Differential response rates in postal and Web-based surveys in older respondents

Abstract

Response rates were compared across a postal and a Web-based survey containing the same questions about preferences for future for the design of future nursing homes which were sent to a random sample of 10,000 individuals aged 50-75 years. The 10,000 individuals were randomly allocated to receive a postal questionnaire or a letter with a Web link to an online version of the same questionnaire. The Web-based survey gave a significantly lower response rate than the postal survey. The Web-based version improved the sample representativeness with respect to gender but worsened the representativeness with respect to age. Respondents' characteristics in the Web-based survey differed significantly from those of respondents in the postal survey with respect to income, education, civil status and health status. The Web-based version improved data quality by significantly lowering the number of item non-response and 'don't know' answers. Respondents found it easier to answer the Web-based survey. The cost per response was significantly higher for the Web-based survey because of the significantly lower response rate compared to the postal survey.