Regression analysis or Propensity score matching

[Rob Wood]

Hi there,

I wondered if anyone could offer some insight or provide some literature on when it is statistically 'better' to perform a regression analysis over a propensity score matching analysis and vice versa.

As far as I'm aware they both attempt to take into account covariates in one method or another, but when is one technique superior to another?

Any information on this topic would be appreciated.

Thanks,

[Carlo Lazzaro]

Robwood (please, as per FAQ, re-register with your full name and surname. Just click on the Contact us button and follow the instructions):
- your question is a bit too broad to elicit helpful replies;
- a description of a prominent, recently revised textbook on this topic can be found at http://www.stata.com/bookstore/prope...core-analysis/.

Kind regards,
Carlo

[Melissa Garrido]

Hi,
Propensity scores attempt to account for observed covariates that are related to both the treatment and the outcome, while regression analysis accounts for observed covariates that are related to the outcome (and ignores any relationship between the the covariates and the treatment).

Hope this helps,
Melissa

[Clyde Schechter]

And the two approaches are not mutually exclusive. One can use propensity scores to set up matched-tuples for regression analysis, or for weighting a regression analysis.

[Rob Wood]

Thank you all for you insight thus far, it has been very useful.

Thanks,