High group level validity but high random error of a self-report travel diary, as assessed by wearable cameras

Abstract Self-report remains the most common method for collecting epidemiological evidence of the links between travel and health outcomes. This study assesses the validity and reliability of a self-reported travel diary (a modified version of a well-established UK travel diary; The National Travel Survey (NTS)) by comparison with wearable camera data. Across four locations (Oxford, UK; Romford, UK; San Diego, USA; and Auckland, New Zealand) we collected 3–4 days of SenseCam (wearable camera) and travel diary data from 84 adult participants (purposive sample). Compliance with the data collection protocol was high and inspection of the crude results suggests acceptable agreement between measures for total days of data collected (diary=278; SenseCam=274), daily journey frequency (diary=4.78; SenseCam=4.64) and average journey duration in minutes (diary=17:46; SenseCam=15:40). Once these data were examined for total daily time spent travelling in minutes agreement was poorer (diary=84:53; SenseCam=72:35). Analysis of matched pairs of journey measurements ( n =1127) suggests a positive bias on self-reported journey duration of 2:08 min (95% CI=1:48–2:28; 95% limits-of-agreement=−9:10 to 13:26). Similar analysis of diary days matched to complete SenseCam days ( n =201) showed a very small positive bias with a very large limits-of-agreement (1:41 min; 95% CI=−2:00 to 5:24; 95% limits-of-agreement=−50:29 to 53:41). These results suggest self-reported journey and daily travel exposure data are relatively valid at a population level, though corrections according to reported bias could be considered. The large limits of agreement for matched journey and diary summary analysis suggest self-report diaries may be unsuitable for assessment of an individual׳s travel behaviour.