1 The Data

The data were collected from twins enrolled in the Australian National Health and Medical Research Council Twin register. In this sample, mailed questionnaires were sent to the 5,967 pairs of twins on the register between November 1980 and March 1982 (see also Chapter 10). Among the items on the questionnaire were those from the state depression scale of the Delusions-Symptoms States Inventory (DSSI; Bedford et al., 1976) and a single item regarding marital status. The analyses performed here focus on the like-sex MZ and DZ female pairs who returned completed questionnaires. The ages of the respondents ranged from 18 to 88 years; however, due to possible differences in variance components across age cohorts, we have limited our analysis to those twins who were age 30 or less at the time of their response. There were 570 female MZ pairs in this young cohort, with mean age 23.77 years (SD=3.65); and 349 DZ pairs, with mean age 23.66 years (SD=3.93).

Using responses to the marital status item, pairs were subdivided into those who were concordant for being married (or living in a marriage type relationship); those who were concordant for being unmarried; and those who were discordant for marital status. In the discordant pairs, the data were reordered so that the first twin was always unmarried. Depression scores were derived by summing the 7 DSSI item scores, and then taking a log-transformation of the data [ $x' = log_{10}(x+1)$] to reduce heteroscedasticity. Covariance matrices of depression scores were computed for the six zygosity groups after linear and quadratic effects of age were removed. The matrices are provided in the Mx scripts in Appendices  and , while the correlations and sample sizes are shown in Table 9.3. We note (i) that in all cases, MZ correlations are greater than the corresponding DZ correlations; and (ii) that for concordant married and discordant pairs, the MZ:DZ ratio is greater than 2:1, suggesting the presence of genetic dominance.

Table 9.3: Sample sizes and correlations for Australian female twin depression data.

Zygosity Group	N	r
MZ - Concordant single	254	0.409
DZ - Concordant single	155	0.221
MZ - Concordant married	177	0.382
DZ - Concordant married	107	0.098
MZ - Discordant	139	0.324
DZ - Discordant	87	0.059

Before proceeding with the G $\times$ E interaction analyses, we tested whether there was a G - E correlation involving marital status and depression. To do so, cross-correlations between twins' marital status and cotwins' depression score were computed. In all but one case (DZ twin 1's depression with cotwin's marital status; $r=-0.156$, $p < 0.01$), the correlations were not significant. This near absence of significant correlations implies that a genetic predisposition to depression does not lead to an increased probability of remaining single, and indicates that a G - E correlation need not be modeled.