著者

山田 一成

出版者

東洋大学社会学部

雑誌

東洋大学社会学部紀要 = The Bulletin of Faculty of Sociology,Toyo University (ISSN:04959892)

巻号頁・発行日

vol.56, no.2, pp.79-94, 2019-03

---

One of the benefits of Web surveys is that a variety of paradata regarding how respondents answer questions are generated by the survey process and that these paradata can be used to evaluate response quality of Web surveys. However, few studies have examined whether respondent interruptions alter response quality. To address this discrepancy, this article investigates the interruptions and their consequences for response time and response quality in volunteer panel Web surveys. Data were based on two Web surveys in the Tokyo metropolitan area (one of 1,444 Japanese adults in January 2017, and the other of 622 Japanese adults in January 2018). The results revealed the following: (1) interruptions are common (more than 40% of all respondents) and they add substantially to the duration of surveys, but (2) interruptions do not appreciably alter response quality, and (3) interruptions do not have a significant effect on the number of straight lining answers on the Likert grid containing 15 items, nor do they reduce the number of extremely short response time to questions. Thus, while interruptions may be an issue for survey administration, they do not appear to reduce the quality of data collected by researchers through volunteer panel Web surveys. Finally, the paper concludes by discussing implications and limitations of this research.