Article Text


Collecting information on marital status: a methodological note
  1. G Barrett,
  2. K Wellings
  1. London School of Hygiene and Tropical Medicine, London
  1. Correspondence to:
 Dr G Barrett, Health Promotion Research Unit, Department of Public Health and Policy, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK;

Statistics from

Marital status is often probed in epidemiological, public health, and social surveys, and has been shown to be associated with various health outcomes. However, in the past 25 years there has been a trend towards cohabitation, and higher rates of divorce and remarriage.1 Birth registration data also show that nearly 40% of births are outside marriage, a large proportion of which are to cohabiting couples.2 The way in which data are collected has tended to reflect the increase in cohabitation; “living with partner” tends to be an additional category of marital status, for example, Johnson et al.3 In our recent work (a study to develop a measure of unplanned pregnancy) we wanted to identify both women's marital status and who they were living with because previous research has shown that the life experiences of certain groups (for example, never married cohabiting mothers, previously married cohabiting mothers, and never married lone mothers) are different.4 Collecting these data was more complicated that we had expected.

We piloted our questions with 26 women and used interview techniques to assess their understanding, similar to a method used by Donovan et al.5 We soon realised there was a problem with version 1 of the marital status questions (table 1). One interviewee (aged 22, living with partner) left the question blank and afterwards reported that she had been confused over which to tick. A second interviewee (aged 42, living with partner for past 18 years) ticked “single” but put a question mark against it and afterwards commented on her dissatisfaction with this question. A third woman (aged 37, living with partner) ticked “single” but stated that she disliked “single” because of its connotations of “single mother”.

Table 1

Questions piloted

In version 2, we placed the “living with” question first and changed “single” to “unmarried”. Seven women completed these questions, and it was clear that no problems had been solved. One woman (age 37, living with partner for past 12 years) ticked “unmarried”. Afterwards it emerged that she had previously been married for five years—so was technically in the “divorced/separated” category. She was answering the question with her current partnership in mind, rather than her legal status with regard to marriage. Two other women ticked “married” but hesitated some time over the question. We felt there was sufficient doubt about the validity of the question and decided to omit it altogether in version 3, and separated out “husband” and “partner” in the remaining question. This question worked well with the remaining pilot interviewees and the 1000 plus women who took part in the field testing.


We were surprised at the number of problems we encountered with the marital status question. Generally, women did not see the categories as representing their legal state in relation to marriage, but understood them in a more colloquial sense. However, by omitting the marital status question we are simply focusing on the presence of a partner in the home; we are not able to identify particularly subgroups of interest (for example, never married lone mothers, previously married cohabiting mothers). As partnership histories become more complicated, we cannot assume that old relationships between marital status and health outcomes still stand, and research is required to see if there are new patterns. However, asking about marital status is no longer straightforward; research instruments must take account of the fact that responses will be made in terms of lay conventions rather than official categories, and there may be disparities and discontinuities between the two. Questions that can collect complex information about partnership/marital histories, yet are valid in lay terms, and are reliable and concise, need to be developed.


View Abstract


  • Funding: GB was funded by a Medical Research Council/London Region special training fellowship in health services research.

  • Conflicts of interest: none.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • In this issue
    John R Ashton Carlos Alvarez-Dardet