-
Ever since learning how to refer to Stata's labeled values by name (from👇), I've used it a TON and loved it. But today I ran into a sneaky bug in my code, and if you use the "label":valuelabel construction you should understand this and be careful. 1/ @AmandaYAgan/1126149848548376577Permalink On twitter.com
♻️ 11 Retweets
❤️ 22 Favorites
Mood +4 🙂
-
Here's a short snipped that demonstrated the problem. Note that in line 12, if county == "Baltimore":county FAILS to find the relevant observation. (The construction in line 20 works fine.) 2/Permalink On twitter.com
♻️ 3 Retweets
Mood -2 🙁
-
The problem can arise when you have more than one numerical value with the same label, like label define county 0 "Baltimore" 1 "Baltimore" 2 "New York" When Stata sees "Baltimore":county, it evaluates to 0 so misses Baltimore observations with county == 1. 3/Permalink On twitter.com
♻️ 3 Retweets
❤️ 2 Favorites
Mood 0
-
Of course *after* banging my head against the wall and finally figuring out what was going on, I found a Tweet where brilliant @davidmolitor warns about this and suggests -labelbook- as a way of finding value labels that are not "unique at full length." 4/ @davidmolitor/1126275796706918400On twitter.com
♻️ 3 Retweets
❤️ 3 Favorites
Mood +2 🙂
-
@davidmolitor Last tweet in this thread will be to re-up, once again, my suggestion to check out stata_kernel for running Stata code interactively. If you are used to a bunch of selecting lines then Apple-Shift-D (or non-Mac equivalent) this will rock your world. 5/5 @lukestein/1100227720162598913Permalink On twitter.com
♻️ 4 Retweets
❤️ 5 Favorites
Mood 0