-codebook, problems- with problems indicating incompletely labeled vars

[Dirk Enzmann]

-codebook, problems- can indicate if variables are incompletely labeled. However, it also reports variables as incompletely labeled if the labels correspond exactly to the values of a variable (which I suspect is unintended -- at least to me, because it is not documented), see for example:

* Example generated by -dataex-. For more info, type help dataex
clear
input byte(smident1 smident2)
. .
. .
. .
5 5
2 2
4 4
4 5
4 4
1 1
1 2
1 1
2 3
1 3
2 3
3 4
4 3
1 2
3 3
4 3
2 3
3 4
2 3
2 3
4 5
2 3
3 5
3 2
1 2
1 1
4 5
3 2
5 5
1 1
3 3
3 3
3 3
4 4
3 3
2 3
2 2
3 5
3 3
1 4
1 2
1 2
2 1
2 2
5 5
. .
3 4
1 1
2 3
3 4
3 2
2 2
. .
2 3
. .
3 4
4 4
4 4
4 4
4 3
4 3
2 2
4 2
3 4
3 3
3 4
1 2
3 4
2 2
2 2
4 5
3 4
3 3
3 3
5 4
5 5
3 4
2 3
4 4
1 2
1 1
1 3
3 5
3 3
4 4
3 3
5 5
3 4
1 2
4 3
1 1
3 3
3 2
4 4
2 1
1 5
5 5
end
label values smident1 smident1
label def smident1 1 "not at all", modify
label def smident1 5 "completely", modify
label values smident2 smident2
label def smident2 1 "not at all", modify
label def smident2 2 "2", modify
label def smident2 3 "3", modify
label def smident2 4 "4", modify
label def smident2 5 "completely", modify

recode smident1 smident2 (. = .a)

cap which fre
if _rc ssc install fre

codebook, problems
fre smident1 smident2
label list smident1 smident2

The result of -codebook, problems- indicates that both variables have incompletely labeled variables, although (to my mind) smident2 is completely labeled:

. codebook, problems

Potential problems in dataset [unnamed]

               Potential problem   Variables
----------------------------------------------------
       incompletely labeled vars   smident1 smident2
----------------------------------------------------

. fre smident1 smident2

smident1
-------------------------------------------------------------------
                      |      Freq.    Percent      Valid       Cum.
----------------------+--------------------------------------------
Valid   1  not at all |         19      19.00      20.21      20.21
        2             |         19      19.00      20.21      40.43
        3             |         30      30.00      31.91      72.34
        4             |         19      19.00      20.21      92.55
        5  completely |          7       7.00       7.45     100.00
        Total         |         94      94.00     100.00           
Missing .a            |          6       6.00                      
Total                 |        100     100.00                      
-------------------------------------------------------------------

smident2
-------------------------------------------------------------------
                      |      Freq.    Percent      Valid       Cum.
----------------------+--------------------------------------------
Valid   1  not at all |          9       9.00       9.57       9.57
        2  2          |         20      20.00      21.28      30.85
        3  3          |         29      29.00      30.85      61.70
        4  4          |         22      22.00      23.40      85.11
        5  completely |         14      14.00      14.89     100.00
        Total         |         94      94.00     100.00           
Missing .a            |          6       6.00                      
Total                 |        100     100.00                      
-------------------------------------------------------------------

. label list smident1 smident2
smident1:
           1 not at all
           5 completely
smident2:
           1 not at all
           2 2
           3 3
           4 4
           5 completely

This is not due to the fact, that missing values are not labeled (-codebook, problems- does not care about the labeling of extended missing values): If I modify the labels of smident2, this variable is now treated as completely labeled although the extended missing values are not:

. lab def smident2 2 "2nd" 3 "3rd" 4 "4th", modify

. codebook, problems

Potential problems in dataset [unnamed]

               Potential problem   Variables
--------------------------------------------------
       incompletely labeled vars   smident1
--------------------------------------------------

. fre smident1 smident2

smident1
-------------------------------------------------------------------
                      |      Freq.    Percent      Valid       Cum.
----------------------+--------------------------------------------
Valid   1  not at all |         19      19.00      20.21      20.21
        2             |         19      19.00      20.21      40.43
        3             |         30      30.00      31.91      72.34
        4             |         19      19.00      20.21      92.55
        5  completely |          7       7.00       7.45     100.00
        Total         |         94      94.00     100.00           
Missing .a            |          6       6.00                      
Total                 |        100     100.00                      
-------------------------------------------------------------------

smident2
-------------------------------------------------------------------
                      |      Freq.    Percent      Valid       Cum.
----------------------+--------------------------------------------
Valid   1  not at all |          9       9.00       9.57       9.57
        2  2nd        |         20      20.00      21.28      30.85
        3  3rd        |         29      29.00      30.85      61.70
        4  4th        |         22      22.00      23.40      85.11
        5  completely |         14      14.00      14.89     100.00
        Total         |         94      94.00     100.00           
Missing .a            |          6       6.00                      
Total                 |        100     100.00                      
-------------------------------------------------------------------

[daniel klein]

Technically, what happens is this:

codebook.ado uses the extended macro function label lblname # to obtain the label attached to # in value label lblname. This extended macro function retuns # if no label is attached to # (unless the option strict is specified). Thus, there is no way of distinguishing between

. label list foo
foo:

and

. label list foo
foo:
          42 42

Whether the latter should be considered labeled depends on how we define a label. If the definition is a purely technical one, e.g., a(ny) string, then the codebook should not report missing labels. Note that labelbook with the problems option will report numeric-to-numeric mappings as a potential problem. Personally, I do not like numeric labels. If the label matches the underlying value, there is no need for a label because all commands will simply output the numeric value. If the label does not match the underlying value confusion is a matter of time.

Anyway, I believe this should be clearly documented.

[daniel klein]

duplicate