Social class attribution: mortality and life expectancy analysis

Louisa Blackwell, Brian Johnson and Bola Akinwale, Office for National Statistics

[Project number 20096]

Linkage of census and death records in the LS has allowed more reliable statistics of socio-demographic differences in mortality (see for example LS Series No. 6).

In most previous LS analyses, the method for assigning social class to LS members used social class at entry to the study, which for most members was 1971. LS members with no assigned occupation in 1971 or classified as armed forces, inadequately described, students, retired, permanently sick or housewives were given the social class of their spouse, father or mother (in this order).

This project will review the current rules for social class attribution in mortality analyses, taking into account:

  • The implications of women's increased labour force attachment since 1971.
  • Whether it would be appropriate to use household rather than individual attribution rules. A primacy rule would classify all household members according to the most advantaged social class position of members within the household. Associated issues are the relative positions of IIIN and IIIM in the 'hierarchy' and questions raised in the literature on social classification about the validity of assuming that resources are shared within households.
  • The life stage at which it is most appropriate to attribute social class and whether individual social mobility since 1971 should be reflected in the attribution rules.
  • In individual attribution, whether it is appropriate to use former or later social class if an individual is not in employment at the preferred time point.

Alternative algorithms will be derived and adjudicated through assessment of differences in the social profiles they produce. Validation will involve the creation of life expectancy by social class, based on selected algorithms.

The project will also assess comparability between NS-SEC and RG Social Class attribution.

The LS is unique in allowing the prospective analysis of mortality by socio-demographic characteristics.

Top of page