Device Effects and Response Quality in Online Surveys: Evidence from Czech GGS Paradata

This study examines device effects on response behaviour and data quality in the Czech Generations and Gender Survey, Wave II, using keystroke-level paradata from a fully mobile-optimised CAWI instrument. We compare respondents completing the survey on PCs/laptops and smartphones across indicators capturing cognitive load, satisficing, and respondent effort, including response latency, completion time, answer changes, page revisits, item nonresponse, midpoint use, “don’t know” responses, and straightlining. Descriptive analyses and multivariate models controlling for age, gender, and education show that most device differences are small or negligible. Smartphone use is not associated with longer completion times or systematically lower engagement once demographic composition is accounted for. The only robust and substantively meaningful device effect is a higher incidence of straightlining among smartphone respondents, alongside a very small increase in item nonresponse. Education consistently attenuates device-related differences. Overall, the findings indicate broad comparability across devices in a mobile-optimised survey, while highlighting specific behavioural domains where smartphones continue to shape response processes.