Likert Scaling

History and Development

Rensis Likert, in collaboration with Gardner Murphy, began development of his method of summated ratings in 1929 in response to Louis L. Thurstone's equal-appearing intervals method. Despite the strengths of Thurstone's approach, the method was criticized for its unwieldiness. Specifically, the method required burdensome calculations in a time before the advent of computers and required a sample of participants to act as judges in the pretesting of potential items. As a result, creating scales using the [Page 537]Thurstone method proved to be time-consuming. The method was also criticized by Likert and others for making a number of statistical assumptions that were unverified at the time (e.g., that the scale values of the items are independent of the attitudes of the pretesting judges). Likert developed his method with the goal of creating a reliable and valid attitude scaling procedure that was less time-consuming to construct and that did not require unnecessary assumptions. The procedure that Likert presented in 1932 has changed little throughout the years, apart from capitalizing on the statistical computation advantages offered by modern computers.

Figure 1 Sample Instructions and Items for a Likert Summated Ratings Scale

Constructing a Likert Scale

The first step in creating a Likert scale is to specify the attitudinal object. The more well-defined the subject under consideration is, the better. Once the attitudinal object has been determined, the researcher generates a pool of potential items. The aim of this phase is to create statements reflecting a broad diversity of viewpoints on the topic. Although belief or cognitive items (e.g., Items 1, 3, 4, and 6 in Figure 1) are most typical, feeling or affective items (e.g., Item 2) and behavioral tendency or conative items (e.g., Item 5) are also common. In contrast with some methods that use statements varying along the favorable-unfavorable continuum (e.g., Thurstone scaling), only statements that are clearly positive or negative (although not at the most extreme) should be selected for a Likert scale. The reasoning behind this recommendation is that extremely worded or neutral statements do not discriminate among respondents, that is, they do not distinguish people with very positive attitudes from people with moderately positive attitudes, negative attitudes, and so forth. Care should be taken to avoid biased or leading statements, extremely worded or neutral items, and items containing well-known facts.

The item pool usually consists of at least 25 statements, and can be as many as 50 or more. The primary reason for using multiple items rather than a single statement is that each statement may have ambiguities and subtle biases, leading people to respond in a certain way. By summing or averaging across multiple related items, the impact of biases and imperfections contained in individual items can be minimized. A secondary reason for using multiple items concerns breadth. Attitudes are often multifaceted, involving cognitions, emotions, and behavioral tendencies. A single item is unlikely to capture the full scope of the attitude in question; using multiple items potentially ameliorates this problem.

...