Identifying values that differ for a given group

[Erika Kociolek]

I have a dataset that looks something like this:

ID var1 flag_same
1 3 0
1 3 0
1 2 0
1 2 0
1 2 0
2 1 0
2 2 0
3 2 1
3 2 1

I'd like to create a variable that indicates when the values of var1 for a given value of ID are completely the same - i.e., generate the variable flag_same. Any ideas about how to do this would be so helpful. It would be great if this could work for both numeric and string variables.

Best,
Erika

[Roman Mostazir]

Its not clear what the 'flag_same' variable will contain if the condition matches. The following command will create 'flag_same' with 1=where ID and var1 mathces for first observation and 'zero' for the rest of the matches.

egen flag_same = tag(ID var1)
lis, clean noobs

 ID   var1   flag_same  
     1      3          1  
     1      3          0  
     1      2          1  
     1      2          0  
     1      2          0  
     2      1          1  
     2      2          1  
     3      2          1  
     3      2          0

[Erika Kociolek]

In the fake dataset above, ID 3 appears twice in the dataset. The two values of var1 for ID 3 are the same, so I'd like to create a variable, flag_same, which is equal to 1 for both of those records. For the other two IDs, 1 and 2, the values of var1 are different, so I'd like the variable flag_same to be equal to 0 for all of those records. Hopefully that makes sense.

[Roman Mostazir]

The following code is based on your ID ranges from 1 to 3. Replace the values according to your data:

gen flag =.

forval i=1/3 { //Change the value here based on your ID
su var1 if ID==`i'
replace flag = 1 if ID==`i' & var1==r(mean)
}
replace flag = 0 if flag==. & !missing(var1) //This ensures that if you have a value missing
for var1, you will not get a value in 'flag'

lis, clean noobs

   ID   var1   flag  
     1      3      0  
     1      3      0  
     1      2      0  
     1      2      0  
     1      2      0  
     2      1      0  
     2      2      0  
     3      2      1  
     3      2      1

PS: There may be other easier ways of doing it, but that is what coming to my mind at the moment.

[Nick Cox]

This is an FAQ: http://www.stata.com/support/faqs/da...ions-in-group/

There is an one-line solution, but anything shorter would appeal too.

 
bysort ID (var1): gen  flag_same = var[1] == var[_N]

[Erika Kociolek]

Thank you, Nick and Roman.

Nick - your solution works perfectly. Thanks!