25-year trends and socio-demographic differences in response rates: Finnish adult health behaviour survey.

When estimating population level changes in health indicators, the declining response rate, especially if also the characteristics of non-respondents are changing may bias the outcome. There is evidence that survey response rates are declining in many countries. It is also known that respondents and non-respondents differ in their socio-economic and demographic status as well as in their health and health behaviours. There is no information about the changes in the differences between respondents and non-respondents over time. Our purpose was to investigate the changes over time in the differences between respondents and non-respondents in respect to their sex, age, marital status and educational level. The data from the Finnish Adult Health Behaviour Survey (1978-2002) was used. The response rate declined over the past 25 years for both men and women in all age groups. The decline was faster among men than women, and also faster in younger age groups than older age groups. There is a marked difference in the response rate between married and non-married persons but it did not change over time. Also the response rate between different educational levels differed for both men and women, and this difference increased over the years. The declining response rate and at the same time occurring change in the non-respondent characteristics will decrease the representativeness of the results, limit the comparability of the results with other surveys, increase the bias of the trend estimates and limit the comparability of the results between population groups.