SAS catalog: value label sets with long names aren't being matched

[evanmiller]

SAS catalog files can name their formats (value label sets) with both 8-byte short name and a 32-byte long name. At present ReadStat is returning the short name, which may not be the correct behavior. @cullenjd has reported that his value labels are not being properly applied (see previous discussion at #34). It is likely an issue with ReadStat but could also be something with haven.

[evanmiller]

I've pushed a possible fix to ReadStat here: WizardMac/ReadStat@ac1a2db

[evanmiller]

@cullenjd Please try:

devtools::install_github("evanmiller/haven", ref="update-readstat")

[cullenjd]

@evanmiller - I updated but I'm still not getting the value labels from the catalog file.

table(opdata$workstat)

0 1 2 3
5604 2441 67 42

Should be:

"No" "Yes" "Don't Know" "Refused"
5604 2441 67 42

[evanmiller]

@cullenjd - Do you have a small sample data file you can share?

[cullenjd]

@evanmiller
https://www.dropbox.com/s/hdg3ng810wgyrjv/clear.sas7bdat?dl=0

[evanmiller]

I am seeing the labels:

> df$workstat
<Labelled>
  [1]  2  0  1  0  0  0  0  1  0  0  0  0  0  1  1  0  0  0  1  0  0  0  0  0  0  1  0  0  0  0  1  0  0  0  0  1  0  0  0  0  0  0  0  1  1  0  0  0  1  0  1
 [52]  0  1  0  0  0  0  1  1  1 NA  1  0  1  0  0  0  0  1  1  1  0  0  0  1  0  0  0  0  0 NA  0  0  1  0 NA  0  0  0  0  0  0  1  0  0  0  0  0  0  0
attr(,"label")
[1] workstat: H1. At the time of your fracture, were you working full- or part-time for pay?

Labels:
 value                                  label is_na
     1                                  A.Yes FALSE
     0                                   B.No FALSE
     2                           C.Don't know FALSE
     3 D.Patient declined to answer (Refused) FALSE

It appears the table doesn't produce labels -- I think this is a separate issue (@hadley?).

In any event it appears that labels are imported correctly with the patch. I will leave this issue open until the changes are merged into haven/master.

[gergness]

@cullenjd - the variables are imported as labelled class vectors and require some more manipulation before you can refer to them by their labelled values. I think the idea is that haven imports the data in this intermediate form, and expects the user to convert them to more standard R structures that suit the particular analysis.

In your example, use as_factor() to convert the variable to a factor, and table() will use the labels.

opdata$workstat <- as_factor(opdata$workstat)
table(opdata$workstat)