ANOVA or MANOVA

[Noel Carter]

Hi,

I just need some confirmation or not about this. I will try and describe my data in simple terms.

I have three groups (cell treatments) and we are measuring a state within those groups (cell death state) and there is a distribution within this state that is NOT independant ie I think they are a dependent continuous variable (early, mid, late and final cell death).

I think I am right that i need a MANOVA but just want some confirmation by anyone who is happy to help

Cheers

Noel

 

[Karabiner]

So you have got 4 subgroups in each cell treatment, i.e. cells in early/mid/etc. death status?
What is your precise research question, what do you want to find out?

With kind regards

Karabiner

 

[Noel Carter]

Hi Karabiner

The research question is do the treatments inhibit/slow down the process of cell death. So do you see differences between the treatments by seeing the distribution skewed to most of the cells being in the early stage. Does that answer your question?

Cheers
Noel

 

[Karabiner]

So for each cell you have 1 ordinal scaled measure (its stage), and no longitudinal data?
You could compare groups using the Kruskal-Wallis H-test.

With kind regards

Karabiner

 

[Noel Carter]

Hi Karabiner,

I am not sure as what you are measuring in fact is a population of cells everytime you take a measurement (say 10,000 cells per assay) and looking at the distribution of the cells at the different stages. The measurement allows you to grade the cells to define the stage. Not sure if this is continuous or not due to the set up. I guess the population has been sectioned in groups. In hindsight this might have been done differently, but the data I have is the data I have.

Noel

 

[Karabiner]

Does that mean that e.g. in group 1 there are 10,000 cells, and it is known that 3950 of them are in stage 1, and 2340 are in stage 2 etc.? This could be the basis for performing a Kruskal-Wallis H test.

With kind regards

Karabiner