Comparison of the responses of cross-sectional web- and paper-based surveys on lifestyle behaviors of Japanese adolescents

Objective: In Japan, paper-based surveys are currently more effective than web-based surveys. This study compared the response rates and lifestyle behaviors between paper- and web-based surveys conducted among Japanese junior and senior high schools. Methods: In total, there were 42 and 64 junior and senior high schools, respectively, for the web-based surveys and 20 and 27 junior and senior high schools, respectively, for the paper-based surveys. The questionnaire covered lifestyle behaviors (e.g., alcohol consumption, smoking, exercise, eating, and sleep status), mental health, and plans to attend college. School- and student-level response rates by survey method were assessed, and so was the effect on the reporting of each lifestyle behavior using logistic regression models. Results: The school response rates were 16.0% and 38.3% for web- and paper-based surveys, respectively. The student response rates were 88.7% and 77.2%, respectively. The web-based group had significantly more female participants and lower response rates for higher grades in senior high schools. The odds of lifetime and current alcohol consumption and poor mental health were lower, whereas those of lifetime use of conventional cigarettes, shorter sleep duration, and plans to continue attending college were higher among web-based (vs. paper-based) participants. Conclusions: The school response rate was poor in the web-based survey. However, whether the differences in lifestyle behavior are attributed to selection bias or the survey method remains unclear.