|
Abstract:
|
In this paper, I deal with web surveys response times as well as their association with data quality and I provide guidelines and suggestions for web survey scholars and practitioners working with response times. Firstly, I highlight the limited usefulness of the total time spent to complete the whole questionnaire and I argue that our primary focus should be on the response times of each item separately. Furthermore, I show that occurrence of long item response times are usually random and not associated with specific users or items and that extremely long times usually occur after an external distraction (e.g. an incoming email or phone call). Then, I suggest a method that we can use to flag responses that were given in extremely short time (i.e. a response time so short that shows that the flagged respondents instead of reading and comprehending the question, they have scanned the question text). Finally, I compare the suggested method with the method that is most widely used today for speeding detection.
|
