NotesFAQContact Us
Collection
Advanced
Search Tips
ERIC Number: ED411262
Record Type: Non-Journal
Publication Date: 1997-Mar
Pages: 21
Abstractor: N/A
Reference Count: N/A
ISBN: N/A
ISSN: N/A
Assessing Reliability of Student Ratings of Advisor: A Comparison of Univariate and Multivariate Generalizability Approaches.
Sun, Anji; Valiga, Michael J.
In this study, the reliability of the American College Testing (ACT) Program's "Survey of Academic Advising" (SAA) was examined using both univariate and multivariate generalizability theory approaches. The primary purpose of the study was to compare the results of three generalizability theory models (a random univariate model, a mixed univariate model, and a multivariate model) and examine their utility for assessing the reliability of the data. Data were from SAA history files for 10 postsecondary institutions, with 10 advisors from each institution, and 10 students for each advisor. Results of the study demonstrate empirically that, when there is only one item for each level of the fixed facet in the multivariate model, the multivariate G coefficients are identical to the G coefficients produced by the mixed univariate model in which the item is a fixed facet. Although the multivariate and mixed univariate models produced higher G coefficients than did the random univariate model, the choice of using a univariate or multivariate approach should depend on the nature of the items and scales in the instrument and the assumptions underlying the universe of generalization. However, since the multivariate model (with one item per level of the fixed facet) yields the same G coefficient as the mixed univariate model (with item facet fixed) and since statistical procedures for the univariate model are simpler and more available, the mixed univariate model could be used to determine the desired level of generalizability/ reliability of the measurement data. (Contains 4 tables and 11 references.) (Author/SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A