Gen command replace "0 real changes made" not working for numeric variable

[Yaminette Diaz-Linhart]

Hello,

I am trying to generate a variable that categorizes departments into 3 groups.
I have successfully done this before, however, it doesn't seem to be working for this particular variable. I searched the forum/manual and did not find an answer.

Code:
gen group=.
replace group = 1 if Dept_ID==4748 | Dept_ID==4834 | Dept_ID==4580 | Dept_ID==4472 | Dept_ID==4014

Output:
. gen group=.
(3,111 missing values generated)

. replace group = 1 if Dept_ID==4748 | Dept_ID==4834 | Dept_ID==4580 | Dept_ID==4472 | Dept_ID==4014
(0 real changes made)

I then tried to use the egen command, didn't work either.

Code:
egen group1 = anycount(Dept_ID), values(4748 4834 4580 4472 4014)

tab group1, m

Output:

Dept_ID == |
4748 4834 |
4580 4472 |
4014 | Freq. Percent Cum.
------------+-----------------------------------
0 | 3,111 100.00 100.00
------------+-----------------------------------
Total | 3,111 100.00

I then tried to see if there was something different about this variable.
It is not a string variable (although I did try that as well). It is a numeric long variable and strangely says the range is 1-17. This might mean that my Dept_ID numbers must be labels (?) rather than the actual value.
Could this be the problem? I cannot figure out how to look at the underlying value...

codebook Dept_ID

---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Dept_ID (unlabeled)
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Type: Numeric (long)
Label: Dept_ID

Range: [1,17] Units: 1
Unique values: 17 Missing .: 58/3,111

Examples: 4 4683
6 4688
10 4807
14 4937

describe Dept_ID

Variable Storage Display Value
name type format label Variable label
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Dept_ID long %12.0g Dept_ID



It is my first time posting and I did my best to follow all instructions, please forgive me if I missed something.
Also note that this .dta file is downloaded from R and I can't help but feel like this might be contributing to these issues.

Thank you for any help you can provide!

[Nick Cox]

The values of that variable run from 1 to 17. The text like 4738 and 4834 constitutes value labels. There aren’t any numeric values like 4834, which is why nothing changed. See

help label

for more information.

[Nick Cox]

To add to #2, at some point earlier you or someone else applied encode to a string variable. Your string variable had (mostly) values like

"4748" "4834" "4580" "4472" "4014"

but there were also 58 missing values. The effect of encode -- with nothing else said -- is to map distinct values in alphanumeric order to integers 1 up and assign the string values as value labels.

Consider this sill example where matters are more obvious.

* Example generated by -dataex-. For more info, type help dataex
clear
input str6 fruit
"orange"
"banana"
"apple" 
"apple" 
"banana"
end

.encode fruit, gen(encoded)

. tab encoded

    encoded |      Freq.     Percent        Cum.
------------+-----------------------------------
      apple |          2       40.00       40.00
     banana |          2       40.00       80.00
     orange |          1       20.00      100.00
------------+-----------------------------------
      Total |          5      100.00

. tab encoded, nolabel

    encoded |      Freq.     Percent        Cum.
------------+-----------------------------------
          1 |          2       40.00       40.00
          2 |          2       40.00       80.00
          3 |          1       20.00      100.00
------------+-----------------------------------
      Total |          5      100.00

. label li encoded
encoded:
           1 apple
           2 banana
           3 orange

Why your data were previously string I cannot say.