toward a psychology of person-environment (pe) correlation: the role of spouse selection
TRANSCRIPT
Journal of Penonility and Social Psychotojy1984, Vol 47. No 2, 361-377
Copyright 1984 by theAmerican Psychological Asociauon. Inc.
Toward a Psychology of Person-Environment (PE) Correlation:The Role of Spouse Selection
David M. BussHarvard University
Identifying the important ways in which attributes of persons correlate with featuresof the interpersonal environment is perhaps one of the most intriguing issues facingthe development of an interactional psychology. Little empirical progress has yetbeen made, however, in documenting the domains within which person-environment(PE) correspondence occurs, in identifying the mechanisms by which such cor-respondence is brought about, and in exploring the consequences of obtained cor-respondence. One hypothesis of this article is that nonrandom spouse selection(assortative marriage) is one mechanism by which correspondences between personsand their interpersonal environments are created. An empirical study of 93 marriedcouples examined spouse correlations within eight interpersonal categories (e.g.,dominance, submissivness, extraversion) by using self and observer reports of theperformance frequencies of 800 acts. Substantial spouse correspondence was found,particularly for the domains of extraversion, dominance, quarrelsomeness, andingenuousness. In addition, changes in degree of spouse correspondence were as-sociated with length of marital relationship. Discussion focuses on different typesof PE correspondence, implications for the study of adult personality development,and the emergence of a psychology of PE correlation.
Who marries whom is the subject of intenseinterest in groups ranging from tabloid readersto behavior geneticists. Part of the intrigue ofnonrandom marriage, of which assortativemarriage1 is one of the most prevalent ex-amples, lies with the range of implications thattranscends disciplinary boundaries. It is aninteresting sociological process in its own rightbecause nearly all members of society even-tually enter into a marital relationship (Price& Vandenberg, 1980). At a psychological level,moderate levels of assortment have been foundfor certain intellectual, attitude, and person-ality variables (Buss, 1984; Jensen, 1978; Van-denberg, 1972). And at the genetic level, theprocess of assortative mating has consequencesfor changes in the genetic structure in the pop-ulation in subsequent generations.
The author thanks Michael Barnes, Kenneth Craik, Ka-thy McCartney, Dan Ozer, Carolyn Phinney, RobertPlomin, and Jerry Wiggins for helpful comments on anearlier draft of this paper. Special thanks go to MichaelBarnes for lending his expertise at every stage of this project
Requests for reprints should be sent to David M. Buss,Department of Psychology and Social Relations, HarvardUniversity, 33 Kirkland Street, Cambridge, Massachusetts02138.
But the study of assortative marriage hasremained largely within the province of be-havior geneticists, perhaps because the processhas such profound genetic consequences. First,heritability coefficients tend to be inflated byassortment (Plomin, DeFries, & Roberts,1977), and thus must be considered in manystudies attempting to derive accurate estimatesof genetic influence. Second, assortative matingfor heritable characteristics tends to increasethe dispersion of those characteristics in sub-sequent generations (Jensen, 1978; Vanden-berg, 1972). Thus, greater frequencies of in-dividuals are found at the tails of the distri-bution as assortative mating makes subsequentdistributions more platykurtic. A third con-sequence is that traits that are initially un-correlated in the population (e.g., dominanceand conscientiousness) may become correlatedthrough the process of assortative mating. Andfourth, the process influences the correlationsamong most biological relatives (e.g., betweenparents and children) on those heritable char-
1 Assortative mating may be defined as the nonrandomcoupling of individuals based on resemblance on one ormore characteristics.
361
362 DAVID M. BUSS
acteristics for which assortment occurs. Forthese reasons, assortment is critical to behav-ioral genetic analysis.
The importance of assortative marriage forpersonality psychology, however, has not beenelaborated as fully. This article will examineseveral ways in which spouse correspondencein the interpersonal domain and processescontingent on that correspondence can addresstwo key issues in personality psychology: (a)specifying the ways in which individuals createtheir own environments, and (b) identifyingpotent variables that affect adult personalitydevelopment. A conceptual framework withinwhich these processes might operate is pre-sented, followed by an empirical study of mar-ried couples.
An important conceptual advance in un-derstanding the different ways in which cor-respondence between dispositions and envi-ronments can occur pertains to the family ofconcepts subsumed by genotype-environment(GE) correlation (Plomin, DeFries, & Loehlin,1977; see also Scarr & McCartney, 1983). Ac-tive genotype-environment correlation entailscreating or selecting environments that cor-respond to (and perhaps reinforce) initial pre-dispositions. Highly gregarious individuals, forexample, may seek social stimulation bythrowing parties, calling friends, and initiatingconversations with strangers. Those high onintelligence may attend lectures, read books,and initiate academic conversations, therebycreating environments that correspond to andreinforce initial propensities.
Active GE correlation may be contrastedwith evocative (or reactive) GE correlation inwhich differential responses from others areevoked depending upon differences in a givenattribute. For example, parents and peers tendto react to highly active children with behaviorsdesigned to reduce the noise and intensity oftheir behavior, resulting sometimes in strifeand competitiveness, whereas interactions withless active children are generally more har-monious (Buss, 1981a; Buss, Block, & Block,1980). Similarly, society may like to cut downindividuals who are highly dominant "to helpthe humble inherit the earth" (Cattell, 1973,p. 145).
Both active and evocative (or reactive) GEcorrelations may be contrasted with passiveGE correlation, in which parents provide both
genes and environments that happen to becorrelated, independent of any response fromor reaction to the child. Parents who are ge-netically gifted in verbal abilities, for example,in addition to passing on genes favorable toverbal skills, may provide environmental con-texts (e.g., books, verbal exchange) that cor-relate with and facilitate initial abilities.
Genotype-environment correlation, ratherthan GE interaction (as in nonlinear statisticalcombinations of effects in an analysis of vari-ance (ANOVA) framework) more closely resem-bles the concept of "interactionism" in thedynamic sense of interchange between personsand environments (Plomin et al., 1977).
In spite of the potential importance of thisconcept, little empirical progress has yet beenmade in documenting the domains of person-environment (PE) correspondence, nor inidentifying the mechanisms by which suchcorrespondence is reached (but see Buss,1981a; Snyder, 1981; Snyder & Gangestad,1982). One thesis of this article is that theprocess of assortative marriage provides oneimportant avenue through which PE corre-lation occurs. The social environment is de-fined largely by significant others who occupyand behave within it. And because the spouseis perhaps one of the most important personsoccupying that environment, selection of apartner based on one's personal characteristicsbecomes an important mechanism for estab-lishing correspondence between person andenvironment As such, the study of the basisof spouse selection can provide clues to themanner in which persons actively create theirown environments.
Identifying important correspondences be-tween attributes of persons and features of theirinterpersonal environment may have conse-quences for adult personality development. Todate, we have not identified any large envi-ronmental variables that account for muchvariance in personality traits (Loehlin & Ni-chols, 1976; Rowe & Plomin, 1981). By puttingsimilar individuals into close proximity, as-sortative marriage may be an important areawithin which the key environmental variablesthat affect adult development may be identi-fied.
What mechanisms affecting adult person-ality development might be contingent onspouse similarity? First, initial similarity might
PERSON-ENVIRONMENT CORRELATION 363
create a marital environment that correspondsto and reinforces initial tendencies. Spousesimilarity in extraversion, for example, mayresult in mutually reinforcing tendencies toselect some environments (e.g., parties, night-clubs) rather than others (e.g., quiet eveningat home). Shared environments create sharedexperiences, and deviation from the maritalmean may be resisted. The crystallization ofexisting tendencies resulting from exposure tocommon and mutually selected environmentsis one developmental process that might bebased on initial spouse similarity.
A second process is suggested by the de-velopmental concept of "pacers" (e.g., Loe-vinger, 1976). Effective developmental pacersare generally those that provide a model thatis only moderately discrepant from the targetindividual. For example, moral developmentmay be accelerated by interaction with some-one who is advanced by a single stage withina given stage sequence but less so by one whois two or more stages advanced. Assortativemarriage for certain characteristics (e.g., moralstage, maturity, social skills) may result in morecases of moderate discrepancy between spousesthan would occur if mating were random.Thus, spouses may serve as pacers for eachother, creating environments that affect adultpersonality development—a process estab-lished by the initial marital assortment.
In sum, the process of assortative marriage,although previously studied primarily withinthe contexts of behavioral genetics and soci-ology, could have important linkages to per-sonality theory and research (a) by specifyingone way in which persons create their ownenvironments (PE correlation) and (b) byidentifying variables that might affect adultpersonality development (e.g., processes of so-lidification and pacing).
The first step in exploring these processesis to document the domains within which cor-respondence between attributes of persons andfeatures of their interpersonal environmentmay be found. One potentially fruitful areafor examining PE correlation is with marriedcouples. The correspondence between an at-tribute or behavior of the husband and an at-tribute or behavior of his wife (and vice-versa)identifies one potentially important form ofPE correlation. But even though studies of as-sortative marriage with respect to intellectual
and cognitive abilities have been numerous(see Jensen, 1978; Johnson, Ahern, & Cole,1980; Watkins & Meredith, 1981; Zonderman,Vandenberg, Spuhler, & Fain, 1977), studieson spouse correlation with respect to person-ality and interpersonal behavior have beenfewer in number. As one step in bringing theprocess of assortative marriage into the domainof personality and interactional psychology, Iconducted the present study to assess, in arelatively comprehensive way, the interpersonaldomains within which spouse correlation oc-curs.
The Wiggins (1979) circumplex model ofthe interpersonal domain was chosen as a rel-atively comprehensive system from which tosample systematically from the interpersonaldomain. For the purposes of this study, I se-lected every other point on the circumplex,resulting in eight categories: dominance, ex-traverted, agreeable, ingenuous, submissive,introverted, quarrelsome, and calculating. Toobtain a relatively comprehensive assessmentof each of these eight interpersonal disposi-tions, 100 acts, independently nominated asbelonging to each disposition, were employed(see Method section).
Previous studies of spouse correlation haveused standard self-report scales. In the presentstudy, self-reports of performance with respectto each of the 800 acts (100 within each ofeight categories) were supplemented by ob-server (spouse) reports of act performance inorder to assess convergence. The purposes ofthe present study may be briefly outlined: (a)to identify the interpersonal domains withinwhich spouse correlations occur, (b) to ex-amine whether obtained spouse correlationsare accounted for by the process of initialmarital assortment, or are due to age, to co-hort, or to convergence over the course of themarriage; and (c) to explore developmentaltrends in spouse correspondence as a functionof marriage length and number of years ofassociation.
The Act Nominations
Method
SubjectsSeveral samples of subjects participated in the act nom-
ination stage. The first consisted of 75 undergraduate stu-dents who provided act nominations for the category of
364 DAVID M. BUSS
dominance (see Buss & Craik, 1980). Thirty-seven un-degraduates provided nominations for submissiveness (seeBuss & Craik, 1981). The remaining act nominations,obtained from independent samples for this study, were(sample size in parentheses) quarrelsome (79), agreeable(83), calculating (76), ingenuous (78), extroverted (78),and introverted (79).
ProcedureEach participant received a sheet with standard instruc-
tions, the basic form of which read: "Think of the threemost dominant [submissive, quarrelsome, agreeable, etc.]females you know. With these females in mind, write downfive acts or behaviors they have performed (or might per-form) that reflect or exemplify their dominance [submis-siveness, quarrelsomeness, etc.]." Five lines were providedupon which act nominations could be written. The in-structions were then repeated, altering the sex of actor tomale (e.g., "Think of the three most dominant males youknow . . .").
Results
The eight lists of acts generated in this waywere subsequently reduced by eliminating re-dundancies, "nonact" statements (e.g., adjec-tives), general tendency statements (e.g., "shetends to exercise a lot"), and statements con-sidered too vague to constitute observable acts.Some act nominations (e.g., general tendencystatements) were converted into act descrip-tions by appropriate rephrasing. The final listswere examined for grammatical errors, whichwere then corrected.
The target number of 100 acts within eachcategory was obtained for each of the eightcategories and then prepared for the subse-quent study. Three forms of each act werecreated: two in the third-person singular (onefor "he . . ." and one for "she . . .") and onein the first-person singular. For example, theact "He introduced himself to the new neigh-bor" became "I introduced myself to the newneighbor" on the first-person singular form ofthe act The 800 acts thus obtained were in-termingled and packaged into Act Reports, de-scribed below.
Main Study: Spouse Correlations
Method
Subjects
One hundred and eight-six individuals composing 93married couples participated in the main study. Subjectswere obtained by placing newspaper advertisements andflyers throughout the larger Boston area. Both the adver-
tisements and the flyers indicated that a study was beingconducted using married couples and that personal feed-back and a small sum of money would be given as a tokenof appreciation for participation.
MaterialsAmong a larger battery of tests and measures were the
following instruments used for the present study:Confidential biographical questionnaire. This ques-
tionnaire asked a variety of questions about physical char-acteristics, demographic characteristics, consumptionhabits, academic achievements, background marital in-formation, and marital satisfaction. Of particular impor-tance for the present study, in order to establish compa-rability to published studies on assortative marriage andspouse correlation, were the variables of age, height, weight,smoking and drinking behavior, previous marriages, num-ber of years known, number of years married, and sat-isfaction with marriage.
General Vocabulary Test. This multiple choice vocab-ulary test (Gough & Sampson, 1974) consists of SO words.Among the four multiple-choice options, subjects selectthe option that is most similar in meaning to the initialword. The vocabulary score is the sum across the SO itemsof the correct answers.
Self-Act Reports The retrospective self-report of actperformance was obtained through two forms with 400acts (intermingled from the eight categories) composingeach form. The instructions were as follows: "The followingpages contain 400 human acts beginning with act (1) toact (400). For each act, please indicate how often you haveperformed it (if at all) within the past three months." Athree month time frame was chosen to allow enough timefor a sufficient number of the acts to have occurred.
Observer-Act Reports. In order to obtain an indepen-dent assessment of act performance, the spouse of eachparticipant completed a parallel Act Report form on whichtheir spouse's performance of each of the 800 acts wasreported. Instructions were similar to those for the Self-Act Report.
Procedure
Data gathering for the main study occurred in two ses-sions, separated by several days. Each session lasted aboutthree hours, although the time needed to complete theprocedures varied across individuals. In the first session,participants completed the confidential biographical ques-tionnaire, the Self-Act Reports, and other measures. Inthe second session, participants completed the Observer-Act Report, the vocabulary test, and several other measures.Subjects were tested in groups that ranged from two (asingle couple) to 14 (seven couples). Each couple was sep-arated for the duration of the testing session to preventdiscussion of the forms.
Results
Spouse Correlations in BackgroundVariables
To establish comparability between thefindings of the present study and those reported
PERSON-ENVIRONMENT CORRELATION 365
Table 1Spouse Correlations in Age, Physical Traits, and Background Variables
Variables
AgeReported height (in inches)Reported weight (in pounds)
Reported smoking frequencyReported drinking (alcohol)
Previously married (yes-no)Number of years knownNumber of years marriedSatisfaction with marriage
Vocabulary test
Unadjustedcorrelation
Age
.66—*
. 3 9 W
.30*
Adjusted forspouses'ages
: and physical variables
3 7 » *.29*
Consumption variables
.43*** .41***
. 3 3 " .34*"
.37"*
.99*"9 9«.»7 4 « .
.34***
Marital valiables
. 3 3 "
.9S>...
.99"*
.72*"
Cognitive variable
.34*"
Husbands
M
28.770
167
0.51.4
SD
5.52.8
24
1.30.8
8%6.5 3.83.7 3.67.2 1.3
34.8 9.0
Wives
M
27.565
127
0.31.3
7%6.73.77.1
30.9
SD
5.22.7
21
0.90.8
3.93.61.5
10.4
• p < .05. •• p < .01. ••* p < .001.
in the behavioral genetics literature, Table 1shows descriptive statistics and spouse corre-lations for age, physical variables, smoking andalcohol consumption, vocabulary, and severalmarital variables. The spouse correlations forage, height, and weight are comparable to thosefound in previous studies. For example, Priceand Vandenberg (1980) report spouse corre-lations for age of .83 (Swedish couples) and.91 (American couples), whereas the presentstudy found a correlation of .86.
Similarly, the spouse correlations for thesmoking and alcohol consumption variablesare similar to those reported in the literature.Price and Vandenberg found (for Americancouples) correlations of .46 and .41 for drink-ing and smoking, whereas the present studyfound correlations of .33 and .43. Controllingfor age does not seem to affect these corre-lations much.
The cognitive measure (vocabulary) showsa spouse correlation of .34 in this study. Pre-vious studies (e.g., Johnson et al., 1980; Price& Vandenberg, 1980; Watkins & Meredith,1981; Zonderman et al., 1977) show similarmagnitudes for a variety of cognitive measures.Johnson et al. (1980) report a mean correlation
across a large array of studies (corrected foriV)of.35.
The marital variables are of interest for sev-eral reasons. First, the .99 correlations betweenspouses on years known and years married,while not surprising, lend credibility to thecare and accuracy with which the procedureswere completed. Second, the positive corre-lation (r = .37) on previously married repli-cates the often noted trend for the divorcedto marry the divorced and the previously un-married to marry previously unmarried. Andthird, the striking correlation between spousesfor marital satisfaction (.74) suggests thatspouses agree highly on perceived maritalquality.
Like previous studies in the assortativemarriage literature, the present study finds nostrong effects for controlling statistically forthe ages of the spouses. This suggests that co-hort effects can be ruled out as a hypothesisin accounting for the observed correlationsbetween the spouses on these variables. In sum,the results in Table 1 suggest strong compa-rability between the present study and thosereported in the literature.
366 DAVID M. BUSS
Table 2Spouse Correlations in Act Performance: Composites Within Eight Categories
Category
ExtrovertedIntrovertedQuarrelsomeAgreeableDominantSubmissiveCalculatingIngenuous
Mean r
S Data
Unadjustedcorrelation
.22*
.09
.24*
.14
.22*
.20
.08
.37***
.20
Adjustedforage
.21*
.08
.24*
.14
.23*
.19
.10
.37***
.20
O Data
Unadjustedcorrelation
.48***
.20
.27*
.22*
. 2 8 "
.31**
.28**
.46***
.31
Adjustedforage
.49***
.20
.27*
.23*
.27*
.29**
.27*
.45***
.31
Note. S = Self, O = Observer.* p < .05. ** p< .01. *** p< .001.
Spouse Correlations in Eight DispositionalCategories
The major novel measures in this study wereassessments of the performance frequencies ofthe 800 independently generated interpersonalbehaviors. Recall that two independent as-sessments of the frequencies with which theseacts were performed within the past threemonths were obtained: self and observer(spouse) generated. Composite scores werecomputed for each category, consisting of the25 most prototypical acts within each (see Buss&Craik, 1980, 1981, 1983, 1984), excludingacts with extreme base rates (less than 10% orgreater than 90% endorsement). These com-posites were computed for each subject, sep-arately for each of the two data sources.
This study sought to examine spouse cor-relations (reflecting either similarity of dissim-ilarity) of acts performed in the 3-month pe-riod prior to assessment. In this context, it isappropriate to examine correlations from thetwo data sources separately so that those de-rived from the self-report data source can becorroborated by those derived from the spouse-observer data source. This dual analysis allowsexamination of whether the patterns of resultsreplicate across different data sources. Thus,two sets of correlations are shown in Table 2:between husband and wife (a) from their re-spective self-reports of act performance and(b) from their reports of each others act per-
formance. Two columns of correlations areshown for each data source—unadjusted andadjusted (through partialling) for spouses' ages.
All correlations are positive, suggesting thatsimilarity, rather than complementarity, is therule in the interpersonal domain. The cate-gories showing the greatest spouse similarityacross data sources (self (S) and observer (O)reports) are extraverted, quarrelsome, domi-nant, and ingenuous. Of interest is the findingthat observer act reports tend to yield highercorrelations than do self-reports of act per-formance (average difference = .11 points).Although the present data do not afford anexplanation for this difference, several spec-ulations may be offered.
First, the observer reports were obtained inthe second testing session, and thus may havebeen placed in a more familial context (e.g.,reflecting the family party line) than were theself-reports of act performance, which wereobtained in the initial session. Second, the ob-server reports may have focused attention moreexclusively on public observable acts, yieldinghigher spouse similarity, while the self-reportspermit reporting of private acts not performedin the presence of the spouse. Third, the dif-ferences in the perspectives of the self and ob-server (e.g., in terms of impression manage-ment) may have been sufficient to cause theobtained discrepancies (e.g., Laing, Phillipson,& Lee, 1966). Clarification of these alternativeexplanations must await replication and al-
PERSON-ENVIRONMENT CORRELATION 367
Table 3Spouse Correlations in Extroverted Act Performance
Extraverted acts
I went to a discoI went to a nightclubI went skiingI rode on a motorcycleI danced in front of a crowdI dressed in flashy clothesI played sports over the weekendI drank a lot of alcohol at the partyI spoke openly about sexI wore a sexy outfit to the dance
Spouse correlations
SData
.74***
.69"*
.51***
.64*"
.52***
.48***
.46*»*
.46*"
.42***
.38***
O Data
.58***
.57*"
.64***
.81***
.63***
.26*
. 3 0 "
.45*"
.43*"
.29**
Base
S Data
113120
' 5313524436011
rates
OData
1027176
303018385515
S X O
.67***
.65***. .72*"
.91***
.58*"
.39***
.59***
.55***
.44*"
.52"*
Note. S » Self, O = Observer.*p< .05. ** p< .01. •*• p < .001.
tentative research designs (e.g., enlisting otherobservers such as friends to record act per-formance).
In order to examine spouse correlations ingreater detail, correlations were computed foreach of these 800 acts, for each data sourceand with spouse's ages partialled and unpar-tialled.2 The resulting 3200 correlation coef-ficients pose formidable reportorial problems.In addition, one would expect a certain num-ber of these to appear significant by chancealone. To reduce this formidable array and todecrease the probability that chance findingswould be reported, two criteria were adopted,both of which had to be fulfilled: (a) that theunpartialled correlation coefficients for theself-reported act performance had to be sig-nificant beyond the .05 level (two-tailed), and(b) that the unpartialled correlations derivedfrom the observer data source also had to besignificant beyond the .05 level (two-tailed).
Application of these criteria yielded 112(14%) of the 800 acts as having significantspouse correlations across data sources, whereapproximately two would be expected bychance alone. Of these, 58 were significant be-yond the .01 level (two-tailed) for both selfand observer data sources, where less than onewould be expected by chance alone. All werepositive in direction. The frequencies withineach of the eight categories are as follows: ex-traverted (26), introverted (9), quarrelsome(11), agreeable (14), dominant (18), submissive(10), calculating (9), and ingenuous (15). Thus,
the interpersonal categories of extraverted anddominant showed the highest frequency of sig-nificant spouse correlations, whereas the cat-egories of introverted and calculating showedthe lowest frequency of significant spouse cor-relations.
For reportorial purposes, Tables 3-10 showonly the 10 (or nine) acts with the largest cor-relation magnitudes (from the self-reportedperformance) for each of the eight categories.Also reported are the base rates (percentageof the sample who reported performing eachact) and the S X O correlations for each actTo conserve space, adjusted correlations arenot reported.3 All adjusted correlations werewithin .05 correlation points of the unadjustedvalues.
Table 3 shows the acts with the largest spousecorrelations from the extraverted act category.The magnitudes are striking, more so for thiscategory than for the others. Reviewing thecontent of these acts reveals that spouses arecorrelated in the environments they select: at-tending discos, nightclubs, ski resorts, sporting
2 Correlations were computed twice, once using reportedfrequency and once using dichotomized (yes-no) scores,indexing whether or not the act had been reported at leastonce. They were highly similar, therefore, only the formerare reported here. The full set of analyses may be obtainedfrom the author.
3 These additional analyses may be obtained from theauthor.
368
Table 4Spouse Correlations in Introverted Act
Introverted acts
I unplugged my phone for theevening
I watched the soap opera on TVI jogged aloneI escaped to a wild natural locationI stayed at home to watch TV
rather than attend the partyI went for a long walk aloneI pretended I was sick to avoid
attending the partyI gave one-word answers to personal
questionsI wrote a poem
DAVID M
Performance
. BUSS
Spouse correlations
S Data
.56*"
.56"*
.44*"
.43*"
. 3 3 "
. 2 8 "
.26*
.26*
.25*
OData
.67*"
.34*"
.22*
. 3 1 "
. 3 1 "
.21*
.41"*
.25*
.38"*
Base
SData
27333027
3758
16
3718
rates
OData
24293024
3242
15
3716
S X O
.57*"
.73"*
.78*"
.41*"
.29"*
.28"*
. 2 0 "
.26*"
.66*"
Note S = Self, O = Observer.• p < .05. " p < .01. " • p < .001.
events, and parties. In addition, the acts suggestcorrespondence in inhibition-disinhibition ofdress (e.g., flashy clothes, sexy outfits) as wellas behavior (e.g., dancing in front of crowds,talking openly about sex).
Table 4, snowing the acts with the largestspouse correlations within the introverted cat-egory, supports the findings from the extra-verted category. Spouses are correlated in actsthat entail avoiding the loud and uninhibited
settings seen in Table 3. For example, spousesshow correspondence in watching TV (soapoperas), walking alone, avoiding parties, andescaping to nature. The similarity of somecouples in loud group activities appears to bereplaced in other couples by similarity in en-joyment of quiescence and solitude.
Tables 5 and 6 show the spouse correlationsin the categories of quarrelsome and agreeable.These acts appear to differ from those of the
Table 5Spouse Correlations in Quarrelsome Act Performance
Quarrelsome acts
I made fun of him or her for having arunny nose
I yelled at my partnerI criticized someone for failing to put
the napkin on his or her lapI cursed at my parentsI made belittling comments about the
people who walked byI argued about whose turn it was to
driveI played my stereo loudly at 1 a.m.I made fun of his driving abilityI argued about who was the best novelistI criticized a minority group for being
lazy
Spouse
S Data
.53*"
.50***
.48"*
.48"*
.46"*
.40"*
. 3 3 "
.29"
. 2 9 "
.23*
correlations
O Data
.34"*
.53"*
.65"*
. 2 9 "
.35*"
.22*
.30"
. 3 5 "
. 3 5 "
.25*
Base
S Data
1282
711
49
11103617
18
rates
O Data
1073
1014
35
169
3216
17
S X O
.51*"
.54"*
.68*"
.41"*
.48"*
.38"*
.39*"
.47"*
.37*"
.28*"
Note. S = Self, O = Observer.• p < .05. " p < .01. * " p < .001.
PERSON-ENVIRONMENT CORRELATION 369
Table 6Spouse Correlations in Agreeable Act Performance
Agreeable acts
I drank a lot at the social gatheringwhen everyone else did
I tried to please others at the dinnerparty
I remained patient when the car ran outof gas
I made witty remarks at the partyI took my friend to the baseball gameI showed sympathy with my friend's
troublesI watched a different TV show because
someone else wanted toI picked up the tab for lunchI compromised about where to go out
to eatI told a joke to lighten up a tense
situation
Spouse correlations
S Data
.44"*
.37*"
. 3 2 "
. 3 2 "
. 3 1 "
. 3 1 "
.26*
.25*
.23*
.22*
OData
. 3 3 "
.24*
.24*
.38"*
.50"*
. 3 2 "
.29"
.37"*
. 2 6 "
. 3 3 "
Base
S Data
72
70
765
5
95
5755
86
61
rates
O Data
80
75
1172
5
88
5952
83
46
S X O
.45"*
.31*"
.32*"
.51*"
.08
. 1 9 "
.26"*
.45"*
.26*"
.20"
Note. S = Self, O = Observer.* p < .05. " p < .01. " * p < .001.
extraverted and introverted categories in thatthey seem to refer more to specific actionsrather than to the selection of settings. Thus,spouses are correlated in yelling at each other,belittling passersby, and generally criticizing,cursing, arguing, and making fun of others.They also show congruence in trying to pleaseothers, making witting comments, showing
sympathy, and compromising about restaurantchoice.
The findings for dominant and submissiveacts, shown in Tables 7 and 8, amplify thethemes seen earlier. For example, there appearto be correlations between spouses in planningparties (probably the same party held by thecouple together) as well as in dominant acts
Table 7Spouse Correlations in Dominant Act Performance
Dominant acts
I advocated an idea that was ahead of thetimes
I made a bold sexual advanceI decided which TV programs we would
watchI refused to accept the compromiseI took the initiative in planning the partyI directed the conversation around to
myself and my doingsI hung up the phone on my partnerI haggled over pricesI readily used the authority of my positionI deliberately arrived late for the meeting
Spouse correlations
S Data
.45*"
.44*"
.38"*
.36"*
.35*"
. 3 3 "
. 3 3 "
. 3 2 "
. 3 1 "
. 3 0 "
OData
. 2 8 "
.39"*
.25*
. 3 3 "
.27**
.25*
.22*
.24*
.24*
.22*
Base rates
SData O
5556
723431
4329554618
Data
4871
673439
2520493515
S X O
. 41"*
.47*"
.43*"
.28"*
.35*"
.07
.39"*
.40*"
.56*"
.17*
Note S = Self, O = Observer.* p < .05. " p < .01. " • p < .001.
370 DAVID M. BUSS
Table 8Spouse Correlations in Submissive Act Performance
Submissive acts
1 smoked marijuana when everyone elsedid even though I didn't want to
I made coffee for my lover in the morningI requested no compensation when an
acquaintance lived in my apartmentAlthough not really interested in it, I
watched the big TV show becauseeveryone would be talking about it thenext day
I picked up the visitor at the airport eventhough bus transportation was available
I agreed to go out with someone I didn'tlike
When the three of us set out on thejourney, I took the back seat of the car
I straightened up my room when mypartner asked me to
I used my car for the group trip anddidn't ask for gas money from theothers
I was passive in the sexual encounter
Spouse correlations
S Data
. 6 2 W
.59***
.56«*
.52**»
.SO***
.35**»
.28"
.27**
.27"
.22*
OData
.25*
.36***
4 9 « »
.32"
.29~
.23*
.25*
.22*
.25*
.2A*
Base
SData
0656
18
09
32
20
53
68
2750
rates
OData
0753
15
12
26
29
36
73
5744
S x O
AT**.13***
.13*
.46*"
.31~»
.38***
.42»«*
.32***
.48***
Note S = Self, O = Observer.*p< .05. •* p < .01. *** p< .001.
of a quarrelsome nature (e.g., hanging up thephone on the partner). But several otherthemes warrant notice. Acts that reflect anarrogant tone show spouse correlations: de-liberately arriving late, refusing to compro-mise, and directing the conversation aroundto the self.
In addition, spouse correlations with respectto the sexual acts seem intriguing, and perhapsparadoxical at first. Spouses show congruenceboth in making bold sexual advances and inremaining passive in the sexual encounter. Al-though one might expect complementaritywith respect to these acts, these results suggestthat, on the contrary, behavior in the sexualdomain appears to be similar on the boldness-passivity dimension.
Tables 9 and 10 show spouse correlationsfor calculating and ingenuous acts. Several ofthe calculating acts suggest common actionsperformed by the members of the couple to-gether not answering the phone when homeand going to a friend's house for food. Others,however, suggest correspondence between in-dividual acts performed separately: taking
credit for a co-workers idea, letting someoneelse pay for cocktails, and wearing sexy clothesto impress someone.
The ingenuous acts suggest a selection ofcommon settings and activities: going tochurch, going to the beach, and letting a friendstay in the apartment while away. Others,however, suggest mutuality of affection. Thus,spouses are correlated on the acts of saying "Ilove you" and hugging a friend. Leaving thecar and apartment unlocked and answeringthe door without asking who was there suggestsimilarity in general trust. The correlation be-tween spouses on the act "I cosigned a loanfor my friend" might be a spurious result ofthe low base rate shown for that act Mostother base rates are sufficiently large to ruleout this interpretation generally.
In sum, spouse correlations with respect toa variety of acts within eight interpersonal cat-egories are striking in magnitude. Severalthemes emerge from these spouse correlations:(a) commonality in the selection of settings(e.g., attending church, going to the beach,attending discos, nightclubs, ski resorts,
PERSON-ENVIRONMENT CORRELATION 371
Table 9Spouse Correlations in Calculating Act Performance
Spouse correlations Base Rates
Calculating acts SData OData SData OData S X O
I did not answer the phone, even thoughI was home .47*** .52"* 49
I let someone else pay for the cocktails .43*** .51*** 511 went over to a friend's place to get food .38*** .31*** 34I wore seductive clothes .37*** .50*** 27I told everyone I was broke, even though
I wasn't .36*** .42*** 15I paid my bills the last day they were due
to obtain the highest interest .32** .28** 44I wore sexy clothes to impress someone .31** .29** 26I took credit for a co-worker's idea .24* .65*** 7
Note. S = Self, O = Observer.• p < .05. ** p < .01. *** p < .001.
37452730
11
38254
.60***
.52***
.35***
.60*—
.46***
.52***
.40***39***
dances, baseball games); (b) similarity in attire(e.g., flashy clothes, seductive clothes, sexyoutfits); (c) mutuality in physical and verbalaffection (e.g., hugging, saying "I love you,"active-passive orientation in the sexual do-main); (d) correspondence in social orientation(e.g., staying at home to watch TV, taking longwalks alone, dancing in front of crowds, de-clining party invitations, planning parties); and(e) correspondence in positive or negative affecttoward others and perhaps toward each other
(e.g., yelling at partner, criticizing a minoritygroup, showing sympathy with a friend).
Two additional findings warrant comment.First, although a few of the acts may showsignificant spouse correlations due to extremebase rates, most base rates appear to be mod-erate. Thus, extreme base rate can be generallyruled out as an explanation for the sizablespouse correlations for these acts. Second, theS X O correlations are generally high, but sev-eral factor may have attenuated them. For ex-
Table 10Spouse Correlations in Ingenuous Act Performance
Ingenuous Acts
I cosigned a loan for my friendI went to churchI went to the beach by myselfI let a friend stay at my apartment while
I was awayI left my apartment unlocked at nightI did not object to my partner spending
time with members of the opposite sex1 left my car unlocked while I ran an
errand1 opened my door without asking who
was thereI told someone "I love you"I hugged my friend
Spouse Correlations
SData
.80***
.69*"
.58***
.41***
.40***
.38***
.36***
.35***
.29**
.26*
OData
.49***
.69***
.45***
.27**
.21*
.46*"
.48***
.21*
.40"*
.24*
Base
SData
244
5
2022
54
47
514997
Rates
O Data
342
6
1119
51
34
393294
S X O
.77*"
.81*"
.46*"
.38***
.36***
.42*"
.61***
.33***
.47"*
.41***
Note. S = Self, O = Observer.*p< .05. ** p < .01. •** p < .001.
372 DAVID M. BUSS
ample, the S and O data assessments wereseparated by several days, and correlations lessthan unity may reflect veridical differences inthe act performance from which the reportsdraw. Another attenuating factor pertains tothe performance of private acts, unseen by thespouse, which might emerge on the self-reportbut not on the observer report. This suggestiongains empirical credibility from the generallylower base rates shown for the O than for theS data source. Finally, the differing perspectivesentailed in observing and reporting on the selfversus observing and reporting on others wouldserve to attenuate the S X O correlations. Theseissues are interesting in their own right anddeserve research attention. In the present con-text, the it may be noted that the obtainedS X O correlations for specific acts are generallyquite robust.
Changes in Spouse CorrespondenceWith Time
This study also sought to assess the degreeto which spouses converge or diverge in actperformance in the interpersonal domainacross time. As Price and Vandenberg (1980)note, assessing the linear effects of length ofmarriage does not test directly for the presenceor absence of phenotypic convergence or di-vergence during marriage. Rather, change inthe degree of spouse similarity as a functionof length of marriage (either convergent or di-vergent) is appropriately assessed by the in-teraction between a time variable and onespouse member's score in predicting the otherspouse members score. Price and Vandenberg(1980) recommend using hierarchical multipleregression in which the years of marriage areentered as the first step, one spouse's score ona given variable is entered as the second step,and the interaction term is entered as the thirdstep. The interaction term, when entered asthe last step in hierarchical multiple regression,is independent of the other effects. Testing forthe statistical significance of the increment inR provides a test of whether there is changein the degree of similarity over time. A positivebeta (0) suggests convergence; a negative beta(—0) suggests divergence.
The procedure suggested by Price and Van-denberg (1980) was followed closely in theanalysis of changes in spouse correspondenceover time on each of the eight act composites
and on each of the 800 acts. However, eventhough Price and Vandenberg employed num-ber of years married as the time variable, itcould be argued that the length of time thecouples knew each other might be a bettertime variable, particularly in an era wheremany couples live together and interact in-tensively prior to marriage. Thus, hierarchicalmultiple regressions were performed on eachof the eight act composites and each of the800 acts; once by using number of years mar-ried as the time variable and once by usingnumber of years known as the time variable.
Regressions for the eight act compositesshowed that none of the dispositions! cate-gories yielded significant R increments for theinteraction term across the two data sources.Thus, no general increases or decreases inspouse correspondence are associated withyears known or years married in this data set.Of the 800 specific act level multiple regres-sions performed by using the length of mar-riage variable, 75 interactions proved to besignificant beyond the .05 level. For the anal-ogous regressions performed by using thelength of time known, 93 interactions (about12%) proved to be significant beyond the .05level. Thus, it appears that length of timeknown might be a a slightly better time vari-able in predicting changes in spouse corre-spondence than is length of marriage. The di-rections of the two sets of analyses, however,were quite similar, thus, only the results of theregressions using the length of time knownwill be reported (the complete set of analysesmay be obtained from the author).
The first finding was unexpected: There weremany more acts on which couples became lesssimilar with time (divergence) than vice-versa(convergence). Of the 93 significant interac-tions, 56 showed divergence and 37 showedconvergence. The interpersonal categorieswithin which these significant interactions oc-curred are (convergent, then divergent in pa-rentheses): extraverted (7, 5), introverted(3, 3), quarrelsome (10, 12), agreeable (0, 8),dominant (2,8), submissive (3,10), calculating(6,3), and ingenuous (6,7). Thus interpersonalacts within three categories (dominant, sub-missive, and agreeable) primarily show diver-gence with time, whereas the acts that showconvergence are distributed across categoriesand not concentrated heavily within any.
PERSON-ENVIRONMENT CORRELATION 373
Space limitations preclude presenting theresults from all of these analyses. Instead, Ta-bles 11-13 show selected regressions: those forthe acts showing divergence within the cate-gories of dominance and submissiveness, anda subset of the acts that show the greatestamount of convergence. In interpreting theseresults, it should be noted that because thisstudy is not longitudinal, convergence and di-vergence cannot strictly be claimed. As a cros-sectional study, an interpretation of these re-sults in terms of convergence and divergencerequires the (untested) assumption that dif-ferences in similarity between the couples oflong and short duration reflect change overtime.
As can be seen in Table 11, the dominantacts that show divergence are primarily"egoistic" or "agentic" in nature (see Buss,1981b for a discussion of different themes ofdominant acts). Flattering others to get one'sway, telling others to perform one's menialtasks, and directing the conversation aroundto the self all seem to embody displays of in-
fluence, but for selfish rather than for groupaims. The submissive acts that show divergence(Table 12) seem to involve being the recipientof these selfish dominant acts: accepting verbalabuse from others, taking on all the blamewhen the group failed, and following the dic-tates of others.
One speculative interpretation may beplaced on these results: That in spite of initiallevels of assortment for dominance, and despitethe results reported earlier that couples arequite similar with respect to many dominantacts, spouses may, through the course of mar-riage, diverge with respect to dominance andsubmissiveness. That is, they may take on dif-ferent roles within the marriage—one assum-ing the role of director, the other assuming therole of follower who submits to the requestsof the emerging dominant parter. This finding,however, appears to occur only with a subsetof these acts and not for the disposition gen-erally.
In contrast to the results suggesting diver-gence, no prominent themes seem to emerge
Table 11Dominant Acts Showing Divergence
Wife's variable
I flattered him in order to get my way.
I reported someone who had broken arule.
When someone cut ahead of me in line,I protested loudly.
I settled the dispute among othermembers of the group.
I told others to perform menial tasksinstead of doing them myself.
I directed the conversation around tomyself and my doings.
I managed to get my own way.
I initiated the conversation with thestranger.
Hierarchicalstep
years knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteraction
MultipleR
.04
.08
.28
.03
.04
.23
.07
.13
.29
.10
.30
.50
.06
.23
.35
.08
.34
.41
.05
.17
.31
.18
.25
.35
R2
.00
.01
.08
.00
.00
.05
.01
.02
.09
.01
.09
.25
.00
.05
.12
.01
.11
.17
.00
.03
.10
.03
.06
.12
Significanceof increment
nsns.05nsns.05nsns.05nsns.01nsns.05ns.05.05nsns.05nsns.05
Note, ns = not significant.
374 DAVID M. BUSS
from the regressions showing convergence, orincreasing spouse similarity with time (see Ta-ble 13). Instead, they seem to be a hodgepodgeof acts: some reflecting isolation and separation(e.g., ignoring the partner, saying "do it your-se l f ) and others reflecting possible joint ac-tivities (e.g., watching a soap opera on TV,throwing a party). Perhaps the present set ofeight interpersonal categories were insufficientfor the task of discerning the important themesof interpersonal convergence that may haveoccurred.
Discussion
Identifying correlations between attributesof persons and features of their interpersonalenvironment is perhaps one of the most in-triguing issues facing the incipient field of in-teractional psychology. Little empirical prog-ress has yet been made, however, in docu-
menting the domains within which PEcorrespondence occurs, in identifying themechanisms by which such correspondence isbrought about, and in exploring the potentialconsequences of obtained correspondence(Buss, 1983, in press-a, in press-b). This articlesought to create a bridge between the studyof assortative marriage (traditionally residingwithin behavioral genetics and sociology) andpersonality psychology. It was suggested thatassortative marriage may be one mechanismby which correspondences between personsand interpersonal environments are created.In addition, it was suggested that within suchcorrespondence may lie important clues tovariables that affect adult personality devel-opment.
The first step requires documentation of thedomains within which PE correspondence oc-curs (Buss, in press-b). This study sought toassess, in a relatively comprehensive way,
Table 12Submissive Acts Showing Divergence
Wife's Variable
When the group failed at its task, 1took on all the blame for it.
I accepted verbal abuse withoutdefending myself.
I laughed at a joke that was not funny.
Although not really interested in it, Iwatched the big TV show becauseeveryone would be talking about itthe next day.
I let someone cut into the parkingspace I was waiting for.
I asked my partner if it would be OKto meet his parents.
I killed a fly when someone asked meto.
I let my friend decide which jacket Ishould buy.
On the new assignments, I followed mysupervisor's directions withoutquestion.
Hierarchicalstep
years knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteraction
years knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteraction
MultipleR
.03
.20
.70
.01
.04
.34
.02
.15
.29
.06
.52
.66
.00
.10
.33
.11
.29
.37
.10
.14
.27
.09
.38
.58
.08
.11
.46
R*
.00
.04
.49
.00
.00
.11
.00
.02
.08
.00
.27
.44
.00
.01
.11
.01
.08
.14
.01
.02
.08
.01
.14
.34
.01
.01
.21
Significanceof increment
nsns
.001nsns.01nsns.05ns
.001
.001
nsns
.01nsns
.05nsns
.05ns
.05
.001nsns
.001
Note, ns = not significant.
PERSON-ENVIRONMENT CORRELATION 375
spouse correspondence across a large array ofspecific interpersonal behaviors within eightdisposition^ categories. In light of the averagespouse correlation reported in the literatureof. 15 for traditional personality scales (Jensen,1978), the magnitudes of the present resultsstand out, although some are clearly due tocouple co-acting rather than to independentcorrespondence of performance.
The present study found substantial spousecorrelations with respect to a variety of specificinterpersonal behaviors across all eight dis-positional categories examined. These corre-lations were greatest for the categories of ex-traversion, dominance, quarrelsomeness, andingenuousness, and lowest for the category ofcalculating acts, although significant correla-tions across data sources were found withinall eight interpersonal categories. Severalthemes emerged: (a) correlation in the selectionof extraverted settings (e.g., going to discos,nightclubs, ski resorts, dances, parties); (b)correspondence in interpersonal intimacy (e.g.,talking openly about sex, passivity and bold-ness in the sexual arena, hugging friends, and
expressing love verbally); (c) similarity in pre-ferred leisure activities (e.g., jogging, watchingsoap operas, writing poems); (d) mutuality inquarrelsome responses (e.g., yelling at one'spartner, arguing about novelists); (e) correla-tion of extraverted attire (e.g., flashy clothes,sexy outfits); and (f) similarity in calculatinginterpersonal tactics (e.g., deliberately arrivinglate for a meeting, directing the conversationaround to the self, taking credit for a co-work-er's idea).
These results suggest two forms of PE cor-relation. The first is direct—a correspondencebetween an attribute of one person (e.g., ex-traversion) and that person's interpersonal en-vironment (spouse's extraversion). The secondis suggested by examining the environmentsimplied by similarity on specific acts(e.g., parties, nightclubs, church, wilderness).Spouses are thus correlated with respect to theenvironments they select. Correlated "niche-picking" (Scarr & McCartney, 1983) repre-sents a second way in which PE (or environ-ment-environment) correlation occurs.
Another purpose of this study was to ex-
Table 13Sample Acts Showing Convergence
Wife's Variable
I ignored my partner at the dance.
I insisted on playing a record I knewmy friend didn't like.
I was the first to stand up to delivera standing ovation.
I invited a stranger to stay at myapartment.
I did not start a single conversationat the party.
When someone asked me for a favor.I said "do it yourself."
I watched the soap opera on TV.
I threw a big party.
Hierarchicalstep
years knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's variableinteractionyears knownhusband's scoreinteraction
MultipleR
.05
.65
.82
.01
.07
.58
.21
.27
.57
.05
.52
.63
.01
.48
.60
.13
.23
.44
.12
.56
.65
.05
.42
.52
R2
.00
.42
.68
.00
.00
.33
.04
.07
.33
.00
.27
.39
.00
.23
.36
.02
.05
.19
.01
.32
.42
.00
.18
.27
Significanceof increment
ns.001.001nsns
.001nsns
.001ns
.001
.001ns
.01
.001nsns
.01ns
.001
.01ns
.01
.01
Note, ns = not significant
376 DAVID M. BUSS
amine the degree to which spouses becomemore or less congruent over time. Becausespouses may have converged prior to the actualmarriage, the number of years of the rela-tionship might be a better index than numberof years of the marriage. Both indices wereused in a series of hierarchical multiple regres-sions designed to assess changes in spouse sim-ilarity. Regressions using the number of yearsknown emerged more robustly in predictingchanges in spouse correspondence.
Although no significant trends were foundfor the act composites, one surprising resultwas that divergence was found more frequentlythan convergence at the single-act level. Theacts that showed convergence were dispersedthroughout the eight interpersonal categories,suggesting that the category system employedin this study may be insensitive to the majordimensions along which spouses become moresimilar with time. The acts that showed di-vergence were more clearly concentrated inthe dominant, submissive, and agreeable cat-egories. These results must be viewed as pro-visional in that interpretation of the differencesbetween older and younger couples on simi-larity in terms of convergence or divergenceis based on the untested assumption that thelocus of the differences is temporal rather thangenerational. Longitudinal studies are neededto verify the temporal interpretation.
What are the implications of the results forpersonality development? First, the search forbig variables affecting personality has met withlittle success (see, e.g., Loehlin & Nichols,1976; Willerman, 1979). The present resultssuggest that one's marital partner may be onesuch big variable. One way in which this vari-able might operate entails the establishment(and selection) of environments that reinforceinitial propensities. Mutuality of niche pickingproduces a self-perpetuating cycle in whichcommon environments create shared experi-ences and maintain shared interests. Deviationfrom the marital mean by one partner maybe strongly resisted by the other. The processes(e.g., reinforcement contingencies) set intomotion by initial spouse similarity may serveto maintain personality consistency frequentlyfound in longitudinal studies (e.g., Block, 1971;Costa & McCrae, 1980; Costa, McCrae, &Arenberg, 1980).
Another way in which the marital partnermight operate as a big variable pertains to the
finding that older couples are less similar thanyounger couples for specific acts within thedomains of dominance and submissiveness.Perhaps this is one area where complemen-tarity begins to emerge only after the rela-tionship progresses considerably. In spite ofinitial similarity, spouses may adopt comple-mentary roles as one partner gains relativeascendance over the other.
An interesting parallel may be noted be-tween the present results and recent findingssummarized by Rowe and Plomin (1981) inthe domain of child development. Rowe andPlomin conclude that most environmentalvariables affecting child development areprobably nonshared factors operating withinfamilies (causing siblings to become differentfrom each other) rather than shared influences(causing family members to become moresimilar to each other). The present study foundstriking spouse similarity in interpersonal be-havior, yet notably more divergence than con-vergence associated with length of time spousesknew each other, particularly in the domainsof dominance, submissiveness, and agreeable-ness. These preliminary findings suggest that,as with child personality development, theremay be important factors operating withinmarriages causing spouses to become moredissimilar to each other, in spite of considerablemutuality exhibited in interpersonal behavior.
The present research must be viewed as onlya preliminary attempt to uncover the domainswithin which correlations between persons andinterpersonal environments occur and to dis-cover the mechanisms responsible for obtainedcorrelations. The results are sufficiently robustacross data sources to suggest that substantialspouse correspondence can be found in theinterpersonal domain, particularly with ref-erence to dominance, extraversion, quarrel-someness, and ingenuousness. Because ad-justing for spouses' ages did not decrease thecorrelations appreciably, age and cohort canbe ruled out as factors responsible for obtainedcorrelations. Further, the hypothesis that ob-tained spouse correlations result from con-vergence over the course of the marriage isnot supported by these data. Initial assortmentis implicated as one cause of obtained spousecorrespondence.
Although spouse selection provides a dra-matic example of choosing one's interpersonalenvironment, other domains (e.g., assortative
PERSON-ENVIRONMENT CORRELATION 377
friendship) and different processes (e.g., evo-cation and alteration) can be subsumed prof-itably by a PE correlation framework. In ev-eryday life, individuals rarely assign themselvesrandomly to conditions. The present resultssuggest that personality assessment can be ex-tricated from broader personal contexts onlyat a cost, and that an integral part of person-ality is the process of selecting, evoking, andaltering environments within which one re-sides. The study of spouse selection is but onestep toward developing a psychology of PEcorrelation.
ReferencesBlock, J. (1971). Lives through time Berkeley, CA: Bancroft
Books.Buss, D. M. (1981a). Predicting parent-child interactions
from children's activity level. Development Psychology,17, 59-65.
Buss, D. M. (1981 b). Sex differences in the evaluation andperformance of dominant acts. Journal of Personalityand Social Psychology, 40, 147-154.
Buss, D. M. (1983). Evolutionary biology and personalitypsychology: Implications of genetic variability. Person-ality and Individual Differences, 4. 51-63.
Buss, D. M. (1984). Marital assortment for personalitydispositions: Assessment with three different datasources. Behavior Genetics, 14, 111-123.
Buss, D. M. (in press-a). A conception of the interpersonalenvironment from the act frequency perspective. In R.Hogan & W. Jones (Eds.), Perspectives in personality:Theory, measurement and interpersonal dynamicsGreenwich, CT: JAI Press.
Buss, D. M. (in press-b). Evolutionary biology and per-sonality psychology: Toward a conception of humannature and individual differences. American Psychologist
Buss, D. M., Block, J. H., & Block, J. (1980). Preschoolactivity level: Personality correlates and developmentalimplications. Child Development, 51, 401-408.
Buss, D. M., & Craik, K. H. (1980). The frequency conceptof disposition: Dominance and prototypically dominantacts. Journal of Personality, 48. 379-392.
Buss, D. M., & Craik, K. H. (1981). The act frequencyanalysis of interpersonal dispositions: Aloofness, gre-gariousness, dominance, and submissiveness. Journalof Personality. 49, 174-192.
Buss, D. M., & Craik, K. H. (1983). The act frequencyapproach to personality. Psychological Review, 90, 105-126.
Buss, D. M., & Craik, K. H. (1984). Acts, dispositions,and personality. In B. A. Maher & W. B. Maher (Eds.),Progress in experimental personality research- Normalpersonality processes (Vol. 13. pp. 241 -301). New York:Academic Press.
Cattell, R. B. (1973). Personality and mood by question-naire: A handbook of interpretive theory psychometricsand practical procedures. San Francisco: Jossey-Bass.
Costa, P. T, Jr., & McCrae, R. R. (1980). Still stable afterall these years: Personality as a key to some issues in
aging. In P. B. Baltes & O. G. Brim (Eds.), Life spandevelopment and behavior (Vol. 3). New York; AcademicPress.
Costa, P. T, Jr., McCrae, R., & Arenberg, D. (1980). En-during dispositions in adult males. Journal of Personality,and Social Psychology, 38, 793-800.
Gough, H. G., &. Sampson, H. (1974). The General Vo-cabulary Test. University of California: Institute of Per-sonality Assessment and Research.
Jensen, A. R. (1978). Genetic and behavioral effects ofnonrandom mating. In R. T. Osborne, C. E. Noble, &N. Wey (Eds.), Human variation- Biopsychology of age,race, and sex (pp. 51-105). New York: Academic Press.
Johnson, R. C, Ahem, F. M., & Cole, R.E. (1980). Secularchange in degree of assortative mating for ability? Be-havior Genetics, 10, 1-7.
Laing, R. D., Phillipson, H., & Lee, A. R. (1966). Inter-personal perception: A theory and a method of research.London: Tavistock.
Loehlin, J. C , & Nichols, R. C. (1976). Heredity, envi-ronment, and personality Austin: University of TexasPress.
Loevinger, J. (1976). Ego development- Conceptions andtheories. San Francisco: Jossey-Bass.
Plomin, R., DeFries, J. C , & Loehlin, J. C. (1977). Ge-notype-environment interaction and correlation in theanalysis of human behavior. Psychological Bulletin, 84,309-322.
Plomin, R., DeFries, J. C , & Roberts, M. K. (1977). As-sortative mating by unwed biological parents of adoptedchildren. Science, 196. 449-450.
Price, R. A., & Vandenberg, S. G. (1980). Spouse similarityin American and Swedish couples. Behavior Genetics,10, 59-71.
Rowe, D. G, & Plomin, R. (1981). The importance ofnonshared (El) environmental influences in behavioraldevelopment. Developmental Psychology. 17. 517-531.
Scarr,S.,& McCartney, K. (1983). How people make theirown environments: A theory of genotype —• environ-ment effects. Child Development, 54, 424-435.
Snyder, M. (1981). On the influence of individuals onsituations. In N. Cantor & J. F. Kilstrom (Eds.), Per-sonality, cognition, and social interaction. Hillsdale, NJ.:Erlbaum.
Snyder, M., & Gangestad, S. (1982). Choosing social sit-uations: Two investigations of self-monitoring processes.Journal of Personality and Social Psychology, 43, 123—135.
Vandenberg, S.G. (1972). Assortative mating, or who mar-ries whom? Behavior Genetics. 2, 127-157.
Watkins, M. P., & Meredith, W. (1981). Spouse similarityin newlyweds with respect to specific cognitive abilities,socioeconomic status, and education. Behavior Genetics,11, 1-21.
Wiggins, J. S. (1979). A psychological taxonomy of traitdescriptive terms: I. The interpersonal domain. Journalof Personality and Social Psychology, 37, 395-412.
Willerman, L. (1979). Effects of families on intellectualdevelopment. American Psychologist, 34, 923-929.
Zonderman, A. B., Vandenberg, S. G., Spuhler, K. P., &Fain, P. R. (1977). Assortative marriage for cognitiveabilities. Behavior Genetics, 7, 261-271.
Received April 27, 1983 •