two continuous & two categorical variables: how to compare

[Alexei]

Dear all,

I have the following biological data:

• 1st variable (continuous, normally distributed): concentration of a hormone in blood
• 2nd variable (continuous): age of donors
• 3rd variable (categorical): genotype (control vs mutation-carrier)
• 4th variable (categorical): gender (male vs female)

I want to test whether there is a difference in the hormone levels between control donors and mutation carriers, but age and gender might bias the direct comparison using a t-test. For instance, I see that age affects the hormone levels in the control group, but not in the mutation group.

What sort of normalization/correction test shall I apply prior to comparing horomonal levels between the control and mutation groups?

Thanks a lot for helping!

Best wishes,
Alexei

University of Luebeck, Germany

 

[rogojel]

hi,
I think the best would be to simply build a regression model with one continuous and two discrete variables with interactions amongst them. That will give you all the detailed infos you need.
regards

 

[Alexei]

Thanks for your suggestion, rogojel!

What I've learned so far is that I can use the analysis of covariance (age as a covariant) with two categorical variables. However, it assumes that my continuous data can be modelled with a linear regression. But, as fas as I understood, I should first test whether linear regression is applicable for my data by analyzing residual plots. And I should test for potential outliers (measuring Mahalanobis’ distance or Cook’s D).
Am I thinking in the right direction?
Is there anything I'm still missing here?

Thank you!

 

[rogojel]

yes,
that makes sense. In order to have residuals you should first build the regression model - same for Cooks distance.
Regatds