3
$\begingroup$

Is it possible to compare two groups (e.g., t-test, Wilcoxon, Mann-Whitney test, etc.) while controlling for something?

E.g., I want to compare home ownership between military veterans and non-veterans, while controlling for educational attainment.

I can easily compare home ownership between military veterans and non-veterans (e.g. using the Wilcox test). But how can I do the same while controlling for educational attainment?

I think an ANCOVA does this, yes?

$\endgroup$
2
  • 2
    $\begingroup$ Use linear regression? $\endgroup$
    – Repmat
    Commented Jul 22, 2016 at 18:06
  • 1
    $\begingroup$ @Repmat Ah -- you're right... just have military status and education as independent variables, and the beta coefficient for each independent variable will automatically control for the other independent variable. Thanks! $\endgroup$ Commented Jul 22, 2016 at 18:19

1 Answer 1

1
$\begingroup$

Given that you have a large enough sample size, you can subset the data by educational attainment, and then compare vets vs. non-vets.

Example:

1) no HS: vets vs. non-vets

2) HS only: vets vs. non-vets

3) some college: vets vs. non-vets

.

.

.

$\endgroup$
2
  • $\begingroup$ Thanks! How about ANCOVA -- would that be okay too? $\endgroup$ Commented Jul 22, 2016 at 18:39
  • 1
    $\begingroup$ I think the way I described is simpler and easier to interpret... no need to overcomplicate :) $\endgroup$ Commented Jul 28, 2016 at 17:34

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.