Can i use MANOVA to determine whether there are any differences between (3 categories of a single study group) on 4 continuous dependent variable?
Dear Ragab,
I think the answer is yes, but you should check the test assumptions first.
Hi Ragab, Following link of IDRE, University of California has very helpful table to decide which statistical test you need to use.
http://www.ats.ucla.edu/stat/mult_pkg/whatstat/
Many other decision tree like the one below may also help.
https://statswithcats.wordpress.com/2010/08/27/the-right-tool-for-the-job/
Regards
Dear Peter Samuels and Arif Ulhaq
Thank you in advance for trying to help me. Indeed, I'm bit confused. In my study, i have one independent variable (that can be polychotomized) and one dependent variable that has been measured at 4 intervals and i want to predict changes of the dependent variable over time in a short term cohort study.
in MANOVA, we can only estimate the mean differences of the dependent variable at each follow up point when the independent variable is polychotomized to three categories. But, we can't make prediction how much would be the changes of the dependent variable for each increased unit of the dependent variable.
My question is: can MANOVA be used instead of regression analysis in this case just to clarify the changes over the follow up period without prediction?
if not? what is the regression model that fit this design?
Regards
Multivariate Analysis of Variance (MANOVA) simultaneously analyzes the
relationship of 2 or more metric dependent variables and several non‐metric independent variables. A related procedure is multivariate analysis of covariance (MANCOVA) which can be used to control factors other than the included independent variables.
Do you have enough replications?
Did you perform an experimental design?
Thank you Carlos for your interest. the study design is "a prospective cohort". no intervention has been implemented. I just follow up the recovery over time and the assessment of the dependent variables was carried out at 4 intervals.
Ragab,
what you comment corresponds to the design of the research and not to the experimental design.
For what you mention
Gx: O1 - O2 - O3 - O4
The problem for performing a multivariate analysis is the number of replicates per treatment. according to the literature it is at least 20 units.
Hi Carlos,
What exactly to you mean by replication in this case? The number of combinations that has to be tested?
I'm not sure that I got what do you mean. But, my answer would be 3 factors at 4 intervals and 30 replicates.
I have overlooked many websites and I knew that i can determine the number of experiments I will do by multiplying 3X4X n, where n is the number of replications.
- Badges
- Science topic
- Similar topics
- Statistics
- Time Series