86
mative context has a significant effect. By contrast, all the structural fac-
tors are significant predictors of the proportion of female alters, pointing
out the role of structural differentiation processes on the gender balance
of personal networks (Model 4b). Individuals born in 1970–1975 and
individuals with higher education have a higher proportion of female
alters. Confirming our previous findings of an overall orientation to
cross-gender preferences (with the exception of Swiss women), women
show a lower proportion of female alters than men.
Concerning the duration of the relationships, we decided to analyse
only the predictors of very old acquaintances (more than 75% in ego’s life)
and very recent acquaintances (less than 25% in ego’s life) since these were
the two acquaintanceship indicators that showed more variation between
countries. Model 5 shows the predictors of the proportion of very recent
acquaintances and Model 6 shows the predictors of the proportion of very
old acquaintances (Table 3.8). In both models, the variable “country” is a
strong predictor of duration length, as its effect remains significant while
accounting for the other factors. On the one side, we can see that being
born in Lithuania is positively associated with the proportion of very
recent acquaintances and is negatively associated with the sharing of very
old acquaintances. By contrast, Model 6a shows that individuals born in
Switzerland include a higher proportion of very old acquaintances. As
stated before, this association is linked to the higher integration of chil-
dren and grandchildren in Lithuania when compared with the two other
countries (in Portugal, the share of grandchildren is also high, but the
inclusion of ascendants decreases the salience of the former tie). Structural
factors also play a role on the duration of relationships. Women have a
higher proportion of old acquaintances compared to men as it is more
likely that women, as caregivers, include parents and parents-in-law in
their personal networks, increasing the salience of long-lasting acquain-
tances. Models 5b and 5c show that those who were born in 1970–1975
show a lower proportion of very recent alters compared to those who
were born in 1950–1955. Again this is linked to the demographical res-
ervoir of both cohorts. If, on the one hand, some individuals in the
youngest cohort are still childless, on the other hand, the focus on ascen-
dants makes the inclusion of very recent acquaintances less likely. Instead,
the oldest cohort is more likely to include grandchildren and also
R. Gouveia et al.