The question is asked by John Sides here in response to this article. The article argues that the idea respondents cheats on self-completed surveys is a myth.
I think there are two main reasons why people don’t cheat in online surveys, one methodological and one theoretical.
First, in the study presented in the article only 2 out of 505 people cheated (less than 0.5 pct.). However, there is a big difference in the design compared to a real world setting. The respondents used a computer in a non-familiar room with no information about potential monitoring mechanisms (which, of course, were present). I recognize that it is hard to control for cheating behavior in a non-lab setting, but there are technological ways to overcome this problem. I may return to this problem in a later post.
Second, flip the question: Why would people cheat? It’s anonymous, nothing to win and nothing to lose. Imagine that people got $1 for each correctly answered question. Oh my, people would cheat! But when the gain of cheating is more or less non-existent, it is no mystery people aren’t motivated to cheat.
To sum up, it would be strange if anonymous people would cheat in a non-familiar setting without knowing whether or not he/she is being monitored, when the gains are low.