1. Report “Family breakdown in the UK”

The conflation of marriage and cohabitation in government statistics – a denial of difference rendered untenable by an analysis of outcomes

UK Bristol Community Family Trust; Harry Benson; September 2006

http://www.socialjusticechallenge.com/uploads/tx_ev3evnews/Family_breakdown_in_the_UK.pdf

The author is grateful to Stephen McKay of Bristol University for sharing his analysis of Millennium Cohort Study data and the trustees of Bristol Community Family Trust for their funding grant. The author also thanks Samantha Callan of Loughborough University and Robert Rowthorn of Cambridge University for their helpful comments and suggestions.

Abstract

Despite a great deal of evidence that marriage benefits and protects adults and children, successive UK governments have eroded and dismantled policy mechanisms that distinguish married from unmarried cohabiting families. Following the abolition of the term “marital status” in 2003, recent government-sponsored family research refers only to “couple parent families”. This combined category conceals significant differences between unmarried and married couple outcomes typically demonstrated by overseas and earlier UK research.

Analysis of data from the Millennium Cohort Study, the most up-to-date large scale UK panel survey of new parents, shows substantial differences in family stability between married and unmarried couples in the early years of parenthood, even after discounting socio-economic factors such as age, income, education and race. Most notably, the difference in family breakdown risk between married and cohabiting couples is sufficient that even the poorest 20% of married couples are more stable than all but the richest 20% of cohabiting couples.

Given the central importance accorded to ensuring optimal outcomes for children in public policy, these findings demonstrate that the lack of distinction between marriage and cohabitation in government policy and research is untenable. Moreover this conflation of terms is at odds with the increasing requirement incumbent upon modern states to be transparent in their functioning and accountable for clearly stated target delivery through the release of comprehensive statistics.

Introduction

In recent years, it has become politically unacceptable to privilege marriage and to treat the institution as anything other than one of several equally acceptable lifestyle choices (e.g. Rowthorn, 2001). As more couples choose to cohabit prior to and, less commonly, as an alternative to getting married, government policy has also reflected this new social norm.

Fiscal policy has long ceased to distinguish married couples from unmarried couples who live together as if married, and more recently, the commissioning of government-sponsored family research has taken the same approach. Married couples have become just one of several couple types – married or unmarried, biological or step-parents – to be viewed as “couple families” or “couple parent families” (Barnes & al, 2004, 2005; Lyon et al, 2006).

Although overseas studies continue to suggest that married families consistently do better than unmarried families on important outcome measures (Carlson, 2006; Manning & Brown, 2006), equivalent UK studies are becoming hard or impossible to find. Using data rebased from Manning & Brown (2006), the chart below illustrates the potential differences to be found when comparing the risk of poverty amongst US children by parent marital status. In this case, the relatively low risk faced by all children with “couple parents” would have concealed the relatively high risk faced by those children with either “cohabiting biological parents” or “cohabiting step-parents”.

Some social commentators and researchers now either dismiss marriage out of hand – “what really matters, at least for the wellbeing of children, is not the matrimonial status of their parents but the ability of two adults to remain together in relative harmony for as long as possible” (Roberts, 2006) – or over-emphasise the rather obvious point that “marriage does not always enhance ones well-being” (Walker, 2000; Glenn & Sylvester, 2005). Others dismiss differences on the evidence of qualitative interviews rather than hard outcomes (Eekelaar & MacLean, 2004). It has to be appreciated however that although analysis of marriage and marital status has received little attention in recent UK outcome research, the distinctions are still present in the underlying data.

Family stability, and its converse, family breakdown, is a simple but compelling measure with which to investigate different outcomes between married and unmarried couples with young babies. The most recent UK analysis of this kind is seven years old and relies on a relatively small dataset (Kiernan, 1999). The aim of this study is to investigate differences in family stability amongst parents of young children, using the Millennium Cohort Study (MCS). The MCS is a robust source of up-to-date, large scale, and publicly available data with which to investigate any differences in outcomes based on family structure, whilst controlling for income and other potential confounds. Data covering the first three years of a child’s life were released in May 2006.

Family trends

Three major family trends have taken place in the UK since 1970.

1. Fewer people are marrying. The number of UK weddings reduced from a peak of 470,000 in 1970 to 306,000 in 2003. Marriage rates per year have declined from 70 to 26 weddings per 1,000 adults. Remarriages have increased from 17% to 40% of all weddings. Civil weddings have increased from 40% to 66% of all weddings. Men and women are also getting married five years older.

2. More people are divorcing. The number of UK divorces increased from 63,000 in 1970 to 167,000 in 2004. Whilst divorce rates per year increased from 4 to 13 divorces per 1,000 marriages during the 1970s and early 1980s, divorce rates have barely changed in the subsequent two decades. Analysis based on ONS data estimates lifetime divorce risk at 45% (Benson, 2005). This figure is likely to be lower for first marriages and higher for remarriages, where annual divorce rates are around 80% higher than for first marriages.

3. More people are cohabiting. The number of UK children born outside marriage increased from 8% of all births in 1970 to 41% in 2003 (Office of National Statistics). In some areas of the country, children born to married parents are now in the minority.

Family policy

Government policy may be both consequence and contributing cause of these changes in family trends. One example is the 1969 Divorce Act. In the years preceding the Act, rising divorce rates increased social pressure on legislators to change the law. In the years immediately subsequent to the act, divorce rates temporarily peaked as pent up demand for divorce was released.

Another example is the Married Couples Allowance. During the Conservatives last period in office (1979-1997), the value of this tax break for married couples was eroded from the equivalent of over 4% of GDP in 1979 to 0.9% of GDP in 1997 (Lindsay et al, 2000). The incoming Labour administration abolished it altogether – except for older couples. Both tax and benefit systems now address couples as “married” or “living together as if married”. In other words, there is no longer any fiscal distinction between married and unmarried couples. A neutral fiscal policy on marriage reflects what appears to be the political view that couples should not be distinguished by their marital status.

Although this policy is neutral in terms of immediate financial consequences, policy does continue to favour marriage in terms of longer term financial consequences. Inheritance tax, transferable allowances and pension rights still provide advantages to married couples. The introduction of civil partnerships has extended these advantages to same-sex couples.

Research policy

As a knock-on effect of the neutral fiscal policy towards married and unmarried couples, the government decided in 2003 the term “marital status” would no longer appear on government forms. This policy was announced in a government paper summarising responses to the consultation on pending civil partnership legislation (Smith, 2003).

This change in policy has influenced the commissioning of new research by government departments. For example, the Families and Children Study, commissioned by the Department of Work and Pensions, changed the way it looked at family outcomes. Prior to 2003, FACS research distinguished family outcomes according to a variety of family structures, including marriage (Marsh & Perry, 2003). After 2004, FACS research refers more narrowly only to “couple parent” families and “lone parent” families (Barnes et al, 2004, 2005; Lyon et al, 2006). The commissioning researcher at DWP has clarified that this change was in line with tax and benefit policy, which does not distinguish between married and unmarried couples (personal email communication, Elizabeth Rayner, September 2005). Other government sponsored publications – e.g. Social Trends, Family Resources Survey, Labour Force Review – also refer extensively to ‘couples’ without differentiation. Where marriage is distinguished, it involves population data rather than an analysis of outcomes – e.g. Population Trends.

Marriage benefits and protections

Nevertheless, it remains well known that being married is consistently associated with a range of better outcomes for both adults and children (Waite & Gallagher, 2000). Conversely, family breakdown is consistently associated with a range of poorer outcomes for both adults and children (Brown, 2004, McLanahan & Sandefur, 1994). Much of this latter evidence suggests that family breakdown causes these poorer outcomes. For example, family breakdown leads to increased risks of poverty, crime, health problems and family breakdown amongst both children and grandchildren (Amato, 2000).

The benefits and protections of marriage are often attributed to selection rather than cause – i.e. people who do better get married. This is undoubtedly true in part. For example, those less educated or on lower income are less likely to marry in the first place and more likely to divorce if they do marry (Bumpass & Sweet, 1989; Ermisch & Francesconi, 2000; Kiernan, 2003). Policy makers sometimes conclude from this that differences in family structure can be accounted for exclusively by selection.

However, a review by Wilson and Oswald (2005) lists 23 longitudinal studies that provide compelling evidence of a causal link between marriage and health, mental health and longevity. The authors conclude that “the size of the health gain is remarkable. It may be as much as the benefit from giving up smoking”. Additional studies also illustrate how marriage – but not cohabitation – improves well-being (Lamb & al, 2003), relationship quality (Kamp Dush & al, 2003) and relationship stability (Marsh & Perry, 2003).

Although most of these longitudinal studies were conducted amongst US population samples, UK and other European studies do exist. Of particular relevance is an analysis in the FACS study (Marsh & Perry, 2003) showing how the risk of family breakdown amongst low-income families is increased significantly where couples were unmarried.

Family breakdown and prevention

Family breakdown exacts emotional and financial costs on families and society. A report produced for the Lords and Commons Family and Child Protection Group (Lindsay et al, 2000) calculated that the direct annual cost of family breakdown to the taxpayer exceeds £15 billion. The majority of this huge bill represents the excess of income support for single parent families that might reasonably be attributed to family breakdown.

Factoring in a conservative combination of inflation, an 8% increase in lone parent family formation, 19% rise in income support and 41% rise in lone parent premium (Office of National Statistics) the current cost to the taxpayer of family breakdown is now likely to be £20-£24 billion. This equates to an average contribution of £680-820 per taxpayer per year.

Given the centrality of child outcomes to government policy – “Every Child Matters” – it might be assumed that an appropriate focus for government would be to reduce and prevent family breakdown. This does not appear to be the case. The former £5m Marriage and Relationships Support (MARS) fund – now absorbed into the £17m Children, Young People and Families Fund (CYPF) – used to be the main government vehicle for funding couple support until 2003. In 2006-7, the MARS component was cut to under £4m (Percival, 2006), equivalent to 15p per taxpayer per year. Even if the entire CYPF fund could be described as support for voluntary sector programmes that prevent or reduce family breakdown, this still only equates to 58p per taxpayer per year.

Therefore not only does government appear to take no cognisance of the distinctiveness of marriage – the family structure category most associated with a wide range of positive family outcomes – it also contributes very little to support couples and prevent family breakdown. Compared with the billions of taxpayers’ money spent on dealing with the effects of family breakdown, considerably less than 0.1% of this sum is spent on trying to prevent it from happening in the first place.

Family structure and family breakdown

Government policy may now exclude comparisons of married and unmarried families in government-sponsored UK longitudinal studies. However is still possible to conduct such an analysis because the underlying data still exists. It is therefore possible to establish, using large scale up-to-date UK datasets such as FACS or the MCS, the validity or otherwise of the government’s neutral fiscal policy towards marriage.

The most recent UK analysis of this kind was published seven years ago based on a relatively small dataset. This study found that 43% of unmarried parents and 8% of married parents had split up before their child’s fifth birthday (Kiernan, 1999). The absence of subsequent research raises valid questions about the robustness and relevance of these findings today.

The aim of this study is to explore how family structure at birth influences subsequent family stability using a large scale up-to-date government-sponsored UK cohort study. Replicated findings based on more robust evidence would call into question the wisdom of ignoring marriage in both government policy and research.

Method

The Millennium Cohort Study is a large scale longitudinal birth cohort study conducted within the four countries of the United Kingdom. The survey contains a wide range of information about 18,819 babies and their parents in 18,553 families. Parents of babies born between September 2000 and January 2002 were interviewed for the first sweep when their babies were 9 months old and for the second sweep when their babies were 3 years old.

Data was obtained from 15,119 parents during the second sweep. Although response rates were a relatively high 81%, data was unweighted and thus did not take into account any differential rates of attrition.

The key independent variable in this study was family status at the birth of the child. Parents described their status as “married and living together”, “cohabiting/living as married”, “closely involved”, “just friends”, “separated”, “divorced” or “not in any relationship”. This study looked at outcomes both for couples who were explicitly “cohabiting” as well as for couples who were “unmarried”, which included couples who were either “cohabiting” or “closely involved”.

Independent socio-economic variables – such as parental income, employment status, and education levels – were taken either at 9 months or birth as the data allowed.

The key dependent variable was partner status when the child was 3 years old. Data was derived to show status as “same person is partner”, “new partner” or “no partner”. This study assumed that family breakdown had taken place in either of the latter two categories.

Results

Frequency and regression analyses were conducted on the risk of family breakdown by marital status, by socio-economic group, and by each group independent of the others.

Table 1 shows the distribution of all families experiencing breakdown over the first 3 years of a child’s life, based on parents’ marital status at birth. Amongst the entire sample of 15,119 parents, 2,966 experienced family breakdown, an overall risk of 20%. However, the risk of family breakdown during the first 3 years of a child’s life varies greatly depending on marital status.

· Married couples represent 63% of the sample at birth but only 18% of all family breakdown, an overall risk of 6%.

· Unmarried couples – combining those “cohabiting” or “closely involved” – represent 33% of the sample at birth yet 50% of all family breakdown, an overall risk of 32%.

· Cohabiting couples on their own represent 24% of the sample at birth and 25% of all family breakdown, an overall risk of 20%.

· Closely involved couples represent 6% of the sample at birth yet 25% of all family breakdown, an overall risk of 76%.

· Amongst all other categories, family breakdown approaches 100%.

The overall risk of family breakdown is therefore substantially higher amongst all family types compared to married couples.

· The risk is 5.5 times greater for all unmarried couples, 3.5 times for cohabiting couples, and 13.3 times for closely involved couples.

Table 1	Married couple	Unmarried couple	Cohabiting couple	Closely involved	Just friends	Other	Total
Parent status at birth	9487	4629	3672	957	300	703	15119
% of total	63%	33%	24%	6%	2%	5%	100%
Family breakdown	545	1471	742	729	288	662	2966
% of total	18%	50%	25%	25%	10%	22%	100%
Family breakdown risk	6%	32%	20%	76%	96%	94%	20%
Risk vs married couple	1.0	5.5	3.5	13.3	16.7	16.4	3.4

Figure 1 presents the same data in terms of those who were couples at the time of the child’s birth – i.e. excluding parents who were “just friends”, “separated”, “divorced” or “not in a relationship” – comparing the percentage shares of the total sample and those who split up. “Unmarried couple” comprises couples whether “cohabiting” or “closely involved”.

· In this case, married couples represent 67% of all couples but only 27% of family breakdown.

· In contrast, unmarried couples represent 33% of couples and yet 73% of family breakdown.

Figure 2 illustrates the differences in family breakdown risk by couple type. Unmarried couples are more likely to split up than married couples.

Table 2 compares the distribution of population and family breakdown depending on whether the child is the first or subsequent child. Married parents represent a bigger proportion of parents having their second child compared to those having a first child – 68% vs. 55%. They also represent a bigger share of family breakdown – 25% vs. 11% of all family breakdown.

Despite this, family breakdown risks remain similar regardless of whether the baby is the first or subsequent child. For married parents, the risk is 5% for a first child and 6% for subsequent children. For unmarried parents, the risk is 33% for a first child and 31% for subsequent children. The relative risk of family breakdown for unmarried couples is therefore higher for those having a first child at 7.2 times vs. 4.8 times compared to married couples.

Table 2	Married couple	Unmarried couple	Cohabiting couple	Closely involved	Just friends	Other	Total
% of population
First child	55%	37%	29%	8%	2%	6%	6,320
Subsequent child	68%	26%	21%	5%	2%	4%	8,799
% of family breakdown
First child	11%	54%	29%	25%	10%	25%	100%
Subsequent child	25%	45%	21%	24%	10%	20%	100%
Family breakdown risk
First child	5%	33%	22%	74%	95%	96%	22%
Subsequent child	6%	31%	18%	78%	97%	94%	18%
Risk vs married couple
First child	1.0	7.2	4.9	16.4	20.9	21.1	4.9
Subsequent child	1.0	4.8	2.8	12.1	15.0	14.6	2.7

Follow-up analysis of those “closely involved” (not shown) shows this group tend to be younger and less well-educated than other cohabitees. Around half were having their second or subsequent child, the same proportion found amongst cohabitees.

Figures 3 and 4 show the distribution of break-up and actual break-up risk over 3 years depending on whether the baby is the first or subsequent child. Unmarried couples account for 83% of family breakdown amongst couples having their first child but 65% of family breakdown amongst couples having a second or subsequent child. However the absolute risk of family breakdown remains similar for both married couples and unmarried couples, regardless of whether having their first or subsequent child. For married couples, the risk is 5% and 6% respectively. For unmarried couples, the risk is 33% and 31% respectively.

Table 3 shows family breakdown rates amongst couples who were either married or cohabiting when their baby was 9 months old, depending on income. “Income” in this case represents total family income from both work and benefits or tax credits divided into approximate quintiles.

Family breakdown rates over this slightly shorter duration – around 2 years and 3 months – reduce as income increases for both married and cohabiting couples. For married couples, the risk reduces from 8% on low income to 3% on high income. For cohabiting couples, the risk reduces from 23% on low income to 7% on high income. Across every income group, cohabiting couples are at least twice as likely to split up compared to married couples.

Table 3	<£15.6k	<£20.8k	<£31.2k	<£41.6k	>£41.6k	Total
Status at 9 months
Married	1,613	1,545	2,498	1,403	1,656	8,715
Cohabiting	1,129	667	754	356	292	3,198
Family breakdown
Married	126	91	115	37	51	420
Cohabiting	260	83	76	31	20	470
Family breakdown risk
Married	8%	6%	5%	3%	3%	5%
Cohabiting	23%	12%	10%	9%	7%	15%
Cohabiting vs married	2.9	2.1	2.2	3.3	2.2	3.0

Figure 5 illustrates this graphically, showing how both income and marital status independently influence family breakdown risk. The difference in family breakdown risk between married and cohabiting couples is sufficient that even the poorest 20% of married couples are more stable than all but the richest 20% of cohabiting couples.

Figure 6 illustrates a similar pattern for age (data not shown). 95% of married and 91% of cohabiting mothers give birth in their 20s and 30s. Although the risk of family breakdown reduces with age, risk also varies depending on marital status. Even younger married mothers are more stable than older cohabiting mothers.

Finally, a regression analysis was conducted to assess the relative importance of marital status and other socio-economic factors. Data for this analysis included married and cohabiting couples only, not those “closely involved” or in other categories.

Table 4 shows how marital status, age, income education, ethnic group and welfare each independently and significantly influence the risk of family breakdown. Wald numbers suggest marital status and age are more important than income, education, race or welfare.

· Marital status. The odds of a cohabiting couple with a young child splitting up are more than twice that of a married couple of equivalent age, income, education, ethnic group and benefits.

· Age. The odds of a couple in their teens and 20s splitting up are twice that of a couple in their 30s, independent of other factors.

· Education. The odds of couples with less education splitting are higher that for those with more education, although the relationship between risk and education level is not entirely linear.

· Income. The odds of a couple with the lowest family income – less than £15,600 – splitting up are 44% more than that of couples. However rising income does not appear to be a protective factor above this level.

· Ethnic group. The odds of black mothers splitting up are twice those of white mothers, independent of other factors. Asian mothers are most likely to stay together.

· Welfare. The odds of splitting up are 33% higher for those on benefits.

· Birth order. Whether the child is the first or subsequent birth is not a factor.

Table 4: Variables		B	S.E.	Wald	df	Significance	Odds ratio
Marital status (vs married)	Cohabiting	0.78	0.08	101	1	0.000	2.17
Mother's age (vs. 30s)				94	3	0.000
	Teens	1.30	0.18	52	1	0.000	3.68
	20s	0.68	0.08	72	1	0.000	1.98
	40s	-0.06	0.23	0	1	0.789	0.94
Education (vs. NVQ4)				30	6	0.000
	No quals	0.60	0.14	19	1	0.000	1.82
	NVQ 1	0.25	0.15	3	1	0.092	1.28
	NVQ2	0.45	0.10	21	1	0.000	1.57
	NVQ3	0.31	0.12	7	1	0.008	1.36
	NVQ5	0.04	0.24	0	1	0.875	1.04
	Other quals	-0.09	0.30	0	1	0.777	0.92
Family income (vs. mid quintile)				24	6	0.001
	Bottom 20%	0.37	0.11	11	1	0.001	1.44
	Next 20%	0.03	0.11	0	1	0.788	1.03
	Penultimate 20%	-0.26	0.15	3	1	0.082	0.77
	Top 20%	-0.13	0.15	1	1	0.398	0.88
	Don't know	0.19	0.17	1	1	0.261	1.21
	Refused	0.15	0.27	0	1	0.584	1.16
Ethnic group (vs. White)				22	6	0.001
	Indian	-0.54	0.35	2	1	0.126	0.58
	Pakistani	-0.46	0.23	4	1	0.047	0.63
	Bangladeshi	-0.61	0.43	2	1	0.155	0.55
	Black	0.64	0.25	7	1	0.009	1.90
	Mixed / other	0.51	0.23	5	1	0.026	1.66
	Other / unknown	-1.06	1.00	1	1	0.289	0.35
Welfare (vs. not on benefits)	Receives	0.29	0.09	10	1	0.001	1.33
Birth order (vs subsequent child)	First birth	-0.10	0.08	2	1	0.181	0.90
	Constant	-3.56	0.11	1025	1	0.000	0.03
MCS analysis supplied by Stephen McKay of Bristol University

Discussion

This study investigates whether government policy and research are right to dismiss or ignore marriage as a distinct social category with qualitatively different outcomes from other cohabiting family structures. Couples who “live together as if married” may appear to be comparable to couples who are legally married, in that they both live together and have children. But are they really comparable in terms of family outcomes?

Although research in other countries may show outcome differences between married and unmarried families, the relative absence of recent UK research – reinforced by government policy to abolish the term “marital status” – make it unclear whether such findings still apply in the UK.

However our new analysis of Millennium Cohort Study (MCS) data on 15,000 British mothers who gave birth in 2000 or 2001 shows clearly that married families continue to provide significantly more stable homes for their children than do unmarried families.

This robust finding, using a large scale up-to-date dataset, questions the validity and wisdom of recent government policy to treat married and unmarried couples alike, and abolish marital status from government forms. Gliding over any distinction between couple types rules out analyses based on diverse family structures which could have important policy implications. MCS data shows that during the first three years of a child’s life, the risk of family breakdown faced by unmarried parents is 5.5 times greater than that faced by married parents. Amongst unmarried parents who describe themselves as “cohabiting”, the risk is 3.5 times greater. Amongst those who describe themselves as “closely involved”, the risk is 13 times greater. The differential risks associated with family status are broadly similar regardless of whether the child is the first or subsequent birth.

One in three unmarried couple parents – including one in five of those who describe themselves as “cohabiting” – will split up before their child’s third birthday compared to one in seventeen married parents. These figures are similar to those found by Kiernan in 1999. Social trends such as the move towards the separation of marriage and childbearing (Kiernan et al, 1998) may have become more pronounced since that study, but despite its increasingly normative character, the instability associated with cohabitation remains high.

Importantly, this study also shows that income does not account for differences in stability between married and cohabiting couple parents. Whereas the ratio between unmarried and married parent break-up rates is 3:1, the ratio between couples on similar income is still at least 2:1.

Our regression analysis shows that family status and age are more important predictors of family breakdown than either income or education. Moreover, only those on low income are at significantly higher risk, independent of other factors. Government policy aimed at increasing income levels above a threshold of about £15,000 p.a. might reasonably be expected to reduce family breakdown to some extent. But it will not reduce the differential in breakdown risk faced by cohabiting couples compared to married couples. The odds of cohabiting couples splitting up are more than double those of married couples, even after taking age, education, income, ethnic group and benefits into account.

Using a robust dataset, this study therefore replicates the claim that three quarters of family breakdown affecting young children now involves unmarried parents (Benson, 2005). In terms of hard outcomes, these conclusions question the appropriateness of policy-makers and researchers considering marriage and cohabitation as equivalent or even perhaps describing cohabiting couples as “living together as if married”.

Strengths and limitations of the study

The major strength of MCS is that it allows a clean analysis of a large up-to-date dataset of comparable families. The study is “clean” in being able to analyse data from mothers with children of the same age. There are therefore no potential confounds from changing social norms. It is also advantageous to be able to investigate outcomes amongst only those mothers who had their first child, thus discounting any potential sibling effects.

Further investigation is needed to find out why couples described themselves as “closely involved”, implying being a couple, rather than “cohabiting”. Family breakdown risk is especially high amongst the former category, part of which may be due to being younger and less well-educated. It is also possible that some mothers in this category may be “living apart together” (Haskey, 2005), potentially claiming additional lone parent benefits whilst not wishing to admit publicly to being a couple. Recent evidence suggests there appear to be more claimants of lone parent benefits than there are lone parents (Brewer & Shaw, 2006). Further research is needed to establish why those “closely involved” are so unstable and whether this self-description is influenced by welfare policy.

The most obvious limitation of this study is that it only covers outcomes over the first three years of a child’s life. Future MCS sweeps will allow analysis of family outcomes over longer periods. Other datasets – ALSPAC, FACS – already have the potential for such analysis.

This study also concentrated only on change in family structure from birth. Subsequent changes make analysis more complex but may have important consequences. For example, Kiernan (1999) found that family breakdown risk was lower amongst cohabiting couples who subsequently married compared to those who stayed unmarried.

What is it about marriage?

Socio-economic selection effects undoubtedly account for part of the apparent benefits and protections afforded to married couples. But, as Wilson & Oswald (2005) demonstrate in their review of longitudinal studies, marriage brings with it a causal component that is not accounted for by socio-economic background factors.

Although it is not the aim of this study to explain precisely why marriage makes such a difference, it is worth highlighting plausible explanations worthy of further UK research. Commitment. The simplest explanation is that married couples have a higher level of commitment to one another compared to unmarried couples in the first place. This does not have to be true for all unmarried couples, amongst whom a continuum of commitment exists (Smart & Stevens, 1997). However the decision to move in together for an unspecified period of time generally represents a lower barrier-to-entry than the decision to get married for life. Having moved in together, the risk of pregnancy is similar for all couples, whether married or not (Ermisch, 2001). The increasing social norm to cohabit first and marry later also increases the likelihood of unmarried childbirth (Ermisch, 2006).

For married couples, the time involved in bringing up a child fits with the intention to spend a life together. For unmarried couples, the prospect of bringing up a child may set a time horizon beyond the expectation or intention of the relationship. Coming to terms with these long-term consequences may be too much for some couples to resolve.

A compelling new theory also suggests that men and women tend to see commitment in different ways. Whereas women view commitment in terms of attachment – moving in together – men view commitment in terms of a decision – getting married (Stanley & al, 2005). This gender difference in relationship intentions has the potential for considerable misunderstanding.

Communication skills. Resolving intentions may not be helped by the likelihood that couples who cohabit rather than marry may have poorer communication skills (Cohan & Kleinbaum, 2002). Whether or not couples who cohabit start off with this vulnerability, couples who are less committed to their relationship may put less effort into the development of the skills necessary to sustain a long-term relationship.

Father involvement. The willingness and ability of fathers to be involved in the lives of their families also plays a major role in family outcomes. Some studies have shown that the level of father involvement can explain much of the difference in outcomes between the children of married and unmarried families. For example, teenagers with involved fathers were less likely to behave badly regardless whether parents were married, unmarried or single (Carlson, 2006).

Interestingly, mother behaviours towards either parent-child or parent-parent relationship appear less predictive of both child and marital outcomes. Just as Carlson (2006) found that father – but not mother – involvement is a key predictor of teenage behavioural problems, Whitton & al (2002) found that father – but not mother – willingness to sacrifice predicts marital commitment.

Behaviour. There are behavioural differences displayed by married families compared to unmarried families, both before and after the birth of their child. For example, unmarried mothers are less likely to attend ante-natal clinics. A preliminary analysis of MCS data for this study found that 82% of married mothers attended compared to only 64% of those “cohabiting” and 40% of those “closely involved”.

Specialisation and personal autonomy According to Becker (1981), specialisation in household roles is a more efficient arrangement than simply sharing roles. Specialisation allows one spouse to master some skills or responsibilities – such as tax returns – and to relinquish others – such as children’s clothing. Married couples are more likely to specialise household roles compared to unmarried couples, partly due to the length of relationship (Stratton, 2005).

Married couples, especially first time couples, are also more likely than unmarried couples to use joint rather than separate bank accounts (Heimdal & Houseknecht, 2003). For couples to specialise their household roles and to manage their finances in joint name require a focus on “us” rather than “you” and “me”. Both of these findings fit with other research showing that cohabitors are more likely to value their personal autonomy and equity (Clarkberg & al, 1995).

References

· Amato, P. (2000). The consequences of divorce for adults and children. Journal of Marriage and Family, 62, 1269-1287.

· Barnes, M., Willitts, M., Anderson, T., Chaplin, J., Collins, D., Groben, S., Morris, S., Noble, J., Phillips, M., & Sneade, I. (2004). Families and children in Britain: Findings from the 2002 Families and Children Study. Department of Work and Pensions, Research Report No 206.

· Barnes, M., Lyon, N., Morris, S., Robinson, V., & Yee, W.Y (2005). Family life in Britain: Findings from the 2003 Families and Children Study. Department of Work and Pensions, Research Report No 250.

· Becker, G. (1981) A treatise on the family. Cambridge, Mass: Harvard University Press.

· Benson, H. (2005) What interventions strengthen family relationships: A review of the evidence. Paper presented at 2nd National conference on relationships education. London

· Brewer, M. & Shaw, J. (2006). How many lone parents are receiving tax credits? IFS briefing note No 70, Institute for Fiscal Studies.

· Brown, S. (2004). Family Structure and Child Well-Being: The Significance of Parental Cohabitation. Journal of Marriage and Family, 66, 351-367.

· Bumpass, L. & Sweet, J. (1989). National estimates of cohabitation. Demography, 26, 615-625.

· Carlson, M. (2006) Family structure, father involvement, and adolescent behaviour outcomes. Journal of Marriage and Family, 68, 137-154.

· Clarkberg, M., Stolzenberg, R., & Waite, L. (1995) Attitudes, values, and entrance into cohabitational versus marital unions. Social Forces, 74, 609-632

· Cohan, C. & Kleinbaum, S. (2002) Toward a greater understanding of the cohabitation effect: Premarital cohabitation and marital communication. Journal of Marriage and Family, 64, 180-192.

· Eekelaar, J. & MacLean, M. (2004) Marriage and the Moral Bases of Personal Relationships. Journal of Law and Society, 31, 510-538.

· Ermisch, J., & Francesconi, M. (2000). Cohabitation in Great Britain: Not for long, but here to stay. Journal of the Royal Statistical Society (A, 163, Part 2), 153–171.

· Ermisch, J. (2001). When forever is no more: Economic implications of changing family structure. University of Essex: IESR working paper.

· Ermisch, J. (2006) An economic history of bastardy in England and Wales. ISER working paper 2006-15. Colchester: University of Essex.

· Glenn, N. & Sylvester, T (2005). The denial: Downplaying the consequences of family structure for children. Institute for American Values.

· Haskey, J. (2005). Living arrangement in contemporary Britain: Having a partner who usually lives elsewhere and Living Apart Together (LAT). Population Trends, 122, 35-45.

· Heimdal, K.R. & Houseknecht, S.K. (2003). Cohabiting & married couples' income organisation: Approaches in Sweden & the United States. Journal of Marriage and Family, 65, 525-538.

· Kamp Dush, C., Cohan, C., & Amato, P. (2003). The Relationship Between Cohabitation and Marital Quality and Stability: Change Across Cohorts? Journal of Marriage and Family, 65, 539-549.

· Kiernan K., Land H. & Lewis J., (1998). Lone Motherhood in Twentieth-Century Britain. Oxford: Oxford University Press

· Kiernan, K. (1999). Childbearing outside marriage in Western Europe. Population Trends, 98, 11-20.

· Kiernan, K. (2003) Unmarried parenthood: new insights from the Millennium Cohort Study. Population Trends, 114, 26-33.

· Lamb, K., Lee, G., & De Maris, A. (2003). Union Formation and Depression: Selection and Relationship Effects. Journal of Marriage and Family, 65, 953-962.

· Lindsay, D. et al. (2000). The cost of family breakdown. Bedford: Family Matters Institute.

· Lyon, N., Barnes, M, & Sweiry, D. (2006) Families with children in Britain: Findings from the 2004 Families and Children Study (FACS). Department of Work and Pensions Research Report No 340.

· Manning, W. & Brown, S. (2006) Children's Economic Well-Being in Married and Cohabiting Parent Families. Journal of Marriage and Family, 68, 345-362.

· Marsh, A. & Perry, J. (2003). Family change 1999 to 2001. DWP research no 181. CDS: Leeds.

· McLanahan, S. & Sandefur, G. (1994). Growing Up With a Single Parent: What Hurts, What Helps. Cambridge: Harvard University Press.

· Percival, D. (2006) DfES grant funding. 2-in-2-1 newsletter, no 6.29.

· Roberts, Y. (2006) What’s half nothing? Guardian, May 31^st

· Rowthorn, R. (2001) For marriage. Prospect magazine, April issue

· Smart, C. & Stevens, P. (2000) Cohabitation Breakdown. London: Family Policy Studies Centre.

· Smith, J. (2003) Responses to civil partnerships. Women and Equality Unit, Department of Trade and Industry.

· Stanley, S., Kline, G., & Markman, H. (2005). The inertia hypothesis: Sliding vs deciding in the development of risk for couples in marriage. Paper presented at the Cohabitation: Advancing research and theory conference, Bowling green, OH.

· Stratton, L. (2005). The degree of intrahousehold specialization in housework and how specialization varies across couple households. Paper presented at June 2005 SOLE meetings, Bonn.

· Waite, L., & Gallagher, M. (2000). The case for marriage. New York: Doubleday.

· Walker, A. (2000) Refracted knowledge: viewing families through the prism of social science. Journal of Marriage and Family, 62, 595-608.

· Whitton, S. W., Stanley, S. M., & Markman, H. J. (2002). Sacrifice in romantic relationships: An exploration of relevant research and theory. In H. T. Reiss, M. A. Fitzpatrick, A. L. Vangelisti (Eds), Stability and Change in Relationship Behavior across the Lifespan (pp. 156-181). Cambridge: Cambridge University Press.

· Wilson, C. & Oswald, A. (2005) How does marriage affect physical and psychological health? A survey of the longitudinal evidence. Unpublished manuscript, University of East Anglia and Warwick University.