Switching to self-completion protocols impacts item nonresponse patterns. Lessons from the European Social Survey rounds 9, 10 and 11

Surveys are increasingly pivoting from interviewer-assisted to self-completion protocols, yet the consequences for data quality remain unclear. Guided by the survey satisficing theory, this is the first study that inspects the COVID–19 mode change in the European Social Survey Round 10, where eight countries switched to self-completion. At the same time, 19 countries maintained interviewer-assisted modes, with all 27 entirely implementing them in Rounds 9 and 11. We analyse item nonresponse on three items with complex response formats and three standard opinion questions with historically high missingness. We use hierarchical Bayesian beta regression to estimate mode effects. Our findings bring significant considerations for survey studies transitioning from interviewer-assisted to online self-completion mode. Switching to self-completion resulted in a marked increase in nonresponse for the cognitively demanding items, but a decline for the standard opinion items; both effects disappeared when interviewer-assisted modes were used. Our study supports the survey satisficing interpretation: self-completion fosters strong satisficing behaviour on complex survey items, yet it results in weak satisficing on opinion items.