Multivariate analysis versus multiple univariate analyses.

Abstract

The argument for preceding multiple analysis of variance (ANOVAS) with a multivariate analysis of variance (MANOVA) to control for Type I error is challenged. Several situations are discussed in which multiple ANOVAS might be conducted without the necessity of a preliminary MANOVA. Three reasons for considering a multivariate analysis are discussed: to identify outcome variable system constructs, to select variable subsets, and to determine variable relative worth. The analyses discussed in this article are those appropriate in research situations in which analysis of variance techniques are useful. These analyses are used to study the effects of treatment variables on outcome/response variables (in ex post facto as well as experimental studies). We speak of an univariate analysis of variance (ANOVA) when a single outcome variable is involved; when multiple outcome variables are involved, it is a multivariate analysis of variance (MANOVA). (Covariance analyses may also be included.) With multiple outcome variables, the typical analysis approach used in the group-comparison context, at least in the behavioral sciences, is to either (a) conduct multiple ANOVAs or (b) conduct a MANOVA followed by multiple ANOVAS. That these are two popular choices may be concluded from a survey of some prominent behavioral science journals. The 1986 issues of five journals published by the American Psychological Association were surveyed:

Response strategies in an international strategic alliance experimental context: Cross-country differences

“…Third, to examine the direct and moderating effects of country differences and test the hypotheses, we conducted a multivariate analysis of covariance (MANCOVA), followed by a series of post-hoc country comparisons and F-tests. Because the response strategies were interrelated, we manipulated the scenario variables, and we used covariates to control for confounding effects, a MANCOVA was the most appropriate method (Huberty and Morris, 1989). Fourth, to examine the external validity of the findings, we collected additional data from alliance managers in the Netherlands and assessed the measurement equivalence and similarity of their responses with those from the business students in our Dutch sample.…”

Health status of Largescale Sucker (Catostomus macrocheilus) collected along an organic contaminant gradient in the lower Columbia River, Oregon and Washington, USA

“…It is generally inappropriate to conduct univariate analyses following multivariate treatment of the same data (Huberty and Morris, 1989). However, knowledge of site-dependent patterns in the individual variables used for multivariate analyses can be useful to provide specific context.…”

“…This type of questions is best addressed using multivariate statistical approaches (Huberty and Morris, 1989). Therefore, MANOVA was used to determine overall site-associated differences in continuous variables for males and females in each year and in the contaminant content of male livers in 2010.…”

Undergraduate caregivers for individuals with chronic conditions: Stressors and needs

“…Responses for each item are summed to provide a score between 0 and 27 [9]. Scores are grouped into ranges: no depression (0-4); mild depression (5-9); moderate depression (10)(11)(12)(13)(14); moderately severe depression (15)(16)(17)(18)(19); and severe depression (20)(21)(22)(23)(24)(25)(26)(27). The PHQ-9 has been shown to have excellent internal consistency (α's .86 to .89) in validation studies, as well as in the current sample (α =.88).…”