The Impact of Survey Completion with a Mobile Device in a Longitudinal Transport Study
Sascha Hoogendoorn-Lanser, KiM Netherlands Institute for Transport Policy Analysis, Krisje Wijgergangs, KiM Netherlands Institute for Transport Policy Analysis,, Toon Zijlstra, KiM Netherlands Institute for Transport Policy Analysis
The aim of this research is to assess the impact of mobile response on the data quality in a longitudinal panel survey, which is not optimized for mobile use. It concerns the mobile device use by respondents of the Netherlands Mobility Panel (MPN).
THE IMPACT OF SURVEY COMPLETION WITH A MOBILE DEVICE IN A LONGITUDINAL TRANSPORT STUDY
Netherlands Institute for Transport Policy Analysis (KiM)
Nowadays online questionnaires are by definition mixed-mode questionnaires. Respondents use traditional devices (laptops and desktops) and mobile devices (tablets and smartphones). However, these devices differ significantly in terms of their screen sizes, input methods and connectivity. Moreover, we observe differences in use. Desktop computers are stationary, while tablets and smartphones are characteristically mobile. These differences have potential consequences for data quantity and data quality. In this paper we examine the impact of such mode effects in the Netherlands Mobility Panel (MPN), with impact defined as effect times scale.
From 2013 to 2016, we observed a sharp increase in the share of smartphone respondents, from 4% to 18% of the total number of respondents. If the current trends continue, we expect by 2023 there will be 50% desktop or laptop respondents, 15% tablet respondents and 35% smartphone respondents.
The profile of tablet and smartphone respondents differs considerably from that of traditional online respondents; they cannot be readily compared with one other. Among mobile respondents, we observe a relatively high proportion of women, young people, and social and practical internet user profiles. Highly educated people and residents of rural areas are less inclined to use a mobile device for survey administration. We use these insights to match mobile respondents to traditional device users, as such we control for self-selection when modelling the mode-effects. Nearest neighbour matching with a one-to-one ratio is used.
In terms of mode-effects, we examined four potentially relevant effects in the MPN: straightlining, number of given answers, respondent avoidance behaviour, and questionnaire completion times. Virtually all the results are insignificant, which implies that there are virtually no differences in the data due to the devices used for completing the questionnaires. This is a reassuring finding. Mobile respondents do take longer to complete the survey.
Association for European Transport