We estimate the heterogeneity in parental separation effects on adolescent well-being in Britain. A novel feature is that we estimate heterogeneous parental separation effects by using the within-between random effects model that accounts for unobserved heterogeneity (fixed effects) but also allows for heterogeneous effects (random slopes). We estimate the distribution of parental separation effects, and whether the heterogeneity in effects can be accounted for by gender and parental education.

Data are from 15 waves (1994-2009) of the youth sample (ages 11 to 16) of the British Household Panel Study. We have reports from 3882 adolescents whose parents divorced when one or more of their children were part of the youth sample. Outcomes are: self-esteem, life satisfaction, and problems at school.

Findings show that parental separation has a heterogeneous negative effect on the psychological well-being of British adolescents. Average effects are just below the usual significance level (p < 0.05). Yet there is considerable heterogeneity (especially for self-esteem and life satisfaction): some experience no or small effects, and others experience major setbacks (> 0.4 s.d.). For self-esteem and problems in school, interaction models between divorce, and gender, mother’s education, and father’s education do not improve model fit. However, the effect of divorce on life satisfaction is negative at father’s average education, and increasingly so at father’s higher educational levels, accounting for 40% of heterogeneity in separation effects.

Most of the heterogeneity escapes the usual socio-demographic predictors of gender and parental education. The heterogeneity might reflect levels of conflict pre-divorce: if divorce ends a high-conflict family life, the result is “relief”. The heterogeneity might also reflect whether divorce was anticipated: effects are especially pernicious when divorce comes unexpected. Finally, some teenagers might simply be more resilient than others.