0
$\begingroup$

I have a dataset of 90 individuals with 12 binary variables, divided in three groups of the same size.

I want to know if two if these groups have differences with the other one (the control group), which is the best way to do this?

Is it correct to use PCA to do simple ANOVA?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
0
$\begingroup$

Depending on you want to go get.. I would do Anova to detect the behaviour beetwen groups and within the group...

PCA is more useful to reduce the dimension, in some point it doesnt make too much sense to use this method... of course with an homogeneous group with PCA you can get one component but it is not too much useful if you want to get the measure of variance and so on..

Improve this answer
$\endgroup$
2
  • $\begingroup$ I mean using PCA to reduce the dimension to 1, and then use ANOVA with this variable to check if exists any differences between groups $\endgroup$
    – roodry67
    Commented Aug 27, 2016 at 12:17
  • $\begingroup$ You can try it, at the end anova is one tool to confirm that you get homogeneous group( it will confirm that you did a good PCA work) it make sense more in clustering analysis, I will do a Reliability test to make sure you got a good result. $\endgroup$
    – PeCa
    Commented Aug 28, 2016 at 21:17

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.