Research Aims

The study will focus on four inter-related research questions.

1. Who becomes a lone-parent and what are the consequences?

How does lone-parenthood today differ from the past when it was relatively rare? Are lone-parents an increasingly (or decreasingly) "selected" group of the population? How might we expect lone-parents families to fare (in terms of employment, income and poverty) had they not become lone-parents?

2. How long does lone-parenthood last and what is the nature of parents' relationships before and after periods of lone-parenthood?

Most mothers who become lone-parents at the time of their child's birth will have partnered by the time their child is five years old, whilst many married / cohabiting relationships will founder resulting in lone-parenthood for older children. How have these patterns evolved? Do they offer important information for the variation in children's experiences?

3. Are lone-parents becoming more heterogeneous?

Is lone-parenthood becoming increasingly polarized, as has been found in US, with some mothers managing to maintain their incomes through work and maintenance while others fare poorly? Or is lone-parenthood a dominant characteristic leading to poor outcomes for parents and children regardless of background? What are the longer term consequences of lone-parenthood for mothers - does it leave a long term scar even after re-partnering or children leaving home?

4. How does lone-parenthood influence children's outcomes and has this changed over time?

How does family structure, including lone-parenthood and re-partnering (step-parenthood), influence children's outcomes? How does this map onto the patterns in the variation in lone-parenthood described in the preceding questions? To what extent does the background of lone-parents (e.g. age, education) matter in determining children's outcomes? And how does lone-parenthood affect children's social mobility?

Estimation Techniques (Exposure and Outcome Variables)

The analysis will use simple descriptive statistics and panel data techniques to address the research questions set out above. Our first step will be to examine lone-parent status across all data sets. Family status will be defined at the child's birth, at early school age (around age 6), end of primary school (age 11) and end of compulsory schooling (age 16). Re-partnering will be treated as a separate status to intact partnerships from birth. Much of our analysis will focus on lone mothers, who constitute over 90 percent of the lone parent population, although we will also examine lone fathers as a single distinct category where sample sizes allow.

The data will allow us to provide a description of the growth of lone-parenthood over time and across the cohorts. It will also allow us to measure duration of lone-parenthood for children at various ages in the birth cohorts. To do this we will add information of the duration of the relationship prior to the birth to form a typology of lone-parenthood by age of child, duration of lone-parenthood and stability of surrounding relationships. This typology will then be mapped across the cohorts and changes over time will be compared as well as onto child outcomes at ages 6, 11 and 16 to gauge the size of educational deficits at each age for children in or have moved through lone-parent families into re-partnered families. This can also be extended to adult economic and social outcomes, including marriage, fertility and lone-parenthood. So we will assess patterns on social mobility for children growing up in lone-parent families for the early cohorts.

As we are particularly interested in the diversity of experience of lone-parenthood an important question is whether lone-parenthood is more damaging to women's economic position depending on their route into lone-parenthood. We will investigate the influence of education and labour market experience on outcomes for lone-parents; routes into lone-parenthood (including past relationship histories, age of children on becoming a one parent and labour market attachment) and the duration of lone-parenthood. In addition, as paid work has increasingly become the "social norm" for women, has a greater divide developed between lone-parents with strong labour market attachment and earnings and will examine how lone-motherhood has changed in response to increased female labour market opportunities. We will use regression based approaches to condition on observable differences prior to lone-parenthood but it is difficult to also predict what the effects of extremely unhappy relationships would have had if families remained intact. So a number of approaches can be undertaken, each with strengths and weaknesses so as to give as robust a picture as possible. As the data are longitudinal this will allow the use of "fixed effects" estimators, or their equivalents in forms other than linear regression, to further condition out unobservable characteristics of parents and children. Propensity score matching can be used to remove observable differences in socio-economic origins between lone-parent and couple families (including step-parent families), and also to match families by duration of relationships.

The data on children will provide insights into the relationship between the components of our typology and the test scores, so that we can identify which elements of the lone-parent typology appear to lower test scores. We will use regression based approaches to condition on observable differences. In particular we would envisage using propensity score matching to remove observable differences in socio-economic origins between lone-parent and couple families. For those children where lone-parenthood occurs after age 6 or so, and so a pre-lone-parenthood observation of test scores is available a value added model structure will be employed.