I have a data of +500 samples belonging to athletic (+250) and non-athletic (+250) group. In each group I have divided the sample into Low physical activity, moderate physical activity and vigorous physical activity.
Now I want to assess the effect of different physical activity levels on dependent health parameters separately for male and female in each of athletic and non-athletic group. Is MANOVA the correct test to proceed with. Independent variables as gender and physical activity levels.
I would like to add that the no. of samples in low, moderate and vigorous physical activity are not equal or comparable. Athletic group is having minimum no. Of samples in low physical activity (around 20) and highest in vigorous physical activity group (+100). Similarly non-athletic group is having less samples in vigorous (around 10) and highest no. In low physical activity (+100).
Kindly suggest the appropriate test and link for conducting and interpreting the same.
MANOVA is used to determine an F value/statistic to find out if the means of an outcome of interest variable between two populations are significantly different. You can use the F statistic (as well as the p value), when deciding to support or reject the null hypothesis.
Whenever you have two or more dependent variables that are not independent of each other, and may influence each other to some extent, MANOVA procedures are typically used to decrease Type I error, serving as an omnibus or protective function, which many statisticians deem correctly applied and appropriate.
Help is available online for SAS/SPSS information on MANOVA, General Linear Models (GLM), and least squared means (LSM; used with unequal cell size between groups) procedures.
We actually had this same discussion exactly one year ago (Aug 24, 2020) and I walked you through it. So check out your earlier question and my responses. Be safe.
Please categorise your group depending on what you wish to interpret. Also find out the strength of each group under consideration and analysis.
Please discuss with your statistician with your data handy.
Regards
Hello @Michael C.Meyers. I conducted the MANOVA for Physical Activity level (low, moderate and vigorous) and Physical Activity category (Athletic and Non-athletic) along with gender (M and F). I found significant effects by physical activity levels, physical activity category and also by gender. Then I interpreted with the help of t-test (for athletic and non-athletic and for males and females) and contrast k matrix (for low, moderate and vigorous) to examine the significant differences in DVs. I also found an interaction effect by physical activity level nad physical activity category. I know how to report this. To study the significant differences in DVs due to interaction I used estimated marginal means. There I am getting mean, sts error and sig. Values for DVs. I am unsure about reporting sig. Differences due to this interaction effect. Like how can I write VO2 max was higher in low/mod/vig physical activity group from athletic population etc.
Alisha,
A main rule:
You should not interpret the main effects when their interaction is statistically significant unless...
The significant interaction does not affect the main effect. See:
https://www.theanalysisfactor.com
https://www.statsimprove.com
Hope this helps.
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation