How rating scales influence responses' reliability, extreme points, middle point and respondent's preferences

This study sought to analyse different rating scales used in surveys including two ordinal scales with numerals, an ordinal scale with face images, and a ratio scale. First, statistical analyses were performed such as mean comparison, skewness, and kurtosis. Subsequently the scales' critical points, such as the extreme levels and middle points, were analysed. Finally, scale preferences were compared according to gender, age, and education. The study was carried out with 595 people, and the results showed a tendency for scales with fewer response items to obtain higher values. Furthermore, there was a higher incidence of responses at extreme levels, mainly at higher levels. It was possible to identify differences between the scale preferences within groups, and it was observed that the easier scale does not necessarily have more rapport with respondent's feelings.

Automated summary

This study analyzed different rating scales used in surveys, finding that scales with fewer response items tend to obtain higher values, with a higher incidence of responses at higher levels. Differences in scale preferences were identified among groups, and it was observed that the simpler scale does not necessarily have more rapport with respondent's feelings.