Cohabitation and After

Measuring Cohabitation using a British Longitudinal Study

The following results have been collected by CeLSIUS. CeLSIUS is the “Centre for Longitudinal Study Information and User Support“

CeLSIUS is one of the services funded by the ESRC Census Programmes and as such its services are provided free of charge to staff and students of UK academic institutions.

 Source:  and


Marital status has long been recognised as a fundamental socio-demographic characteristic associated with many important behaviours and outcomes; William Farr, one of the first Registrar Generals, noted in 1858 that ‘marriage is a healthy estate’. Married men have been shown to be healthier than single men; married women are less likely to be employed than single women; widows are economically less well off than married women; and so on.

However, since the 1970s there have been ‘considerable changes … amounting to a structural shift in individuals’ demographic behaviour and societal norms’ (Haskey, 2001, Haskey J (2001) Cohabitation in Great Britain: past, present and future trends – and attitudes [Population Trends 103, pp 4-25 ] and among these are increases in divorce and in cohabitation, that is, in couples who live together in intimate relationships without being legally married. These changes are challenging the validity of marital status as a major predictive variable; today’s married man may be healthier than his single peer, but tomorrow his life and habits may be disrupted by divorce. Marital status, measured at one point in time, may no longer have the power to represent a lifetime’s engagement with a partner and family.

These tables examine the rising phenomenon of cohabitation (in this study, people in same-sex cohabiting relationships are not selected for the sample). In the past its prevalence was difficult to assess as social disapproval led to reticence on the part of the cohabiting couple and therefore to caution on the part of survey and census interviewers. However, since the 1970s increasingly straightforward questions have been asked and responses have been elicited. The 2001 Census found that 8.3% of all households were composed of a cohabiting couple, with or without children; and in the 2006 General Household Survey, 13% of people aged 16 to 59 reported themselves as cohabiting.

These substantial proportions raise the question for social scientists of how cohabitation compares with legal marriage as a predictor for other characteristics. Haskey (2001), using cross-sectional and retrospective data in the General Household Surveys of 1998 and earlier, showed that:

  • current cohabitations (i.e. at the time of the survey) were of much shorter average duration (so far) than current marriages;
  • younger people were more likely than older people to be cohabiting;
  • the proportion of people who had at some time cohabited without subsequently marrying that partner was rising fast; and
  • a large and increasing proportion of married couples had cohabited together before marriage.

It has long been observed that cohabitation is not a homogeneous phenomenon. Haskey (2001) used his own findings (above) and those of others on trends in cohabitation and attitudes to it, and suggested that the majority of people cohabiting did so with the intention of forming a life-long union if the relationship continued to be satisfactory, and that many hoped to marry in the future. On the other hand, some people – particularly those who did not plan to have children – cohabited because it suited them in the present and saw no need to make plans for the future. Further study of the duration of cohabitation and its outcome (legal marriage, separation or death) in combination with other characteristics would help to distinguish between different types of cohabitation; these in turn could be compared (in their cross-sectional form) with other marital states to refine the traditional classificatory variable of marital status.

Marriage and cohabitation histories from the General Household Survey or other sources have the disadvantage of being cross-sectional: we can only examine the cohabiting partners by their characteristics at one point in time. The retrospective data which they include suffer from the problems of recall; these are known to be serious with an informal event like the beginning or end of a cohabitation (Lilly, 2000). Longitudinal data from the British Household Panel Survey have been examined (Ermisch & Francesconi, 2000) but are hampered by relatively small numbers of people cohabiting.

Using the ONS Longitudinal Study we benefit from a 1% sample of the population, and can examine characteristics of the cohabiting partners at two points in time, by using successive Censuses. Here we have examined people who reported themselves as cohabiting in the 1991 Census by their partnership status at the 2001 Census. This 2001 outcome includes the categories:

  • respondent is still cohabiting with the 1991 partner
  • respondent is married to the 1991 partner
  • respondent is cohabiting with a different partner
  • respondent is married to a different partner
  • respondent has no current partner

In this way we can contribute to knowledge about cohabitation as an increasingly significant feature of society in the 21st century.


Details of study population

The sample included 536,085 people aged 16-54 years in 1991 who were enumerated in a private household (rather than a communal establishment) at the 1991 Census. Table 1 shows this sample by partnership status at 1991 and whether they were present at the 2001 Census.

There was no direct question about cohabitation in the 1991 Census, although marital status was asked; this means that a cohabiting partnership involving an LS member must be identified using the relationship questions on the census form . Because only relationship to the head of household was collected in 1991, in complex households or where the LS member is not the head of household some partnerships are likely to have been missed. (The online training module on Households and families in the LS discusses these limitations in more detail.) Moreover, for people who were enumerated at an address which was not their usual place of residence, marital status will be known but whether they were cohabiting will not be known. In Table 1, therefore, there will be a slight tendency to undercount people who were cohabiting and overcount other types of partnership status.

Tables 2-7 are based only on people who were identified as cohabiting in 1991, aged 16-54 in 1991 and present at the 2001 Census.

Many variables were imputed at the 2001 Census if cases had missing data; none of these imputed values was used in this study. Cases for whom it was then not possible to derive partnership status at 2001 were dropped; there were 1636 of them, or 9% of the sample of 1991 cohabiters.

A small minority of LS members have different records of sex (for example, they were recorded as a man at a Census but appear on a birth registration as the mother of a child). For this study, we compared each sample member’s recorded sex at the 1991 Census, at the 2001 Census and at their first entry into the LS (i.e. in the CORE1 table, see Data Dictionary). If there was a discrepancy, the sex recorded in two of these three sources was preferred. A similar exercise was carried out for the LS member’s year of birth (which was used to recalculate age where necessary).

The variable ‘Partnership status in 2001‘ [see Annex below] was derived from the variables LARP0 and MSTP0 (for more information on these variables, see the Data Dictionary), plus data from the 1991 and 2001 non-members’ tables. Other members of an LS member’s household are not linked from Census to Census, therefore there is no cross-Census identifier for them. We have used characteristics (sex and date of birth) of the LS member’s cohabitee from the 1991 non-member’s table to determine whether that person was still in the LS member’s household ten years later (in other words, whether s/he can be found in the 2001 non-member’s table). Further details of the criteria used are given under ‘Partnership status in 2001’ above.



How the variable ‘partnership status in 2001’ was constructed:

1 = ‘still cohabiting with 1991 partner’ means:

  • there was someone in the household in 2001 of the same sex as the 1991 partner, and:
  • that person also had at least two of the three date of birth elements (day, month, year) in common with the 1991 partner, and:
  • that person’s relationship to the LS member was coded ‘cohabitee’ OR it was coded ‘unrelated’ and the LS member’s living arrangement (i.e. LARP0) was coded as ‘cohabiting’

2 = ‘married to 1991 partner’ means:

  • there was someone in the household in 2001 of the same sex as the 1991 partner, and:
  • that person also had at least two of the three date of birth elements (day, month, year) in common with the 1991 partner, and:
  • that person’s relationship to the LS member was coded ‘spouse’ or ‘partner or spouse’ and the LS member’s marital status was ‘married (first marriage)’ or ‘remarried’.

3 = ‘cohabiting with new partner’ means:

  • neither of the above applied, and:
  • the LS member’s living arrangement was coded as ‘cohabiting’, and:
  • there was at least one other adult (of either sex) in the household

4 = ‘married to new partner’ means:

  • none of the above applied, and:
  • the LS member’s living arrangement was coded as ‘married and living in a couple’, and:
  • there was at least one other adult (of either sex) in the household

5 = ‘no current partner’ means that none of the above applied.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s