Destining a string variable

[Tariq Abdullah]

Hello,

I'm trying to destring the ID variable into numeric since I need to merge it with my master data which has the same variable into numeric form. In my master data I didn't have any issue turning that string into numeric using the following command.

destring id, replace

However, in my using data despite using the same command above to turn the same variable in the using data to numeric I'm having the following issue.

My master data has the same variable like the following which I could successfully destring without any issue.

* Example generated by -dataex-. For more info, type help dataex
clear
input double id
20120246951
20200247715
20170050877
20170275197
20170197384
20170197870
20220403806
20120321483
20100291496
20230167018
end

Could you help how else I can clean it so that I can finally desiring it ? I took one step more to clean the variable. Though, it showed 75 changes , still the problem persists. I think in my using data below I need to get rid of the decimal and the trailing 0. But, I'm not sure how.

replace id = subinstr(id, ",", "", .)  // Remove commas
(75 real changes made)

destring id, replace
id: contains nonnumeric characters; no replace

* Example generated by -dataex-. For more info, type help dataex
clear
input strL id
"20210012690.0"
"20200326458.0"
"20200075033.0"
"20200075034.0"
"20200075035.0"
"20200075035.0"
"20200075036.0"
"20200075036.0"
"20200075037.0"
"20200075038.0"
end

[Nick Cox]

If the trailing .0 are redundant, then go

replace id = subinstr(id, ".0", "", .)

and then try

destring id, replace

[Maarten Buis]

The example you gave us does not produce that error message. So we cannot say anything definitive. But in general, the error message is clear: the variable id contains nonnumeric characters. That means in one or more observations (rows) there is one or more character in the variable id that Stata cannot interpret as part of a number. That could be anything, but common cases are data providers using a "/" for redacted data or "--" for missing data. So, all I can tell you is: scroll down the variable id and see what non-number character is in there, figure out why that character is there, and what you want to do with it.

[Nick Cox]

Picking up @Maarten Buis's good advice, you can find problem identifiers with

tab id if missing(real(id)), missing

[Tariq Abdullah]

Picking up @Maarten Buis's good advice, you can find problem identifiers with

tab id if missing(real(id)), missing

Mr. Cox,

Thanks so much! This helped me to a great extent. Helped me to find out what was the quirky part of this column. Much appreciated!

[Tariq Abdullah]

The example you gave us does not produce that error message. So we cannot say anything definitive. But in general, the error message is clear: the variable id contains nonnumeric characters. That means in one or more observations (rows) there is one or more character in the variable id that Stata cannot interpret as part of a number. That could be anything, but common cases are data providers using a "/" for redacted data or "--" for missing data. So, all I can tell you is: scroll down the variable id and see what non-number character is in there, figure out why that character is there, and what you want to do with it.

Thanks so much Mr. Buis for showing me the right direction to approach the problem ! Much obliged!