Validating overcrowding measures using the UK Household Longitudinal Study
AuthorsNoriko Cable and Amanda Sacker
PPR, BS and MBS were derived using relevant housing grid information and housing information from the dataset, which were aggregated at a household level (N = 18,848). Raw scores were categorised into ‘under occupied (rooms < people)’, ‘balanced (rooms = people)’, ‘overcrowded (rooms < people)’ according to an established cut-off point for each overcrowding measure. Kappa coefficient was used to assess the level of agreement between overcrowding measures. Construct validity of the measures were tested against log-transformed household equivalised income and housing tenure status as well as with each component of overcrowding measures. Using individual data (N = 38,455), face validity of the overcrowding measures was tested against satisfaction with health and mental health indicated by GHQ12.
Each overcrowding measure has a fair agreement with the others (kappa = 0.44, p<0.001). All overcrowding measures were significantly correlated with income and household tenure in a similar manner. However, components of overcrowding measures were associated differently to these socioeconomic indicators, while they were better correlated with satisfaction with health compared to GHQ12, showing a complex aspect of overcrowding measures.
In sum, use of PPR as a socioeconomic indicator is reasonable. However, given the complexity of the mechanism of health inequalities, the relevant household information is required to understand the link.