0
$\begingroup$

suppose I want to incorporate propensity score matching in analyzing sales. Last year, I sold 100 of 300, so my ratio is 33.33%. This year my items costs 5% less and I sold 300/600, so my ratio is 50%.

this year and last year could potentially have different covariates, so it could be anything from customer gender, age, etc etc.

what I want to do is, relative to the price, I want to use propensity score to balance the covariates, and get a sense of the impact of the 5% price decrease. So for example, maybe after balancing my covariates, I find out that my 300/600 is actually just 200/600.

In this case,what is my treatment and what is my outcome when I put this in R/Python? My understanding is that the general definition of PSM is the probability of treament given the covariate. So is my treatment "sold this year"? But I thought "sold" would be my outcome variable

would appreciate some insight.

thanks!

$\endgroup$

1 Answer 1

0
$\begingroup$

Your treatment is whether or not the price decrease was active, and your outcome is whether or not an item was sold. Note that propensity score matching is not a very good technique for this kind of data. I urge you to consider regression or weighting instead as these don't involve discarding data.

$\endgroup$

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.