Author
Abstract
For social surveys, combining in-home computer-assisted personal interviewing (CAPI) with online self-completion (web) has advantages in terms of both average participation propensities (and therefore response rates) and the distribution of participation propensities over subgroups (and therefore sample composition and representativeness). However, the best way to combine the two modes of data collection is unclear. Different ways may lead to different outcomes, in terms of response rates, sample representativeness, and costs. Findings are presented from a large-scale randomised experiment on a national general population survey in the UK, in which initial use of web with a CAPI-follow up for non-respondents (web-first) is compared to initial use of CAPI with web follow up for non-respondents (CAPI-first). The web-first design is found to result in a higher response rate, a more representative sample, and lower data collection costs.
Paper download