Behavior Genetics, Vol. 20, No. 5, 1990 Does Contact Lead to Similarity or Similarity to Contact? 1 D. T. Lykken, 2,3 M. McGue, 2 T. J. Bouchard, Jr., 2 and A. Tellegen 2 Received 5 Sept. 1989--Final 23 Jan. 1990 Evidence from the Finnish Twin Registry (e.g., Rose et al., 1988) shows that adult monozygotic (MZ) twins are more similar, within pairs, in personality if the cotwins are presently cohabiting or in frequent contact than if they are seldom in contact. Results of a follow-up study led Kaprio et al. (1990) to conclude that "'changes in social contact between monozygotic cotwins precede (and causally contribute to) changes in their intrapair similarity'" (p. 9). If true, this conclusion has important theoretical implications, e.g., many heritability estimates would have to be revised downward. We adduce evidence suggesting that similarity leads to contact, rather than the other way around. Low corre- lations between twins" frequency of contact and their absolute within-pair dif- ference on all traits thus far studied indicates that, whichever the direction of causality, the relationship between MZ within-pair similarity and their frequency of contact is very weak. KEY WORDS: monozygotic (MZ) twins; shared experience; personality; twin closeness; recruit- ment bias. INTRODUCTION Recent research with the Finnish Twin Registry (Rose and Kaprio, 1988; Rose et al., 1988) has indicated that adult monozygotic (MZ) twins who live together or remain in close contact are more similar in their use of alcohol and in Ex- troversion and Introversion than are MZ twins who maintain less contact with This work was supported in part by Grant 2 RO1 MH37860-06 from the National Institute of Mental Health in support of the Minnesota Twin Registry. 1Richard Rose, Jaakko Kapno, and Christopher Williams will be replying to this paper in a sub- sequent issue. ZDepartment of Psychology, Elliott Hail, University of Minnesota, Minneapolis, Minnesota 55455. 3To whom correspondence should be addressed. 547 0001-8244/90/0900~0547506.00/0 9 1990 Plenum Publishing Corporation548 Lykken, McGue, Bouchard, and TeUegen one another. Rose and his collaborators argue from such data that the marked similarity in personality of MZ twins (Bouchard and McGue, 1990; Horn et al., 1976; Loehlin and Nichols, 1976; Tellegen et al., 1988) may be due in large part to the effect of shared experience resulting from the tendency for MZ twins to remain in closer contact than ordinary siblings. This conclusion, if true, would help explain why MZ similarity in personality is greater than would be predicted on genetic grounds when heritability is estimated from nontwin family data; i.e., the excess MZ similarity results from the greater shared experience of MZ twins. This would make it unnecessary to postulate epistatic (Eaves, 1988) or "emergenic'" (Lykken, 1982) genetic models and would generally lower esti- mates of the broad heritability of personality traits. It is obvious that the direction of the causal arrow relating MZ similarity to frequency of contact is ambiguous. For this reason, Rose and his collaborators (Kaprio et al., 1990) investigated the changes in within-pair similarity of MZ twins who had been cohabiting when first assessed. A translated version of the Eysenck Personality Inventory was readministered to some 540 MZ pairs who" were living together when first studied 6 years previously when the twins were aged 19 to 25. Alcohol usage was also reassessed. On retest, about 24% of the pairs were still living together, 64% of them were in daily or weekly contact, and 12% of them saw each other once a month or less. As in the earlier study, the within-pair correlations for Alcohol Use, Extroversion (E), and Neuroticism (N) each increased regularly and substantially as a function of closeness of contact. For this aspect of the findings, the direction of the causal arrow remains ambiguous; we could suppose that the twins still cohabiting were more similar because of their continued close contact or, at least equally plausibly, that the Iess similar twins were now in less frequent contact because they were less similar. Kaprio et al. (1990) focus instead on the similarity of these three groups 6 years earlier when all pairs were still living together. They argue that the similarity --* (causes) contact hypothesis requires that the twins who later came to be in infrequent contact should be less similar in personality when first studied while still cohabiting; this was the pattern clearly found for Extroversion. The contact --+ similarity hypothesis, however, predicts that subsequent changes, if any, in closeness of contact will have resulted from adventitious circumstances and be unrelated to the initial intrapair similarity. This was the pattern found for alcohol use and for Neuroticism. Kaprio et al. (1990) conclude "that changes in social contact precede (and causally contribute to) changes in their intrapair similarity" (p.9) for Neuroticism and alcohol use. In fact, however, all that is known for sure is that, after 6 years, the twins who had lost contact were less similar than they had been previously and less similar than were the twins who remained in close contact. Except for E, the sequence of these changes--loss of contact and loss of similarity in N or alcoholDoes Contact Lead to Similarity? 549 use--remains unknown. It is possible that the twins who became separated did so for reasons unrelated to their mutual closeness or similarity and then, while infrequently in contact, their correlations for N and alcohol consumption dropped from .6 to .2 as these aspects of their personalities were altered differentially by their different environments, i.e., (loss of) contact ~ (loss of) similarity. A different but equally plausible scenario might show some of these young- adult twins starting to drink heavily after leaving the parental home and then becoming estranged from their cotwins as a result of this change in life-style, i.e., (loss of) similarity --~ (loss of) contact. Because these are MZ twins, the causes of within-pair