0
$\begingroup$

I am controlling for education-level (two dummy variables) and another covariate in my regression model. I'm conducting a hierarchical regression analysis. I was confused about how I should be entering all the predictors. Do I enter the dummy variables and other covariate in the first block and then the main predictors in the next block? Or do I enter the dummy variables in the first block, the other covariate in the second block and the main predictors in the third block? Thank you!

$\endgroup$

1 Answer 1

1
$\begingroup$

It depends on what you want to find out, either choice could be correct.

If you are interested in the effects of a) The dummy variable b) The covariate, controlling for the dummy variables and c) the main predictors, controlling for the dummy variable and covariate then you would do the regression in 3 stages.

On the other hand, if you are interested in the effects of a) The dummy variables and covariate and b) the main predictors, controlling for the dummy variable and covariate, then you could do it in 2 stages.

And if you are only interested in the main predictors, controlling for the dummy variable and covariate, you could do it in 2 stages as well.

$\endgroup$
1
  • $\begingroup$ And one stage is often the best - by pre-specifying the entire model and conditioning on everything in the right causal pathway. Note that the original question implicitly assumed no heterogeneity within the 2 education levels, an assumption that is quite strange. $\endgroup$ Aug 31, 2013 at 14:58

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.