Large sociodemographic differences in response rates and methods in a population survey in Sweden

Abstract

Abstract Background Declining response rates in population surveys may decrease the representativeness and introduce bias in survey results. This study investigated sociodemographic differences in response rates and methods in a population survey in Sweden. Methods The study is based on a large population survey conducted in five counties in Mid-Sweden in 2017. Three age-specific questionnaires for persons 18-29, 30-69 and 70 years or older were designed to study living conditions, life style factors and health. The sample was random and stratified by gender, age group and municipality, and comprised 78 000 persons 18 years or older. Up to two reminders were sent, the last reminder consisted of a short version of the questionnaire. The survey could be answered using a written questionnaire or on-line. The web-questionnaire was available in Swedish and English. Results The overall response rate was 44% with the lowest participation rate in ages 18-29 years (26%) and highest in ages 70-84 years (70%). Women, persons with high educational level, and persons born in Sweden were more prone to respond than others. Those aged 18-29 years were overrepresented in the sample and the participation rate would have been 46% if the sample had been unstratified by age. The majority of the respondents answered the written questionnaire, even though it was more common that younger people responded via the web compared to older people. In total, 8% of the respondents replied to the short form and this was most common (17%) among those born outside the Nordic countries. Conclusions In line with previous research, the results indicate large sociodemographic differences in response rates. The response methods also vary between sociodemographic groups, and knowledge about these differences may be helpful when designing future population surveys. Key messages The response rate was higher among older people, women, persons with high educational level, and persons born in Sweden than among other groups. Persons born outside the Nordic countries were most prone to respond to the short form and younger people replied more often to the web-questionnaire than older people.