Unbalanced samples

#1
Hi!

I want to compare between two sample, but they have non equivalent means between demographic parameters such as age and more.
What do you recommend to do in-order to balance between the samples and make following tests like T-test and Chi-square.

Thanks,
Hadar. :wave:
 

hlsmith

Omega Contributor
#2
So you want to compare two samples on some variables, but they differ on some other characteristics that may affect your analyses - is this correct.


Following approaches can be used:


-Match on troublesome variables;

-Control for them in a regression model;


-Use propensity scores (weights) to balance groups


If you have multiple tests, I would imagine the middle option may be best.