[Clyde Schechter]

Yes. You have to save all the labels from the original set in a temporary do-file and then run that after the reshape. Here's the entire code sequence from beginning to end:

ds ce_* cg_*
local stublist `r(varlist)'
local stublist: subinstr local stublist "ce_" "@", all
local stublist: subinstr local stublist "cg_" "@", all
local stublist: list uniq stublist

foreach s of local stublist {
    local t = strtoname(substr(`"`s'"', 2, .))
    local stublist: subinstr local stublist `"`s'"' `"@`t'"'
}

tempfile labeler
label save _all using `labeler'

reshape long `stublist', i(pidp) j(_j) string

gen int date = cond(_j == "ce_", tm(2021m1), tm(2020m9))
format date %tmMonth_CCYY

drop _j

run `labeler'

[daniel klein]

When I use the commands above to reshape the data I miss all the information of the labels of the variables. It only appear the numbers without any other identification (label). Is it possible to keep the label information and reshape the data at the same time? Thank you very mcuh on advance.

That happened in old versions of Stata (before version 12). It usually happened when reshape-ing to wide format. Anyway, I cannot reproduce the problem running the example in #9. I am using Stata 16.1 now, and I get

(output omitted)

. reshape long `stublist', i(pidp) j(_j) string
(note: j = ce_ cg_)

Data                               wide   ->   long
-----------------------------------------------------------------------------
Number of obs.                      100   ->     200
Number of variables                  15   ->      12
j variable (2 values)                     ->   _j
xij variables:
                        ce_semp cg_semp   ->   semp
          ce_parent5plus cg_parent5plus   ->   parent5plus
                    ce_couple cg_couple   ->   couple
                    ce_sex_cv cg_sex_cv   ->   sex_cv
-----------------------------------------------------------------------------

. describe

Contains data
  obs:           200                          
 vars:            12                          
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
              storage   display    value
variable name   type    format     label      variable label
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
pidp            long    %12.0g                
psu             long    %12.0g     psu        
strata          int     %8.0g      strata     
i_hidp          long    %12.0g                
j_hidp          long    %12.0g                
k_hidp          long    %12.0g                
semp            byte    %23.0g     cg_semp    
parent5plus     byte    %8.0g      cg_parent5plus
                                              
couple          byte    %8.0g      cg_couple
                                              
sex_cv          byte    %8.0g      cg_sex_cv
                                              
i_qfhigh        byte    %68.0g     i_qfhigh   
date            int     %tm..                 
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Sorted by: pidp
     Note: Dataset has changed since last saved.

. label list
cg_semp:
          -8 inapplicable
           2 Yes, self-employed only
           4 No
ce_semp:
          -8 inapplicable
           1 Yes, employed only
           2 Yes, self-employed only
           3 Both employed and self-employed
           4 No
cg_parent5plus:
           1 Yes
           2 No
ce_parent5plus:
           1 Yes
           2 No
ce_couple:
           1 Yes
           2 No
cg_couple:
           1 Yes
           2 No
ce_sex_cv:
           1 Male
           2 Female
cg_sex_cv:
           1 Male
           2 Female
i_qfhigh:
          -8 inapplicable
          15 Other school (inc. school leaving exam certificate or matriculation)

with all value labels neatly defined and attached.


I think Clyde's code in this (outdated) FAQ).

[Clyde Schechter]

That's interesting. Before posting #16, I ran the earlier code with O.P.'s example data, and I did reproduce her problem. I'm running version 17 MP4 on Windows.

Specifically, after reshaping, the reshaped variables had labels attached, but the labels were undefined.

[daniel klein]

That's interesting. Before posting #16, I ran the earlier code with O.P.'s example data, and I did reproduce her problem. I'm running version 17 MP4 on Windows.

Specifically, after reshaping, the reshaped variables had labels attached, but the labels were undefined.

Interesting indeed. I only have access to Stata 17 in a remote environment that will not allow me to copy and paste the dataex examples so I cannot quickly reproduce this behavior. However, keeping the label attached and losing the definitions sounds like a bug to me; especially given that Stata 16.1 handles the situation perfectly well.