Best statistical test to encompass all my data for my research?

Hi all,

I've been battling for months to work out which statistical test is the best way to make the most out of the data I have. I have been measuring depth related changes in a material. At depths of 2mm (0-2mm, 2mm-4mm etc) for 10 slices I have measured several quantitative parameters. I want to compare differences in these parameters at different depths and also between the different ages and sexes.

I tried PCA but that was inconclusive, we tried multiple linear regression but decided this wasn't appropriate as the measurements can't be deemed independent, we also tried two-way ANOVAs but we're unsure if this is appropriate. As you can imagine, we have a lot of data and want to make the most of it. I have been using Rstudio for my analysis.

I hope this makes sense, I'm happy to add any information that is needed! Thanks in advance


Well-Known Member

It isn't clear to me what exactly you are checking. Example table?
1. In regression, do you mean the predictors are correlated? What vif.?
2. Multiple linear regression and two-way Anova will have the same results.

Sorry, it wasn't clear, it's quite complex to explain, I'll provide an example table.

1578321537081.png 1578321578142.png
Here's just some of the data we have. So each sample, e.g. 21 y.o male, have had parameters measured from 10 different depths. From here I want to see if the parameters e.g. BVTV, TMD etc, change with depth and how this is, if it is, related to sex and age. I have used t-tests to look at age and sex-related changes separately, but is there a statistical test that will take into account depth, sex and age all together?

Sorry, I'm not quite sure what you're asking in question 1.


Well-Known Member
So in your example 3 predictors : sex, age, depth
And the rest parameters are DVs?

I asked why didn't you use linear regression

Yes, those are my predictors and the parameters are DVs.

Ah ok! Sorry, regression is new to me so I wasn't aware. When we discussed this, I was told that it wouldn't be reliable as the measurements aren't technically independent as each sample has 10 slices, so there are 60 measurements from each sample. Would this not be correct?


Hi, thanks for your reply! I'm not sure if that would work, I initially tried this, but I couldn't find a way to include sex, age and depth all as independent variables. Am I missing something?


TS Contributor
You may be thinking of a 1-way repeated measures design. I am referring to a factorial repeated measures design where sex and age are between subject factors and depth is a within subjects repeated measure.

I found a STATA example with 2 between subjects variables and 1 within subjects repeated measure.
Last edited: