Dear Ka Keat,

I am not experienced with the command, but whenever I get an issue with a variable like this, I would do 2 things.

First I would sort that variable (using command 'sort variableA'), and then go into the dataset and look at each end of that variable, so the first entries and the last entries- is there anything that doesn't look right- an anomaly or a missing value that has not been picked up.

The second thing I would do is get some further info about that data, maybe histogram, but in the first instance 'tabulate variableA' if it is categorical, or 'sum variableA' if it is continuous, then look is there a strange min/max value (a bit similar to the first step in some ways), or anything else that doesn't look right.

Sorry if this answer is trivial or if you have already done it, but might be something to try while someone who has used the command has chance to get back to you!

Dan

Dear Dan,

Thank you very much for your advices!

Really appreciate it.

I just tried them,

but still couldn't find any anomaly or missing values.

I only have 36 units of observations - so it will be quite obvious if there is one.

Will keep exploring and asking around.

& yes, hopefully will hear from someone who are familiar with the command or error code.

Thank you very much & have a great week ahead! =D

Regards,

Ka Keat

Thanks Amir!

I just realise the problem would only occur when I include both Variables A and B together in the same model.

If either of them is included in the model, the analysis runs.

But if both of them are included, the analysis fails and I get code r(504).

I am attaching the graph of variable (outpatient) vs variable B (noncatsurg) below.

Any clue / help is much appreciated.

Thanks!

Hi Amir,

Thank you very much for your reply!

I don't quite understand.

Can you please explain what did you mean by "dividing the variables up with a constant"?

Thanks! Appreciate your advices.

Hi Amir,

Thanks for your advice!

I tried and it worked!

But one thing odd is - my colleague tried to perform the same analysis with R and we got different results.

I think I will post this as a separate question.

Thank you very much again!

Really appreciate your help! =D

Hi everyone!

I communicated with the author of the DEA command on STATA.

and he shared with me the updates he has made and posted on Sourceforge.

https://sourceforge.net/p/deas/code/HEAD/tree/trunk/

Thanks! =)