Non-participation correction and adjustments of weights for the Household section of the Health and Disability survey

The Household section of the Health and Disability survey was collected among people living in households in 2008. 29 931 persons living in mainland France and in overseas “départements”, have responded to the survey, which correspond to a response rate of 76, 6 %. This major structural survey, includes many questions about respondents' functional state (diseases, deficiencies, limitations), social participation restrictions (access to the labour market, educational attainments, aspects of standards of living), environmental factors (like familial or social network, assistive technologies, home fittings, accessibility). This document deals with the sampling framework, the sampling process, non-participation corrections and adjustments. The sample derived from those individuals having answered to the filter-survey “Daily Life and Health, 2007”. Stratification aims to correct for geographical variation and for presumed severity of disability situations. The latter criterion enabled an over-representation of people (presumably) suffering from severe handicaps. This criterion, built up with the filter-survey also enabled to take into accounts individuals in good health, for comparisons, and a large enough sample population to limit the weight dispersion and the consequences of “false negatives”. The non-participation corrections consist of dividing the individual weight by his response probability. This probability is computed with logistic models derived from demographic, geographical and data from the filter-survey. A calibration was performed using demographic data chosen in accordance with the geographical criterion. Eventually, the weight dispersion was found to be high, the inter-decile ratio reaching 40. This happens to be the result first of the sampling process used in the filter survey, based on the annual census survey, 2006, secondly the presence of extensions, and thirdly the stratification, resulting from the variation in the disability situations.