[Lei Chai]

Hi Melissa,
I am trying to run a command, but it does not work. Can you please help me out?
teffects ra (age age2 chrinhsdc EDUC5 incmc workactivity2 workactivity3 workactivity4 religion PROVINCE1 RELIG6C32 RELIG6C33 RELIG6C34) (relationop)
My dependent variable relationop is a four categorical variable (married, cohabiting, LATs and unmarried), the results automatically set married category to be the reference group, how can I change the reference group?

Thanks

Lei

[Melissa Garrido]

Hi Lei,
Try using the "control" option. After the command you've written, include something like the following:

, control (unmarried)

If this doesn't work, consider reposting this question as a brand new post so that others on Statalist are more likely to see it.

Hope this helps,
Melissa

[Ziming Liu]

Dear Melissa,
I have calculated the conditional propensity score for each value of a four-value choices (0,1,2,3) using multinominal probit model. Now I would like to perform a matching method to calculate the average treatment effect (on the treated). Is there a estimator that perform matching with an already existing propensity score please?
By the way, except the IPW method, is this a reasonable way to examine multiple treatment effects please?

Best,
Ziming

Hi Lei,
Try using the "control" option. After the command you've written, include something like the following:

, control (unmarried)

If this doesn't work, consider reposting this question as a brand new post so that others on Statalist are more likely to see it.

Hope this helps,
Melissa

[E. HABTE AWANO]

Dears Rythia, Melissa and Jeff, I read through your exchanges here on use of ipwra to estimate treatment effect when there are more than two categorical treatments. May I chip in a small related question. I have successfully estimated the model using multinomial logit for three treatment alternatives. The treatments are un-ordered. Now I am concerned if that is OK because in the stata manual it talks about mlti-valued treatment ( and gives an example of ordinal nature and doe not say anything if unordered choices apply) and I am not sure if multivalued also refers to multinomial treatment. PLEASE, HELP. Thank you.

[Yuan Chih Fu]

Dear Melissa,
thank you for providing this information about the multiple treatment effects.
I am wondering whether you know how to conduct the sensitivity analysis when the treatment effects are multiple.
I saw that we can use "rbounds" when the treatment effects are binary.
But, I can not find any command that can be used in the multiple treatment effects.
Thank you.

[Nathan Ochatum]

Dear All,
i have data captured over two seasons, the output variable is seed produced in kilograms while the input variables include: the seed planted (x1), fertilizer used (x2) manure (x3) pesticide (x4) labour (x5), land used (x6). i also have other socio economic variable like age (z1), years of schooling (z2), distance to farm (z3), household size (z4), credit recieved (z5), access to extension services-dummy (z6). i have transloged the variables and i am interested in getting technical and allocative efficiency of the farmer over two seasons. i tried to run the command xtfrontier with the option ti, after setting the data as panel. it is not yielding results. help

[Melissa Garrido]

Yuan - unfortunately, I do not know of a command that conducts senistiivty analyses for multiple treatment effects. If you post this question as a new post on Statalist, others might be able to help you.

Nathan - please see the FAQ for Statalist beginners for advice on phrasing your question so that members can assist you (http://www.statalist.org/forums/help)

[Zhang Zhe]

Dear all,I am now doing a paper,there are 4 categories in my treatment variable PT ,like
PT＝1 for receiving treatment 1 only
PT＝2 for receiving treatment2 only
PT＝3 for receiving treatment1 and 2
PT＝4 for not receiving any treatment
i think syntax such as psmatch2 or nnmatch are not suitable for the case,can anynone help me write the syntax？i
found some package about gpsocre or do－respones but i can't understand ,wish someone can write the full syntax here.. I would be very grateful for your help!

[Zhang Zhe]

Hello,

I am trying to run propensity score for multiple treatments (with 4 treatments). The manual for teffects psmatch stated that this command also works for multivariate treatments, where multinomial logit is used. However when I try to use this command as below (note: my outcome variable is pcexp, my treatment variable is treatment, and my covariates are head*, good*, and listrik):

teffects psmatch (pcexp) (treatment head* good* listrik, logit)

I get the error message below:

treatment variable treatment must have 2 levels, but 4 were found

Can anyone suggest anything or any other hints are appreciated.

Best regards,
Rythia

I am now doing a paper,there are 4 categories in my treatment variable PT ,like
PT＝1 for receiving treatment 1 only
PT＝2 for receiving treatment2 only
PT＝3 for receiving treatment1 and 2
PT＝4 for not receiving any treatment
i think syntax such as psmatch2 or nnmatch are not suitable for the case,can you help me write the syntax？
all the best
ZHANg zhe