著者

村尾 博

出版者

青森公立大学

雑誌

青森公立大学経営経済学研究 (ISSN:13419404)

巻号頁・発行日

vol.10, no.2, pp.3-20, 2005-03

---

Likert (1932) developed a scale for the measurement of attitudes, which became known as the "Likert scale." Nowadays, data extracted from a single Likert-type item are often treated as interval data under the name of Likert scale or Likert data. This is unfortunate because such data are illegitimate Likert data. They are ordinal data by construction. In order to distinguish from legitimate Likert data, data extracted from a single Likert-type item are called "Likert type item data" in this paper. This paper investigates the logic of the Likert scale in order to find out whether or not Likert-type item data can be treated as interval data under a certain condition. In Likert's book, the normal population is used as a system of transformation from rank information to quantitative information. This mechanism can be used for the case of Likert-type item data as well. From the viewpoint of emphasizing the importance of normal population, this paper shows a practical condition under which Likert-type item data can be treated as interval data. Furthermore, this paper provides answers to a few questions discussed among researchers using Likert-type item data. Labovitz (1967, 1970) demonstrates that treating ordinal data as interval data by arbitrarily assigning numbers to ordinal categories can be both legitimate and useful. This paper critically analyzes the logic behind Labovitz's idea. Such analysis recognizes the importance of the normal population assumption.