Archive:MANOVA: Difference between revisions

wikademia>Jtneill
wikademia>Wikademia
 
(9 intermediate revisions by 2 users not shown)
Line 1: Line 1:
{{RoundBoxTop|theme=2}}
{{ntnes}}
{{tutorial}}
{{50%done}}
* The purpose of this tutorial is to teach use of '''multivariate analysis of variance (MANOVA)''', with practical exercises based on using [[SPSS]].  
* The purpose of this tutorial is to teach use of '''multivariate analysis of variance (MANOVA)''', with practical exercises based on using [[SPSS]].  
* Note that the MANOVA procedure is not available with the Studentware version of SPSS.
* Note that the MANOVA procedure is not available with the Studentware version of SPSS.
{{RoundBoxBottom}}
{{TOCright}}
==What is MANOVA?==
==What is MANOVA?==
* Developed as a theoretical construct by [[w:Samuel S. Wilks|Samual S. Wilks]] in 1932 (Biometrika).
* Developed as a theoretical construct by [[Wikipedia:Samuel S. Wilks|Samual S. Wilks]] in 1932 (Biometrika).
* An extension of univariate [[ANOVA]] procedures to situations in which there are two or more related [[DV|dependent variables]] (ANOVA analyses only a single DV at a time). DVs should be correlated (but not overly so; otherwise they should be combined) or conceptually related.
* An extension of univariate [[ANOVA]] procedures to situations in which there are two or more related [[DV|dependent variables]] (ANOVA analyses only a single DV at a time). DVs should be correlated (but not overly so; otherwise they should be combined) or conceptually related.
* The MANOVA procedure identifies (inferentially) whether:
* The MANOVA procedure identifies (inferentially) whether:
Line 76: Line 72:
#* The ''F'' test from Box’s ''M'' statistics should be interpreted cautiously because it is a highly sensitive test of the violation of the multivariate normality assumption, particularly with large sample sizes.
#* The ''F'' test from Box’s ''M'' statistics should be interpreted cautiously because it is a highly sensitive test of the violation of the multivariate normality assumption, particularly with large sample sizes.
#* MANOVA is fairly robust to this assumption where there are equal sample sizes for each cell.
#* MANOVA is fairly robust to this assumption where there are equal sample sizes for each cell.
# '''Homogeneity of error variances (Levene's test)'''
#* If this assumption is violated, use a more conservative critical <math>/alpha</math> level for determining significance for that variable in the univariate ''F''-test. Tabachnick and Fidell suggest .025 or .01 rather than the conventional .05 level.


==Test statistics==
==Multivariate test statistics==
Choose from among these test statistics to assess differences across the levels of the IV(s) for the DVs:
Choose from among these multivariate test statistics to assess whether there are statistically significant differences across the levels of the IV(s) for a linear combination of DVs. In general Wilks' <math>/lambda</math> is recommended unless there are problems with small ''N'', unequal ''n''s, violations of assumptions, etc. in which case Pillai's trace is more robust (Tabachnick & Fidell):
;Roy's greatest characteristic root
;Roy's greatest characteristic root
# Tests for differences on only the first discriminant function
# Tests for differences on only the first discriminant function
Line 95: Line 93:
# Considers differences over all the characteristic roots
# Considers differences over all the characteristic roots
# More robust than Wilks'; should be used when sample size decreases, unequal cell sizes or homogeneity of covariances is violated
# More robust than Wilks'; should be used when sample size decreases, unequal cell sizes or homogeneity of covariances is violated
==Tests of between-subject effects==
* If there is a significant multivariate effect, examine the '''Tests of Between-Subjects Effects''' for each of the DVs.
* Since there are multiple tests, control for the Type I error-rate (e.g., use a Bonferroni adjusment - divide the original alpha level by the number of tests).


==Effect sizes==
==Effect sizes==
Also use [[effect size]]s to evaluate strength of the effects (particularly for significant effects):
Also use [[effect size]]s to evaluate strength of the effects (particularly for significant effects):
* Univariate ANOVA: <math>\eta^2</math> gives the proportion of variance in the DV that is attributable to different levels of an IV
* Multivariate ANOVA:
* Multivariate ANOVA:  Wilks' <math>\lambda</math> - multivariate <math>\eta^2</math>: Wilks' <math>\lambda</math> reflects the ratio of within-group variance across all discriminant functions to total variance across all discriminant functions.
** Wilks' <math>\lambda</math> - multivariate <math>\eta^2</math>: Wilks' <math>\lambda</math> reflects the ratio of within-group variance across all discriminant functions to total variance across all discriminant functions.
* Univariate ANOVA:
** <math>\eta^2</math> gives the proportion of variance in the DV that is attributable to different levels of an IV


==Follow-up testing==
==Follow-up testing==