We're implementing comment cards on our site that contain a simple Likert-type scale that uses colored buttons (red, orange, yellow, light green, green), to denote the range of negative to positive ratings.
Are you aware of any research on the effect of color on user responses to Likert-type survey questions?