Measures of reported disability prevalence are often subject to great variability and this can make drawing precise conclusions from them difficult. In this box, we considered two examples where measures varied. The first related to differences between Labour Force Survey and General Health Survey measures of working-age disability prevalence and the second to large changes in the former. We discussed differences in methodology and changes in attitudes as possible drivers.

Measures of reported disability prevalence derived from surveys are often highly sensitive to the survey methodology on which they are based. So different surveys can yield different results.

For example, in their 2015 study of the disability employment gap in the UK, Baumberg et al found that the Labour Force Survey (LFS) and General Household Survey (GHS) yielded contradictory results for the trend in working-age disability prevalence between 1998 and 2012.a In particular, the GHS reports a fall in working-age prevalence in that period, while the LFS reports an increase. The surveys use similar methodologies, but the authors see differences in the definition of disability used, the geographical area covered by each sample and how interviews were conducted as possible sources of the discrepancy. Even after controlling for these, the authors found that a discrepancy remained between the two prevalence estimates.

Methodological differences can also mean that even estimates within surveys are not fully comparable. For example, there have been several changes to the wording of the LFS disability questions since 2010. These harmonise the LFS definition with other household surveys, but mean that the pre-2010 series cannot readily be compared to the post-2010 series.

In November 2017, the ONS suspended publication of its LFS-derived Labour market status of disabled people statistical series for six months due to an apparent discontinuity between the second and third quarter releases in 2017. ONS analysis of the sudden increase in reported prevalence between the releases remains inconclusive, but it may be due in part to changes in respondent behaviour associated with mental health awareness campaigns. This hints at a broader problem in defining the underlying prevalence of disability in the population. Even if survey measures were consistent within and across surveys, the decision to self-report “a longstanding illness/disability/impairment that causes difficulty with day-to-day-activities” (necessarily) is the product of: (i) actual incidence of longstanding conditions; (ii) awareness of ‘disability’; and (iii) the perceived effect of impairments in interfering with everyday life. Changes in awareness and/or the demands of everyday life could therefore lead to changes in self-reported prevalence with no accompanying change in the underlying rate of conditions.

This box was originally published in Welfare trends report – January 2019

a See Baumberg et al (2015), Disability prevalence and disability-related employment gaps in the UK 1998-2012: Different trends in different surveys?