Research on Research
BACKGROUND: In theory, more scale points should permit detection of increasingly subtle differences. After all, shouldn’t a nine-point scale invite finer distinctions than a five-point scale? We wondered whether scales with different numbers of points would produce results that differ in terms of statistical variance, neutral response and completion time.
RESEARCH METHOD: Respondents from a nationally representative online panel of household shoppers answered questions about sports drinks. Respondents rated brands on 28 attributes and were randomly assigned to use otherwise identical scales of 5 (n=416), 7 (n=417) or 9 points (n=417).
FINDINGS: Intriguingly, use of the mid-point decreased as the number of scale points increased (see figure). A series of paired-sample T-tests compared a randomly selected reference item to each of the other items as a test of inter-item discrimination. This process was conducted separately for the five-, seven- and nine-point scales. Seven-point scales demonstrated better inter-item discrimination than both five- and nine-point scales, with 22 significant inter-item differences. The average time to complete the questionnaire was about the same for five- and seven-point scales, but one minute longer for respondents using nine-point scales.
CONCLUSION AND RECOMMENDATIONS: Using scales with more than five points appears to elicit fewer neutral attitude positions. Further, seven-point scales do not increase the time it takes respondents to complete a questionnaire over five-point scales, while nine-point scales do. Additionally, seven-point scales appear to provide superior inter-item variance to both five- and nine-point scales. Thus, while methodology choices always involve many considerations, we generally recommend using seven-point scales to maximize variance and efficiency.