Harry Benson's new study of family breakdown (available for download here ) is based on data from the Millennium Cohort Study of 15,000 mothers with three year old children. All of the mothers gave birth between mid 2000 and early 2002. Information was gathered in two waves – when the child was 9 months and again at 3 years. The size of the sample makes this the largest survey of family breakdown yet conducted in the UK .

The main finding of the study is that married parents were much less likely to split up compared to cohabiting parents or any other category of parent. In absolute terms, 32% of unmarried couples split up compared to 6% of married couples. Unmarried couples comprised mothers who described themselves earlier as “cohabiting” or “closely involved”. Within three years, 20% of the “cohabiting” couples had split compared to 76% of the “closely involved” couples. So compared to married couples, cohabiting couples were 3.5 times more likely to split whereas unmarried couples as a whole were 5.5 times more likely to split.

These numbers confirm that three quarters of family breakdown amongst parents with young children involves the separation of unmarried couples.

The study also sets out to establish whether income and other factors are the real culprits behind family breakdown, not whether couples are married or not, just as many politicians and commentators often assume. Regression analysis of married and cohabiting couples – that allows each factor to be analysed independently of the others – found that age, income, education, ethnic group, benefits receipt all made independent contributions to the risk of family breakdown.

For example, mothers in their 20s were twice as likely to split as mothers in their 30s, all other factors being equal. Mothers with no qualifications were 82% more likely to split compared to mothers with NVQ level 4 or equivalent,. Black mothers were twice as likely, and Pakistani mothers less likely, to split compared to white mothers. Birth order, whether the baby was the first or subsequent, did not appear to increase the odds of family breakdown.

However the headline finding was that marital status topped all of these factors in terms of importance. Cohabiting mothers were more than twice as likely to split compared to married mothers, even after all these other factors had been taken into account. There is clearly therefore a functional difference that increases the vulnerability of cohabiting couples and stability of married couples that cannot be explained by socio-economic background alone.

Of particular note is that the poorest 20% of married couples did better than all but the richest 20% of cohabiting couples. This finding illustrates how policy-makers, cushioned by wealth, may be misled by their relatively positive exposure to cohabitation which is wholly unrepresentative of most people's reality.

Whether the marriage/cohabitation effect is about the type of people or the type of relationship they choose is perhaps a moot point. There is evidence that both explanations have validity. The main policy outcome of this study is to show that it is no longer tenable to claim that cohabiting couples live together “as if married”.

Government policy and research must therefore distinguish family structure by marital status. How it does that, of course, is the next issue.

Reference: Benson, H. (2006) The conflation of marriage and cohabitation in government statistics – a denial of difference rendered untenable by an analysis of outcomes. Bristol Community Family Trust.

Link to this article here.

The full paper here.