variables (on region and area type) can be included as antecedent variables, particularly
for those who have not moved since birth. As we treat the dependent variable as an
ordered discrete variable, we use an ordered probit estimation.
Strand two attempts to consider some of the possible associations between adult
outcomes at the time of the cohort child being nine months old and age at motherhood. It
is worth noting that the cohort child may not be the first child in the family and therefore
some of these mothers may have entered motherhood at some years prior to the
interview. The model will attempt to unpack some of the possible consequences of
entering motherhood at various points in the life course. Strand two estimates the
following equation:
outcome = α+ βagemoth + δantecedent + γcontemporary +ν
where outcome is one of eight adult outcomes considered including the presence of a
partner, equivalised household income and mother’s life satisfaction. As before, α is the
constant, agemoth is the age at first ever motherhood grouped in three year age groups,
antecedent are those variables which are determined before the birth of their first child, ν
is the residual while in addition contemporary are subsequent and contemporary
covariates. The antecedent variables are as described in strand one while the
contemporary variables include qualifications, which may have been gained after
compulsory school leaving, region of residence, and longstanding illness (in some case a
fourth model includes details of a partner’s qualifications and health, or the level of
equivalised net family income). Once again either an ordered probit or probit estimation
method is used. Age at first motherhood is unlikely to have a linear impact on many of
13