Satisfaction guaranteed

Ken drew my attention to a really interesting little piece that shows that when answering questions via a Likert scale people are biased towards responses on the left of the page, a phenomenon known as psuedoneglect. So a descending scale (good,fair,bad) will get a better response than an ascending one. This clearly has implications for survey design as well as helping some of us improve our teaching evaluations. It would be interesting to know if there is any bias when the scale is arranged vertically: the theory would say not.

Michael E.R. Nicholls, Catherine A. Orr, Matia Okubo, Andrea Loftus (2006)
Satisfaction Guaranteed: The Effect of Spatial Biases on Responses to Likert Scales
Psychological Science 17 (12), 1027–1028.