Question regarding loop syntax

[Tim Brooks]

Hi, I'm new to Stata and still figuring out loops.

I'm creating a loop to clean two datasets. First I'm trying to rename some variables, then recode bmi into two categories. I am then trying to extract a substring from patient id to create an id with just numbers. I'm then saving the dataset and doing some checking.

My issue is that when I check the cleaned dataset, while the recoded bmi. has been changed in the dataset and the variable renaming has changed, the new variable (pt_id) has been added to the dataset but it has no values.

Can anyone suggest what I'm doing wrong?

Thanks!

* Define datasets BP1 and BP2
local datasets "BP1.dta BP2.dta"

* Loop through each dataset
foreach dataset in `datasets' {

* Load the dataset
use `dataset', clear

* Rename variables
rename age_group agegp
rename bmi_group bmigp
rename employment_group empgp
rename ses_group sesgp


* Recode bmi level into two groups to match the dataset description
replace bmigp = "0" if bmigp == "underweight" | bmigp == "normal"
replace bmigp = "1" if bmigp == "overweight" | bmigp == "obese"

* Extract numeric part from patient_id
gen pt_id = substr(patient_id, 2, .)
* Verify the extraction worked
list id pt_id in 1/10

* Save the cleaned dataset
local cleaned_filename = "`dataset'_cleaned"
save `cleaned_filename', replace
display "Saved cleaned dataset: `cleaned_filename'.dta"
*Summarize and check dataset
summarize


* Run codebook on bmigp
display "Details for variable bmigp in `dataset'"
codebook bmigp
}

[Nick Cox]

gen pt_id = substr(patient_id, 2, .)

will return an empty string if patient_id is just one character long.

Otherwise here are some suggestions on your code. I created a numeric (0, 1) indicator. which is much more useful than a ("0", "1") indicator.

* Loop through each dataset
foreach dataset in BP1 BP2 {

* Load the dataset
use `dataset', clear

* Rename variables
rename (age_group employment_group ses_group) (agegp empgp sesgp) 

* Recode bmi level into two groups to match the dataset description
gen bmigp == 0 if inlist(bmi_group, "underweight", "normal")
replace bmigp = 1 if inlist(bmi_group, "overweight", "obese"), 

* Extract numeric part from patient_id
gen pt_id = substr(patient_id, 2, .)
* Verify the extraction worked
list id pt_id in 1/10

* Save the cleaned dataset
save "`dataset'_cleaned", replace
display "Saved cleaned dataset: `cleaned_filename'.dta"

*Summarize and check dataset
summarize

* Run codebook on bmigp
display "Details for variable bmigp in `dataset'"
codebook bmigp
}

[Nick Cox]

I should have flagged here my interpretation: although the question is posed as being about loop syntax, that doesn't seem to be the issue, which you did flag concisely:

the new variable (pt_id) has been added to the dataset but it has no values

I take that to mean empty strings. Even your identifiers may be sensitive or confidential, and if so you shouldn't post them here, but I think this boils down to looking more carefully at your identifiers. Why do you want to remove the first character anyway?

Fixed some small errors below.

* Loop through each dataset
foreach dataset in BP1 BP2 {

* Load the dataset
use `dataset', clear

* Rename variables
rename (age_group employment_group ses_group) (agegp empgp sesgp)

* Recode bmi level into two groups to match the dataset description
gen bmigp == 0 if inlist(bmi_group, "underweight", "normal")
replace bmigp = 1 if inlist(bmi_group, "overweight", "obese")

* Extract numeric part from patient_id
gen pt_id = substr(patient_id, 2, .)
* Verify the extraction worked
list id pt_id in 1/10

* Save the cleaned dataset
save "`dataset'_cleaned", replace
display "Saved cleaned dataset: `dataset'_cleaned.dta"

*Summarize and check dataset
summarize

* Run codebook on bmigp
display "Details for variable bmigp in `dataset'"
codebook bmigp

}