Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
ORCID
https://orcid.org/0000-0001-5581-7884
Date of Graduation
Spring 2018
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
Department
Department of Graduate Psychology
Advisor(s)
Deborah Bandalos
Abstract
Attitude scales are an important component of educational and psychological research. One consideration when seeking to make valid inferences from attitudinal data is the issue of the degree to which response options can be assumed to have equal intervals. Many response options on attitudinal measures may produce ordinal-level data rather than interval. This poses a problem for the statistical tests that may be used, as many analyses assume interval-level data. It also poses an interpretational issue if the conceptual distance between response options is not the same – for example, if a researcher believes that someone who answered Agree differs the same amount from someone who answers Strongly Agree as they do from someone who answers Disagree when this may not actually be the case. As a result of the importance of equal-interval response scales, in this study I sought to design a set of equal interval Agree/Disagree Likert-type response options.
To develop these response options, I first asked several hundred undergraduates to assign percentages to a series of Agree/Disagree Likert scale modifiers. I used the median percentages to create a set of equal response options as well as a set of unequal options for comparison. Next, I attached these response options to measures of Mindfulness, Conscientiousness, and Agreeableness and collected data from approximately 2,100 respondents (approximately 1850 completed the equal versions, and approximately 1850 completed the unequal versions). To assess the equal-interval nature of the data, I used the polytomous IRT graded response model to compare spacing between category boundary locations. Equidistance between the category boundary locations would provide evidence that respondents were treating the response scale as equal interval.
Based on the spacing between category boundary locations, the equal response options did not produce data that was equally spaced in an absolute sense. Additionally, they did not produce data that was substantially more equally spaced than the unequal options. Based on these results, response category wording may not make a difference in the spacing of category boundary locations. However, as this was just one study with a limited population, more research in this area is needed.
Recommended Citation
Spratto, Elisabeth M., "In search of equality: Developing an equal interval Likert response scale" (2018). Dissertations, 2014-2019. 172.
https://commons.lib.jmu.edu/diss201019/172
