Overcoming the 65,000 Limit of Labelling Values

[Abiye Kalaiwo]

I am currently trying to apply a value label Mien(label define Mien 1/1000 "Suppressed" 1001/10000000 "Unsuppressed" 0 "Undetectable" -1 "Failed" -2 "Invalid") to a numeric list ranging from -2 to 10000000 (count of about 46000). Unfortunately it is not working. I would appreciate some help. I noticed that there is a numeric limit of about 65,000.

* Example generated by -dataex-. To install: ssc install dataex
clear
input float Result
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-1
-1
-1
-1
-1
end
label values Result Result
label def Result -2 "Invalid", modify
label def Result -1 "Failed", modify

[Clyde Schechter]

That limit of about 65,000 distinct values for a value label is absolute. The best you can do is to create an additional variable Mien_cat that reflects the category:

gen Mien_cat = "Failed" if Mien == -1
replace Mien_cat = "Invalid" if Mien == -2
replace Mien_cat = "Undetectable" if Mien == 0
replace Mien_cat = "Suppressed" if inrange(Mien, 1, 1000)
replace Mien_cat = "Unsuppressed" if inrange(Mien, 1001, 10000000)

[Maarten Buis]

For problems like that, one typically creates a new variable, rather than giving lots of values the same label.

gen indicator = 1 if inrange(Result, 1, 1000)
replace indicator = 0 if Result > 1000 & Result < .
replace indicator = .a if Result == 0
replace indicator = .b if Result == -1
replace indicator = .c if Result == -2

label define mien 1 "Suppressed" 2 "Unsuppressed" .a "Undetectable" .b "Failed" .c "Invalid"
label value indicator mien

[Abiye Kalaiwo]

That limit of about 65,000 distinct values for a value label is absolute. The best you can do is to create an additional variable Mien_cat that reflects the category:

gen Mien_cat = "Failed" if Mien == -1
replace Mien_cat = "Invalid" if Mien == -2
replace Mien_cat = "Undetectable" if Mien == 0
replace Mien_cat = "Suppressed" if inrange(Mien, 1, 1000)
replace Mien_cat = "Unsuppressed" if inrange(Mien, 1001, 10000000)

Thanks Clyde. Exactly what I did. Was checking to see if there was a better way out there. I even tried to create loops. Do you think using a while loop would help? I was thinking of using a loop that loops through the numeric variable until it gets to the end.

[Clyde Schechter]

No, there is nothing you can do to create a value label with more than 65,536 observations. A string variable like that is the best you can do.

[Abiye Kalaiwo]

For problems like that, one typically creates a new variable, rather than giving lots of values the same label.

gen indicator = 1 if inrange(Result, 1, 1000)
replace indicator = 0 if Result > 1000 & Result < .
replace indicator = .a if Result == 0
replace indicator = .b if Result == -1
replace indicator = .c if Result == -2

label define mien 1 "Suppressed" 2 "Unsuppressed" .a "Undetectable" .b "Failed" .c "Invalid"
label value indicator mien

Thanks Maarten. My inclination to want to save the world in one single code got me asking this question. Thanks. This is helpful. Regards