Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
The psychometric utility of the midpoint on a Likert scale
Abstract
Among Likert scale research spanning a number of disciplines are studies reaching conflicting conclusions about the theoretical and psychometric necessity of the midpoint. Typically, midpoint responders have been classified as "ambivalent," "indifferent," and "uncertain." Theoretically, a midpoint response should represent ambivalent attitudes--those who are familiar with what they are rating and have equal feelings of agreement and disagreement. Item Response Theory (IRT) research has not supported the utility of the midpoint. The theory underlying this dissertation is that by providing a second nondirectional category (Don't Know), the midpoint (Undecided) would function more like a theoretical true midpoint, thereby improving scale properties. Hypotheses were studied using Bock's (1972) Nominal Response and Samejima's (1969) Graded Response IRT models. Comparisons of option characteristic curves, discrimination parameters, information functions, relative efficiency, and model fit were made between Scale A (5 alternatives with additional sixth Don't Know alternative) and Scale B (5 alternatives) across 18 items in two demographically similar independent samples. Findings generally supported the hypothesis that providing a sixth Don't Know alternative improves the operating characteristics of the middle category, as well as other categories. Scale A items tended to have discrimination parameters and scale information exceeding that of Scale B, particularly on items measuring general attitudes as opposed to specific behavioral outcomes. The Undecided category had highest probabilities at low negative to middle levels of theta in both scales, and Undecided category Option Characteristic Curves tended to peak above that of other categories, more so on Scale A. Both nondirectional categories appeared to be scale dependent. The Don't Know response option consistently peaked to the left of the scale. Implications are that by using IRT to understand the properties of different options, psychometricians can develop more efficient measures of attitude and practitioners can gain a more precise understanding of the meaning attached to different options. Limitations and future research are discussed.
Subject Area
Educational psychology|Psychological tests|Educational evaluation
Recommended Citation
Harter, James Kevin, "The psychometric utility of the midpoint on a Likert scale" (1997). ETD collection for University of Nebraska-Lincoln. AAI9730273.
https://digitalcommons.unl.edu/dissertations/AAI9730273