that many of these disadvantages, notably low qualifications, are likely to be pre-existing.
Their introduction to the models has the effect of further moderating the relationship
between age at motherhood and the outcome variables, but not usually eliminating,
suggesting a (statistically) independent association between earlier childbearing and the
family’s living standards not accounted for by measured covariates, which again could be
either cause or consequence, but in any case reflects disadvantages for the early starters
not otherwise accounted for. This suggests a particular vulnerability of families where
the mother had her first child in or before her early twenties in addition to the indicators
we have measured.
It is in the analysis of our measures of subjective well being that the age at motherhood
differentials are more nearly completely accounted for when family income is included as
a covariate. This variable mediates almost entirely the relationship between age at
motherhood and our two measures of well-being - mother’s life satisfaction and malaise.
This suggests that material circumstances are a strong determinant of expressing
particularly low life satisfaction, and that if the early mothers had the same incomes as
the later mothers they would be no more dissatisfied. This is not surprising, but it is also
disappointing, for the strong association of well-being with income suggests that early
mothers are not drawing compensating satisfactions from family life and the time they
are spending with their child(ren). We do not actually know how much more or less they
would have enjoyed life in the counterfactual of having had children later, but the
comparison with other, later mothers, suggests that there may be unperceived pitfalls
when having a baby is seen as a fast track to love or adult status.
25