There are three DVs (Job Satisfaction, Work Engagement and Job Search Behavior) whereas one IV (Education) having five groups . Keeping in view research questions, I found differences of means among groups in MANOVA executed through IBM SPSS V 20. The values of Box's test and Leven''s test were highly significant P
Hello Muhammad,
You might try transforming each of the scores to help stabilize variances across your groups. That should help, at least somewhat. One problem with Box test is that it is also sensitive to non-normality (not just heterogeneity of covariance matrices). So, do inspect data sets (start separately by group and by variable) for whether the assumption of normality is plausible. Transformation can help here, as well.
With unequal group sizes (as in your data set), the usual concern about heterogeneity of covariance matrices goes something like this: If group n relates positively to group variance, then F-test becomes too liberal (adjust by adopting a lower alpha level for hypothesis testing). If group n is negatively related to group variance, then F-test becomes too conservative (adjust by adopting higher alpha level, or just acknowledge that statistical power will suffer). In your data set, the patterns are inconsistent and depend on the variable examined.
So, your options are:
1. Try transforming scores (needn't use same transformation for each variable, but do apply for all groups). Re-check normality and homogeneity.
2. Be conservative, and use a smaller alpha level (e.g., .01, .005, or .001 instead of .05).
3. Pray for robustness of MANOVA (e.g., hope that the results are accurate despite data not conforming to assumptions).
4. Unfortunately, hybrid techniques (somewhere between the strictly parametric and non-parametric), such as that of Puri and Sen or Conover and Iman, also require homogeneity. Here are two citations to refer to:
Puri, M. L., and P. K. Sen (1971). Nonparametric methods in multivariate analysis. New York, NY: John Wiley & Sons. doi:10.1002/bimj.19740160414
Conover, W. L., & Iman, R. L. (1981). Rank transformations as a bridge between parametric and nonparametric statistics. The American Statistician, 35, 124-129. doi: 10.1080/00031305.1981.10479327
Good luck with your work!
Dear Muhammad,
I created a movie several years ago. You can get some knowledge about how to run and report MANOVA as well.
Additionally, there is a book in which you can find how to write the report:
Morgan, S., Reichert, T., & Harrison, T. R. (2016). From numbers to words: Reporting statistical results for the social sciences. Routledge.
Wish this helps.
Best,
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA