Data Quality in Web and App Diaries: A Person-Level Comparison

Abstract

The time-use diary is a complex and burdensome data collection instrument. This can negatively affect data quality, leading to less detailed and/or inaccurate activity reporting as the surveyed time period unfolds. However, it can also be argued that data quality may actually improve over time as respondents become more familiar with the diary instrument format and more interested in the diary task. These competing hypotheses have only been partially tested on data from paper and telephone-administered diaries, which are traditionally used for large-scale data collection. Less is known about self-administered modes that make use of new technologies, despite their increasing popularity among researchers. This research note rectifies this omission by comparing diary quality in self-administered web and app diaries, drawing on data from the Millennium Cohort Study. We construct a person-level data quality typology, using information on missing data, episode changes, and reporting of key daily activity domains. Results show significant mode differences on person-level data quality, after controlling for characteristics known to influence diary mode selection and data quality. App diarists were more likely to return two diaries of inconsistent quality. Both respondent fatigue and improvement of completion over time appear more common among app diarists.