Splitting a string using parse

[Hamza Khawar]

I have the following variable "location" which contains the city and state name. I need to split it into a new variable which contains only the city name since I already have the state names.

I have used the following command to split the data:

. split location, parse(,) generate(temp)

Unfortunately, some of the data points such as the 1st one have additional data (18th and vine) which are not required but result in the split data mismatch (I require Kansas City to be in the first column)

How can I split the data so that I can extract only the city name?

Thanks

[Clyde Schechter]

Since you show your data as a screen shot (which in the FAQ you are specifically told not to do) I can't import your example into Stata to test this code. So it may contain errors.

gen rloc = reverse(location)
split rloc, parse(",") gen(temp)
replace temp2 = reverse(temp2)

As I say, not tested because you made it impossible to do so, but I believe this will give you the city in temp2.

There is probably a way to do this with regular expressions as well, though I don't have a good grasp of those.

Please before posting again make sure you read FAQ #12 and in the future follow those instructions for posting example data (the -dataex- command) and code or Stata output (code delimiters). These are the optimal ways to show these things and they facilitate easy transfer of information between questioners and responders.

[Hamza Khawar]

I apologize for not posting the data properly. Thank you for your help though.

[Robert Kogi]

Hi family,
Please help me resolve this. I am trying to split my data which has multiple responses by using the following commnds:
split HowcanonegetinfectedwithHI
split HowcanonegetinfectedwithHI, parse(",") gen (HIVtransmission)
split HowcanonegetinfectedwithHI, parse(",") gen (HIVmodeoftransmission) notrim
split HowcanonegetinfectedwithHI, parse(",") gen (HIVmodeoftransmi) limit(8)
split HowcanonegetinfectedwithHI, parse(",") gen (HIVmodeoftransmi2) destring
tab HIVmodeoftransmi2

The results I obtained had one of the responses missing, and this gave me 361 instead of my total sample size of 403:

HIVmodeoftransmi2 Freq. Percent Cum.

From infected mother to child 20 5.54 5.54
Having unprotected sex with an infec.. 228 63.16 68.70
Kissing an infected person 8 2.22 70.91
Sharing sharps with an infected person 67 18.56 89.47
Through curse 18 4.99 94.46
witchcraft 20 5.54 100.00

Total 361 100.00

[Nick Cox]

I think we need to see the original data that are not shown here by say the results of

tab HowcanonegetinfectedwithHI if missing(HIVmodeoftransmi2), missing

[Robert Kogi]

Thank you Nick. When I type the command, the missing respondents showed up with their various categories. Is there a way I can now add them up to their respective parents categories in stata before running the analysis?

[Nick Cox]

Not trying to be flippant, but the answer to #6 is likely to be Yes, but as in #5 you need to show us those results so that we can advise.