which may indeed be a result of increasing bargaining power on the marriage market.
But it is also associated with the set of variables introduced in these models. Recently
employed and educated women living in advantaged areas are more likely to have
employed partners, though the association with their own qualifications is weakened
when the partner’s qualifications are included (Model 4). The inclusion of area type in
Models 3 and 4 probably accounts for the weakening of the ethnic effect on
unemployment. Model 4 also shows, not surprisingly, that partners with a longstanding
illness are less likely to be employed.
Table 9 considers the employment of the mother at interview. The first model shows that
the later mothers are more likely to be in the labour market than the younger mothers and
this increases monotonically with age. Rather than peaking around 30, the increase
continues strongly across the 30s. Model 2 includes the antecedent variables which
reduces all the effects slightly but maintains the pattern. Women with disrupted families
in childhood, or who left school early are less likely to be employed, as are those in most
ethnic minority groups except Black Caribbean. In model 3 the age pattern is also largely
maintained with much smaller coefficients on age at motherhood, mediated by the battery
of current circumstances and highest qualifications. Women with high qualifications, no
longstanding illness and living in less disadvantaged areas The term registering the
number of other children in the family has the expected negative sign, but the presence of
a partner has a positive coefficient. Since lone motherhood is age-at-motherhood related,
the inclusion of this term helps to moderate the otherwise unexplained age pattern. The
combination of qualification and location information reduces the negative ethnic group
18