming or ranching, and selling or trading; see Waller, Lykken, &
Tellegen, in press). The alpha reliabilities, based on 1,728 men and
2,284 women, are shown in Table 1 together with the intraclass correla-
tions on these variables of 5 12 pairs of adult MZ twins, included as an
indication of the degree of similarity on these traits that can be ex-
pected in pairsofvery similar people. Adding height, weight, ponderal
index, foot length, and years of education made up a total of 88 vari-
ables, all of which were corrected for quadratic regression on age
(McGue & Bouchard, 1984), separately by sex, before correlations were
computed. Although we did not choose this set of variables for the
purpose of studying spousal similarity and although some dimensions
such as physical attractiveness that would be of interest here were not
assessed, they do constitute a broad and varied collection of measures,
many of which would be of interest to prospective mates.
The wide age range of the sample made it possible to investigate
whether spousal similarity increases with years married; are the ob-
served similarities due to assimilation or are they already in place when
couples decide to marry? In this sample of 1,052 twin-spouse couples
and 133 parental couples, absolute within-pair differences on all 88
variables were correlated with years married, to determine, for each
variable, whether spousal similarity increases with cohabitation. Hus-
band-wife correlations within the 1, 185 spousal pairs then were com-
puted for each of these 88 variables.
Results
The correlations between absolute within-pair differences
and years married ranged from —.09 to . 12; the mean correla-
tion was —.006. The most plausible inference is that these cou-
ples were about as similar on these variables when they married
as they were when assessed. This result corroborates previous
reports (Buss, 1984; Caspi et al., 1992).
The 10 variables yielding the largest spousal correlations are
listed in Table 2. Seven of these 10 variables comprise a higher
order factor that can be labeled Traditional Values. People high
on this factor enjoy church activities and have high scores on the
MPQ’s Traditionalism scale; they oppose abortion, favor a
strong national defense, and are politically conservative; they
do not enjoy gambling, visiting nightclubs, or flirting. This is
plainly the dimension of religiosity-conservatism previously
noted as being second only to age in strength of spousal resem-
blance. A second dimension is educational attainment. Finally,
the spousal pairs correlated .34 to .36 in their interests in hunt-
ing, fishing, camping, canoeing, hiking, and so on. Every one
of the 88 spousal correlations was positive, although 47 were
less than .20 and only the 10 listed in Table 2 were greater
than .30.
In the lower half of Table 2, we can see that spouses are
slightly correlated (.28) in (self-reported) height, less so than
opposite-sex dizygotic (DZ) twins (the value shown in the table,
.43, is based on 893 twin pairs). Spousal correlations in talents,
leisure, and occupational interests are low, averaging. 10 to .25,
but they are about as strong as the DZ twin correlations. The
mean correlation of spouses on the 10 MPQ personality scales
(.08) is lower than the mean for DZ twins (. 18).
Discussion of Study 1
With a new set of variables and a large and demographically
representative sample of middle-aged married couples, we con-