然后使用egen命令循环遍历变量的级别

Question

I want to create a new variable that counts all the occurrences of a specified drug code over 27 columns. However, I want to do this for all drug codes that I have available. I am using a loop and the egen function anycount() .

However, when I have tried to run it I get the error:

invalid name
r(198);

Below you can find an example of my data:

clear
input DrugList Drug1 Drug2 Drug3 Drug4 Drug5
1234 7934 1234 . . .
5678 1234 5678 . . .
9876 9876 3456 . . .
3456 9876 . . . .
7934 9876 5678 7934 1234 .
17453 5678 . . . .
end

The code I am using is the following:

levelsof DrugList

foreach drug in `r(levels)'{
    egen d_`drug' = anycount(Drug1-Drug27), values(`drug')
}

The output I expect is this:

  +------------------------------------------------------+
  | d_1234   d_3456   d_5678   d_7934   d_9876   d_17453 |
  |------------------------------------------------------|
  |      1        0        0        1        0         0 |
  |      1        0        1        0        0         0 |
  |      0        1        0        0        1         0 |
  |      0        0        0        0        1         0 |
  |      1        0        1        1        1         0 |
  |      0        0        1        0        0         0 |
  +------------------------------------------------------+

If I run the code without a loop for a specific code then everything works:

egen d_1234 = anycount(Drug1-Drug27), values(1234)

What am I doing wrong?

Answer 1

The problem is that in your original data, you have drug codes in DrugList that contain decimals, negative values or even both. Stata does not consider these legal for variable names.

Consider the following toy example based on a slightly modified version of the data in your question:

clear
input DrugList Drug1 Drug2 Drug3 Drug4 Drug5
1234 7934 1234 . . .
5678 1234 5678 . . .
9876 9876 3456 . . .
3456.46 9876 . . . .
7934 9876 5678 7934 1234 .
17453 5678 . . . .
end

levelsof DrugList, clean

foreach drug in `r(levels)' {
    egen d_`drug' = anycount(Drug1-Drug5), values(`drug')
}
d_3456.4599609375 invalid name
r(198);

A solution that does not require removing the offending observations is to use the floor() function:

clear
input DrugList Drug1 Drug2 Drug3 Drug4 Drug5
1234 7934 1234 . . .
5678 1234 5678 . . .
9876 9876 3456 . . .
3456.46 9876 . . . .
7934 9876 5678 7934 1234 .
17453 5678 . . . .
end

levelsof DrugList, clean

foreach drug in `r(levels)' {
    local drug = floor(`drug')
    egen d_`drug' = anycount(Drug1-Drug5), values(`drug')
}

If you have negative values in DrugList you can use the abs() function instead:

clear
input DrugList Drug1 Drug2 Drug3 Drug4 Drug5
1234 7934 1234 . . .
5678 1234 5678 . . .
9876 9876 3456 . . .
-3456 9876 . . . .
7934 9876 5678 7934 1234 .
17453 5678 . . . .
end

levelsof DrugList, clean

foreach drug in `r(levels)' {
    local drug = abs(`drug')
    egen d_`drug' = anycount(Drug1-Drug5), values(`drug')
}

Of course, you can also combine the aforementioned functions:

clear
input DrugList Drug1 Drug2 Drug3 Drug4 Drug5
1234 7934 1234 . . .
5678 1234 5678 . . .
9876 9876 3456 . . .
-3456.46 9876 . . . .
7934 9876 5678 7934 1234 .
17453 5678 . . . .
end

levelsof DrugList, clean

foreach drug in `r(levels)' {
    local drug = abs(floor(`drug'))
    egen d_`drug' = anycount(Drug1-Drug5), values(`drug')
}

The results in all cases are those expected:

list, separator(0)

     +------------------------------------------------------+
     | d_3456   d_1234   d_5678   d_7934   d_9876   d_17453 |
     |------------------------------------------------------|
  1. |      0        1        0        1        0         0 |
  2. |      0        1        1        0        0         0 |
  3. |      1        0        0        0        1         0 |
  4. |      0        0        0        0        1         0 |
  5. |      0        1        1        1        1         0 |
  6. |      0        0        1        0        0         0 |
     +------------------------------------------------------+