Research Tips

Some Points About Scales

Anyone who has ever created a survey in which respondents were asked to rate something using a scale has probably wondered how many points to use. There are two key issues to consider:

Whether the number of points should be odd or even.
Whether to use a scale with a large or small number of points.

From an analytical perspective, even-numbered scales with a large number of points tend to be preferable because these normally spread responses over a wider distribution, which is useful for many statistical procedures. However, such scales can be problematic in terms of validity and reliability. Using an even-numbered scale (and thus not giving respondents the option of choosing a neutral middle option) may threaten validity if the true attitude of many respondents is neutral. Using a scale with many points can threaten reliability if respondents have difficulty discriminating between different response categories and thus might not give the same responses if asked the same questions a few days later. In general, the more interested and involved respondents are in the subject of a survey, the greater the chances that they will be willing and able to make the fine-grained discriminations required by scales with many points.