1
$\begingroup$

Can propensity scores be used in observational studies to "level the playing field" and get "apples to apples" comparisons of groups when the groups are some characteristic like diabetes status yes/no?

I do not mean static, non-time varying characteristics like sex but instead groups that one might predict themselves from other covariates and then wish to compare between on other outcomes. Thank you.

$\endgroup$
  • $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$ – whuber Nov 22 '16 at 17:44
1
$\begingroup$

Yes; in fact, this is one of the ideal uses of propensity scores, and was used effectively in this paper. You can use propensity scores for anything you wold otherwise use regression for. Just be sure to only include covariates that truly occur before treatment. For example, current weight would be a poor variable to include, since diabetes status affects weight, but, as the authors of the linked paper note, birth weight is an acceptable variable to include. I recommend you read the linked paper to see what such a procedure would look like.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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