Could you please pay attention?’ Comparing in-person and MTurk Responses on a Computer Code Review Task

Abstract

The current study examined the differences in data quality across two environments (i.e., in a laboratory and online via Amazon’s Mechanical Turk) on a computer code review task. Researchers and practitioners often collect data online for the sake of convenience, as well as for obtaining a more generalizable sample of participants. The lack of social contact between the researchers and participants, however, may result in less effort dedicated to the experimental task resulting in poor quality data. The results of the current study showed that data quality—at least when measuring the individual difference variables—was drastically worsened when the experimental task was presented online. In contrast, we observed little differences in the experimental task perceptions across the two samples. Rather, participants spent significantly less time examining the computer code when completing the experiment online. The current study has implications for the effects of using online platforms (like MTurk) to collect experimental data.