ERIC Number: ED420694
Record Type: Non-Journal
Publication Date: 1998-Apr-17
Reference Count: N/A
Heightened Sensitivity of Likert Attitude Scales to Restriction of Sample Range.
Roberts, James S.; Wedell, Douglas H.; Laughlin, James E.
The Likert rating scale procedure is often used in conjunction with a graded disagree-agree response scale to measure attitudes. Item characteristic curves associated with graded disagree-agree responses are generally single-peaked, nonmonotonic functions of true attitude. These characteristics are, thus, more generally consistent with an unfolding model like the Thurstone attitude measurement procedure rather than a cumulative model such as the Likert procedure and the typical item characteristics of the disagree-agree responses can lead to invalid measurement such that individuals with the most extreme attitudes receive Likert scores that are indicative of more moderate opinions. This type of invalidity may be exacerbated when a Likert scale is constructed under restrictions of the sample attitude range. Specifically, a restricted sample range may mask the nonmonotonic response characteristics of a given item so that its characteristic curve appears to be monotonically related to attitude. The item would, thus, appear worthy of inclusion on a Likert scale. If that Likert scale was subsequently used with a broader item sample, the nonmonotonic characteristics of the item would reappear, and the validity problems inherent in measuring individuals with the most extreme attitudes would intensify. (Contains 4 figures, 2 tables and 23 references.) (Author)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A