still does not account for all the disadvantage of all ethnic groups. We have not
investigated the possibility of interactions between age at motherhood and ethnic group;
Robson and Berthoud (2003) report these factors to be independent sources of
disadvantage, as we assume here.
Area Effects
Living in an area of child poverty or of concentration of ethnic minority settlement is
associated with early motherhood and a number of economic disadvantages. The value
of a survey such as this one which has information on individual as well as area level
disadvantages is that it is possible to see how far the ecological association is attributable
to composition or context. The regressions presented here suggest that the individual and
ecological terms share explanatory power. On the whole, the area terms have not been
reduced to insignificance by the inclusion of individual information, (the exceptions
being the absence of a significant estimate for mothers’ employment in ‘disadvantaged’
areas or on malaise in ethnic areas when other variables are included). In some cases the
introduction of area terms appears to have moderated the estimates for individual
characteristics such as ethnicity. Though this conclusion requires further investigation, it
does seem as if the stratification part of the design of the survey is tapping some spatial
patterns whereby inequalities are compounded and perpetuated. This paper is largely
descriptive and therefore the analysis undertaken is relatively simple. However further
work could be undertaken to consider these area effects in more detail. For example, a
multilevel model could be fitted to the data to try to directly model the variation by wards
(clusters). Secondly more work could be undertaken to consider the ecological
27