Impact of Number of Scale Points on Attitude Measurement

This diploma thesis deals with the rating scales, which are widely used in the attitude measurement, namely with the impact of their length (number of scale points) on measurement, both in terms of cognitive processes in the minds of respondents and impact on the data. Hypotheses, formulated on the basis of the literature and tested using data obtained from a split ballot experiment and cognitive interviews, are related to the comparability of results, distribution of responses and task demands. The Net Promoter Score from the environment of market research, expressing respondent's willingness to recommend a service or company used, is used as an application example. It originally uses 0 to 10 eleven-point scale which is here compared to five-point scale from 1 to 5 using various recoding and rescaling techniques. Analysis shows that rating on eleven-point and five-point scales means two different cognitive tasks for respondents (eleven-point scale is a bit more demanding but more accurate in respondents perception) and that the data differ from each other in terms of distribution of responses - on short range scale, midpoint and extreme positive category are selected more often, while on eleven-point scale, milder positive points are used more often