Changes in survey mode (e.g., online, offline) may influence the values of survey responses, and may be particularly problematic when comparing repeated cross-sectional surveys. This paper identifies mode effects by correcting for both unit non-response and sampling selection using the sample...