NotesFAQContact Us
Collection
Advanced
Search Tips
ERIC Number: ED308240
Record Type: Non-Journal
Publication Date: 1989-Mar
Pages: 19
Abstractor: N/A
Reference Count: N/A
ISBN: N/A
ISSN: N/A
Parameters in MANOVA Robustness Studies: Comparability to Empirical Research.
Belli, Gabriella M.
An examination of articles in the last 5 years of the "American Educational Research Journal" (AERJ) was conducted to determine the extent to which choices of levels of parameters in a multivariate analysis were comparable to empirical research that uses multivariate analysis of variance (MANOVA). Originally, a main question was to determine the likelihood of homogeneity of variance and covariance in these studies. A secondary question related to the extent to which empirical researchers using MANOVA techniques reported checks on the tenability of homogeneity before reporting their multivariate results. Of the 212 articles considered, 18 used MANOVA. Six articles that used one-way fixed effects designs included 25 MANOVAs and 5 multivariate analyses of covariance (MANCOVA). Two-way designs were used in 8 articles, with 12 MANOVAs and 9 MANCOVAs. Five articles used higher-order designs in four MANOVAs and three MANCOVAs. In general, for most realistic situations, similar parameters were found in the robustness literature. When the exact design configuration was not available, one could extrapolate between the parameters that were used. Only two articles addressed the issue of robustness. Results indicate that it is imperative that researchers test for assumptions before using multivariate analyses. Three tables give the design features of studies using MANOVA and MANCOVA. The AERJ years, volumes, and pages of the 18 articles assessed are appended. (SLD)
Publication Type: Information Analyses; Reports - Research; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A