use or in regular expression

[Sheran Deng]

Hello,

I am trying to identify for the variable "agency" (a string variable) these observations that contain either "state" or "district". Based on what I read, I used :

gen new_variable = regexs(0) if(regexm(agency, "(state | district)"))

But I think it only identifies the observations that contain "state". I tried a few variations but did not work. Can you advise what is the right code for this. Thank you very much.

[Marco Errico]

I don't fully get your code but, looking to your first sentence, the problem should be easy solved as follow

gen wanted==.
replace wanted =1 if agency=="state" | agency=="district"

[Ali Atia]

My guess is -strpos- is more appropriate for this:

gen new_variable = strpos(agency, "state") | strpos(agency,"district")

[William Lisowski]

Your regular expression is slightly incorrect. Spaces are meaningful characters to match, so you are searching for "state " and for " district" - the latter of which will not be matched if it is at the beginning of the agency with no space character in front of the "d". Change your command to

gen new_variable = regexs(0) if(regexm(agency, "(state|district)")

and I expect it will do what you want, or at least will do what you want somewhat more often than it does now.

[Sheran Deng]

I don't fully get your code but, looking to your first sentence, the problem should be easy solved as follow

gen wanted==.
replace wanted =1 if agency=="state" | agency=="district"

Thanks, Marco, I am trying to select based on if agency contains certain words, instead of matched to certain words. otherwise i would do as you suggested

[Sheran Deng]

My guess is -strpos- is more appropriate for this:

gen new_variable = strpos(agency, "state") | strpos(agency,"district")

Thanks,Ali I will check it out.

[Sheran Deng]

Your regular expression is slightly incorrect. Spaces are meaningful characters to match, so you are searching for "state " and for " district" - the latter of which will not be matched if it is at the beginning of the agency with no space character in front of the "d". Change your command to

gen new_variable = regexs(0) if(regexm(agency, "(state|district)")

and I expect it will do what you want, or at least will do what you want somewhat more often than it does now.

Thanks William, I tried to use what you suggested but it does not identify any observations. I attached the agency column here. I had imagined that there would be a straightforward way to do OR in regular expression......

[William Lisowski]

Of course it didn't work.

The data in your spreadsheet is capitalized, for example, "U.S. Attorney-Eastern District of Pennsylvania". "district" is never going to match "District". In post #1 you told us the regular expression

"(state | district)"

matched observations that contained "state" and my code would match observations that contain "state" or "district". But not "State" or "District".

The technique from post #3 similarly will not match "State" or "District".

You are fortunate that I was able to preview your data without opening it in Excel. Like many members here, I decline to open datasets that can contain malicious code. The Statalist FAQ provides advice on effectively posing your questions, posting data, and sharing Stata output. Please take a few moments to review the FAQ to improve your future posts.

[Sheran Deng]

Of course it didn't work.

The data in your spreadsheet is capitalized, for example, "U.S. Attorney-Eastern District of Pennsylvania". "district" is never going to match "District". In post #1 you told us the regular expression

"(state | district)"

matched observations that contained "state" and my code would match observations that contain "state" or "district". But not "State" or "District".

The technique from post #3 similarly will not match "State" or "District".

You are fortunate that I was able to preview your data without opening it in Excel. Like many members here, I decline to open datasets that can contain malicious code. The Statalist FAQ provides advice on effectively posing your questions, posting data, and sharing Stata output. Please take a few moments to review the FAQ to improve your future posts.

Cool, Thanks!