Dropping observations of a string variable that are not stored in a local macro

[Nicolas Charette]

Hello,

I am using Stata 17/BE for Mac and I want to drop observations if certain strings are NOT observed in a string variable named cours.

I identified the strings I wish to keep with levelsof (there are a lot of them) and I stored them in a local macro:

levelsof cours if niv>=10 & niv!= . & c==".", local(levs)


Now I want to drop the observations that don't contain the strings stored in local(levs) and under one condition (if c=="."). I know I can’t code cours!=levs, but I want is something like:

drop if c=="." & cours!=levs

I hope what I want to do is clear. Could someone please help me?

Thank you very much,

Nicolas Charette

Here is a reduced sample of my data, if it can help:

* Example generated by -dataex-. For more info, type help dataex
clear
input int niv str25 cours byte note str4 c
  . "anglais1recycleANGL1C"      60 "ang"
  . "musique1ercycleMUSI1C"      80 "."  
  . "edu.phy1ercycleEPSA1C"      76 "."  
  . "math.1ercycleMATH1C"        74 "math"
  . "e.m.r.c.1ercycleEMRC1C"     87 "."  
  . "francais1ercycleFRAN1C"     81 "fran"
  . "a.plast.1ercycleARTP1C"     87 "."  
 40 "mathematique22400"          70 "math"
 40 "anglais2ecycle17400"        42 "ang"
 40 "francais14400"              60 "fran"
 60 "anglais17600"               72 "ang"
 60 "francais14600"              62 "fran"
 60 "mathematique22600"          67 "math"
 70 "anglais134104"              51 "ang"
 70 "mathematique63106"          60 "math"
 70 "francais132108"             36 "fran"
  . "ethiqueetculture69102"      86 "."  
  . "science/technologie55104"   81 "."  
  . "musique169104"              87 "."  
  . "geographie95103"            80 "."  
  . "mathematique63106"          84 "math"
  . "educationphysique43102"     73 "."  
  . "anglais134104"              70 "ang"
  . "histoire/education87103"    79 "."  
  . "francais132108"             72 "fran"
 80 "mathematique63206"          80 "math"
 80 "anglais134204"              69 "ang"
 80 "francais132208"             74 "fran"
 90 "francais132308"             65 "fran"
 90 "anglais134304"              60 "ang"
 90 "mathematique63306"          70 "math"
100 "science/technologie55444"   75 "."  
100 "educationphysique43402"     87 "."  
100 "artsplastiques168402"       73 "."  
100 "mathematique63414"          82 "math"
100 "ethiqueetculture69404"      78 "."  
100 "francais132406"             71 "fran"
100 "football/sports46454"       96 "."  
100 "anglais134404"              55 "ang"
100 "histoire/education87404"    83 "."  
110 "anglais134404"              81 "ang"
110 "anglais(pr.local)134451"    81 "ang"
110 "francais132506"             65 "fran"
110 "francais(pr.local)130551"   65 "fran"
111 "ethiqueetculture69502"      82 "."  
111 "anglais(prog.local)134551"  78 "ang"
111 "informatique111551"         91 "."  
111 "anglais134504"              78 "ang"
111 "mathematique63504"          87 "math"
111 "artculinaire115571"        100 "."  
end

[Mike Lacy]

Presuming you mean "observations in which levs contains the value in cours," I think this does what you want

levelsof cours if niv>=10 & niv!= . & c==".", local(levs)
gen byte trouve =  strpos(`"`levs'"', cours) > 0
drop if (c == ".") & !trouve

[Nicolas Charette]

Thank you Mike! It worked like a charm!

Nicolas