Psmatch2 and next steps for survival analysis

[Arnaud Lesegretain]

Hello,

I have run psmatch2 on my observational dataset and would appreciate some guidance on how to use the new variables created by the psmatch2 command for ensuring survival regression analysis.

psmatch2 raceb i.aged i.sex i.geogr, logit odds neighbor(3) ai(3)

- How can I can identify the matched dataset so that I can re-use for my analysis purposes?

- For example, for COX regression and log-rank test for my treatment variable?
* Do i need to stset to a subset of the dataset? based on which variable?
* Should i stratify stcox ? according to which variable? _pscore?
- For the log-rank test, should i stratify according to _pscore?

many thanks
Arnaud

[David Radwin]

For your first question, see https://www.statalist.org/forums/for...94#post1645694

[Anushiya Thanapalan]

I am using a stochastic frontier approach to estimate the profit efficiency. But I am not sure how to estimate the marginal effect of x(inefficiency variables) variable on inefficiency E(u)

[David Radwin]

I am using a stochastic frontier approach to estimate the profit efficiency. But I am not sure how to estimate the marginal effect of x(inefficiency variables) variable on inefficiency E(u)

It sounds like this message is for a different topic, perhaps by mistake. In any case, please follow Statalist advice and start a new thread.

[Arnaud Lesegretain]

Hello David,
Is this correct that psmatch2 does not allow 1-n matching if one choses the "no replacement" option? (I have a error message from Stata)

[David Radwin]

Apparently not. I got an error message, too.

. sysuse nlsw88
(NLSW, 1988 extract)

. psmatch2 union age married south grade tenure, outcome(wage) neighbor(5) noreplace
Matching without replacement is only implemented with 1-to-1 propensity score matching
r(198);