As I noted in a previous post, Mick Couper and I have just finished some research where we varied screen background color to see if it made any difference in breakoffs, in respondent perceptions of time to complete, or in their general sense of the survey experience. We tested three colors: white, blue, and yellow. Here are some self-explanatory screenshots:
The idea for this research came from some work by Gorn ( Gorn et al. (2004), Journal of Marketing Research, XLI, 215-225) which demonstrated that a Web users perception of the length it took to download a large file was affected by the background color. People felt it took longer and abandoned the task more often when they saw a red background than when they saw blue. Duston Pope and I did a small test with a student sample and replicated some of this with respondents who got a blue background in their survey reporting shorter perceived completion times and those how got a pink background, although there were gender effects that confused the issue some. So Mick and I decided to take another stab at it but to stay clear of pink especially because of the potential gender effects.
What did we find? Well, not all that much. We found that the yellow background produced significantly more breakoffs than the blue background, and while white produced more breakoffs than blue the effect was not significant. There were no differences in time to complete, in perceptions of the length of the survey, or in any of the debreifing questions we asked such as ease of reading, enjoyment, and so on.
Based on these results we may want to change our standard here at MSI from white background to blue, even though we would expect only a modest reduction if any in our breakoff rates.