Response rates of self-administered paper-based and digital epidemiological surveys: A systematic review and meta-analysis
Recently, epidemiologists have started adopting digital surveys as a method for data collection. However, it is unknow whether these methods are any better than paper-based surveys.
We undertook a systematic review investigating the response rate differences between questionnaires administered by paper-based and those administered using digital platforms to collect exposure and/or outcome data in epidemiological studies.
We searched through the peer-reviewed published literature for mixed-method epidemiological studies that use paper and digital surveys for data collection. We looked through Medline, Web of Science, CINAHL and PsychINFO. Data were extracted and effect size and study variance were calculated. Using this information, a random-effects model was conducted using R program. Additionally, in order to estimate the relationship between response rate and survey features, sub-group analysis was carried out.
Eighteen studies were selected, off which five had to be excluded from the meta-analysis as they were influential cases. Based on the random-effects model, the response rate difference was 5.90 (1.59 - 10.23) and the overall study heterogeneity was considerably high (I2 = 95%). Sub-group analysis revealed survey features had no impact on the response rate difference.
We conclude that paper surveys a have slightly higher response rate than digital surveys, and that sub-group analysis of study features did not explain this difference. However, due to the clinically and methodological heterogeneity of the reviewed studies, it is difficult to conclude anything from the meta-analytical findings of this review.