PROXY RESPONSES CONTAIN MORE ROUNDED TRIP DEPARTURE/ARRIVAL TIMES: EVIDENCE FROM THE 2017 US NATIONAL HOUSEHOLD TRAVEL SURVEY

Abstract

 Proxy responses in social surveys produce several biases and reduce survey quality. In this study, we used the rounding error in reported trip departure/arrival times to measure quality deterioration due to proxy responses in a travel survey. In self-reported travel surveys, the trip departure/arrival times are often reported as rounded values, such as those ending in 00, 10, 15, 30, and 45 min. We hypothesized that, compared to self-response, proxy response often contains more rounded departure/arrival times. Data from the 2017 US National Household Travel Survey were used to analyze the demographic characteristics and effects of proxy responses. A descriptive analysis revealed that males in their 20s–50s tended to be proxy respondents, whereas males in their 50s or older tended to be proxy completers. Proxy-reported discretionary trips were often misreported, and proxy responses often contained rounded departure/arrival times. To control the effects of demographic variables, we finally estimated a binary logit model to describe the relationships among the self- and proxy response choices, demographic attributes of the respondents, and rounding in the reported times The results revealed a statistically significant association between proxy response and rounded reported time, supporting our hypothesis.