these covariates and so is entered as seven dummy variables of three year age groups.
The omitted category is those who entered motherhood up to the age of 18.
For each of the eight outcomes at least 3 models are estimated. Model one includes just
the age at motherhood terms to establish an unadjusted age profile, model two includes
the antecedent variables and models three (and in some cases four) also include
subsequent and contemporary variables. All of the regressions allow for the weighting
clustering and stratification of the sample design.
RESULTS
Strand one: Analysis of age of motherhood
Table 6 analyses age at motherhood in terms of our measure of antecedent circumstances,
and for the women having their first child some measures of their current location. The
first panel shows that age at first motherhood is associated negatively with family
breakdown in the previous generation, with having been in care, with leaving school
at/before the compulsory school leaving age and with all ethnic groups other than White
and ‘Other’. Bangladeshis and Pakistanis have the earliest births. Similar results are
obtained if we restrict the sample to those who have given birth for the first time in 2000-
1 (but those from the Black Caribbean and Black African groups are no longer
significantly earlier than the Whites). Extending this model on the sub sample to include
information on current location (panel 3) shows a significant association of early
motherhood with living in disadvantaged and ethnic areas. Most of the negative ethnic
effects disappear in this extended model except those for Pakistanis and Bangladeshis,
14