The usual simple rating scale purports to measure direction (important/unimportant, effective/ineffective, etc.) and intensity (very, somewhat) of attitude or opinion in a single assessment. Thus, direction and intensity components may be confounded in simple rating scales, which can increase opportunities for form-related biases such as central tendency error. This study examines response tendencies in simple rating scales using the traditional approach and an alternative two-stage approach, both designed to assess the performance of a major charitable organisation in the United States. Using a measure of the proportion of extreme responses at both the individual scale and respondent levels, results suggest that simple rating scales as generally used (i.e. one-stage) tend to underestimate extreme viewpoints held by people and may be subject to a central tendency form-related error.