Trouble reshaping panel data to long form when some variables are only measured in one wave/year

[Zach Goldberg]

Greetings,

I'm running Stata 15.1 on OSX and working with panel data (2011, 2012, 2016, 2017, 2018). All of my variables are labeled according to the wave/year in which they were measured. To reshape the data from wide to long form, I entered the following syntax:

 reshape long weight_ immi_contribution_ immi_naturalize_ immi_makedifficult_ immi_num_ immi_legal_ third_immi
> _ europe_ mexico_ middleast_ india_ ideo5_ educ_ birthyr_ gender_ pew_religimp_ race_ weight_ pew_churatd_ re
> vamshamed_ prouddem_ proudhis_ proudgrp_ race_deservemore_ race_overcome_ race_tryharder_ race_slave_ amborni
> n_ amcit_ amlived_ amenglish_ amchrstn_ amgovt_ amwhite_ amdiverse_ race_importance_ race_majority_1_ race_ma
> jority_2_ race_majority_3_ race_majority_4_ SOCIAL_CONFORMITY_1_ SOCIAL_CONFORMITY_2_ SOCIAL_CONFORMITY_3_ SO
> CIAL_CONFORMITY_4_ immi_contribution_treat_ immi_contribution_a_ immi_contribution_b_  immigration_cap_ conta
> ct_english_treat_ immigrants_take_ immigrants_culture_ immigrants_security_ immigrants_crime_ immigrants_econ
> omy_ immigration_help_ volunteerorg2_10_ antieur_ egalitarian_opportunies_ egalitarian_worry_less_ egalitaria
> n_no_big_deal_ egalitarian_fewer_problems_ faminc_ state_ party7_ case_identifier europe_ mexico_ middleast_ 
> india_ china_ africa_, i(case_identifier) j(year)
(note: j = 2016 2017 2018)
(note: immi_num_2016 not found)
(note: immi_legal_2016 not found)
(note: third_immi_2016 not found)
(note: europe_2016 not found)
(note: mexico_2016 not found)
(note: middleast_2016 not found)
(note: india_2016 not found)
(note: birthyr_2016 not found)
(note: gender_2016 not found)
(note: immi_contribution_treat_2016 not found)
(note: immi_contribution_a_2016 not found)
(note: immi_contribution_b_2016 not found)
(note: immigration_cap_2016 not found)
(note: contact_english_treat_2016 not found)
(note: immigrants_take_2016 not found)
(note: immigrants_culture_2016 not found)
(note: immigrants_security_2016 not found)
(note: immigrants_crime_2016 not found)
(note: immigrants_economy_2016 not found)
(note: immigration_help_2016 not found)
(note: volunteerorg2_10_2016 not found)
(note: antieur_2016 not found)
(note: egalitarian_opportunies_2016 not found)
(note: egalitarian_worry_less_2016 not found)
(note: egalitarian_no_big_deal_2016 not found)
(note: egalitarian_fewer_problems_2016 not found)
(note: case_identifier2016 not found)
(note: china_2016 not found)
(note: africa_2016 not found)
variable weight_2016 not found
r(111);

...which didn't work. The problem appears to be that certain variables are not available across all years. How can I correct for this?

Below is some sample data, which includes variables that either weren't included in the 2017 wave (immi_contribution_2018) or the 2016 and 2018 waves (immigration_cap_2017).

* Example generated by -dataex-. To install: ssc install dataex
clear
input long case_identifier byte(immi_contribution_2016 immi_contribution_2018 immigration_cap_2017)
    779 1 1 1
   2108 2 3 2
   2597 1 1 2
   4148 1 . .
   4460 3 2 2
   5225 1 1 2
   5903 1 . 2
   6059 3 3 3
   8048 1 . 1
   9869 1 1 2
  13112 1 . .
  14087 1 1 2
  14474 3 . 3
  14507 1 1 2
  15464 1 1 2
  20459 1 1 2
  25193 2 1 1
  27599 3 8 2
2035940 1 1 1
6000479 3 . .
6000794 3 . 3
6001082 2 1 2
6002069 3 3 3
6004538 1 . 2
6006515 3 . 3
6008999 3 . 3
6009677 1 . .
6010778 1 1 2
6013199 3 3 3
6014066 2 . .
6014345 1 . .
6015176 3 3 3
6018461 3 . .
6020087 3 . .
6021503 1 . 2
6022898 3 3 2
6023987 3 . 2
6024683 8 3 3
6026186 1 1 2
6026195 1 1 1
6026246 1 1 2
6026537 3 . .
6028715 3 3 3
6032768 3 . .
6033338 3 3 3
6035645 8 . 2
6036608 1 . .
6038444 1 . .
6041060 3 . .
6046208 3 3 3
6048548 1 . .
6051575 1 . .
6054065 1 1 1
6057581 3 . .
6059078 3 . .
6061217 3 3 3
6061877 1 1 1
6061937 2 1 2
6062156 1 . .
6063833 1 1 2
6064967 1 1 1
6067322 2 1 2
6067754 3 3 1
6070919 3 . 3
6072560 1 1 1
6073346 1 1 2
6074072 3 2 2
6079028 1 . 2
6079631 1 . .
6081782 1 . 2
6082415 1 1 1
6082955 1 . .
6084239 1 . 2
6085712 3 3 3
6086672 3 . 3
6086936 1 . .
6088319 1 1 1
6088628 8 . 2
6089561 2 2 2
6095012 3 3 2
6095318 1 1 2
6095753 2 . 2
6095852 1 1 2
6098411 2 . .
6100238 1 1 2
6100907 3 3 3
6106253 3 . 3
6108272 1 1 1
6109619 1 1 2
6110054 1 2 2
6110435 1 1 1
6110906 3 3 1
6111152 3 . .
6112508 1 1 2
6114890 3 . .
6116483 3 . 3
6116600 1 . 2
6116972 1 . 2
6117203 1 . 1
6117347 1 1 2
end
label values immi_contribution_2016 V323_A
label def V323_A 1 "Mostly make a contribution", modify
label def V323_A 2 "Neither", modify
label def V323_A 3 "Mostly a drain", modify
label def V323_A 8 "Don't know", modify
label values immi_contribution_2018 V993_A
label def V993_A 1 "Mostly make a contribution", modify
label def V993_A 2 "Neither", modify
label def V993_A 3 "Mostly a drain", modify
label def V993_A 8 "Don't know", modify
label values immigration_cap_2017 IMMIGRAT
label def IMMIGRAT 1 "Increase", modify
label def IMMIGRAT 2 "Keep about the same", modify
label def IMMIGRAT 3 "Decrease", modify

Thanks in advance for your help!

[Jorrit Gosens]

The variable stub weight_ is included twice.
Stata reshapes it, and then is asked to reshape it again, and it complains.


Edit:
Also, when Stata says: (note: immi_num_2016 not found), this is just a note, not an error message. Reshape can deal with missing values of j, here years, just fine.