College of Doctoral Studies
RES-845: Module 8 Problem Set Solutions
Factorial (2 x 3) MANOVA
1.
Is there a sufficient co
elation between the dependent variables to justify the use of MANOVA?
YES! THE DEPENDENT VARIABLES ARE BOTH CONCEPTUALLY AND STATISTICALLY (r = .513) RELATED.
2.
Was the assumption of Equality of Covariance Matrices violated? Explain.
NO! RESULT OF THE BOX'S TEST OF EQUALITY OF COVARIANCE MATRICES INDICATED NO VIOLATION (p = .463).
3.
Is there a statistically significant multivariate interaction effect?
YES!
Identify the dependent variable(s) of this interaction effect.
EMOTION ONLY
4.
What would be the proper follow-up tests for a statistically significant interaction effect?
CONDUCT TWO SEPARATE ONE-WAY ANOVAs WITH TREATMENT AS THE INDEPENDENT VARIABLE FOR MALES AND FEMALES. IF THE ONE-WAY ANOVAs REPORT A STATISTICALLY SIGNIFICANT OMNIBUS, THEN PERFORM THE APPROPRIATE POST-HOC.
5.
Identify the proper post hoc analyses for any statistically significant univariate effects. Explain your answer.
THE LSD IS ONE OF THE PROPER POST-HOCS BECAUSE THE EQUAL VARIANCES ASSUMPTION IS NOT VIOLATED.
6.
Is there a statistically significant multivariate gender effect on the dependent variate?
YES! BUT BECAUSE THERE IS AN INTERACTION EFFECT, THIS MAIN EFFECT IS OF LITTLE INTEREST.
7.
Why would a researcher conduct a MANOVA instead of several ANOVAs?
THERE ARE AT LEAST TWO REASONS TO CONDUCT A MANOVA INSTEAD OF A SERIES OF ANOVAs. (1) MANOVA IS A MORE POWERFUL STATISTICAL TECHNIQUE (I.E., IT IS BETTER ABLE TO DETECT DIFFERENCES IF THEY REALLY EXIST), (2) MANOVA CONTROLS FOR AN INFLATED TYPE I ERROR.
8.
Write a Results section for this research.
Co
elations
General Linear Model
Box's Test of Equality of Covariance Matricesa
Multivariate Tests c
Levene's Test of Equality of E
or Variances a
Tests of Between-Subjects Effects
General Linear Model
1. Treatment
2. Gender
3. Treatment * Gender
Univariate Analysis of Variance
Tests of Between-Subjects Effects
Estimated Marginal Means
1. Treatment
2. Gender
3. Treatment * Gender
Univariate Analysis of Variance for MALES
Tests of Between-Subjects Effects
Estimated Marginal Means
Treatment
Post Hoc Tests
Multiple Comparisons
Univariate Analysis of Variance for FEMALES
Tests of Between-Subjects Effects
Estimated Marginal Means
Treatment
Post Hoc Tests Treatment
Multiple Comparisons
Univariate Analysis of Variance for TREATMENT Main Effect
Tests of Between-Subjects Effects
Estimated Marginal Means
Treatment
Post Hoc Tests
Treatment
© 2012. Grand Canyon University. All Rights Reserved.