Hello Statalist, I have a particularly unique, perhaps esoteric question.
I am attempting to create some rules to recode some variables based on a string where there was a maximum of 3 responses allowed, and I have -split- the var which generated 3 new variables (var1, var2, var3). So ideally, if someone responded any combination of numbers as well as the number 5, I would want to recode the "5" to a missing.
So if, for example, someone responded "2, 4, 5", I would replace the var3 response of "5" to a "." but this gets quite burdensome with how many combinations there are. If there had been only two responses allowed, it would be quite easy to do "replace var2 = . if var == "1,5" and then repeat for "2,5" and "3,5" and "4,5".
But I've become curious if there is a way to just indicate "any#, 5" ... perhaps "?,5" or something like that (although I know that specifically will not work).
Thank you in advance for any insight!
Related Posts with Recoding rules based on string variable
Shading between two lines in a graphDear Statalist, I'm trying to create a shaded area between the two borders of the confidence interv…
Extracting the first number from a numeric variable into a new variableHi, I would like to extract the first number of another variable into a new variable. (And an addit…
Bai and Perron (1998) testDear all, I have to test if there are any unknown structural breaks in my panel data. I want to use…
'' xthenreg'' for dynamic threshold modelsDear all it has been 8 months now since we got the new stata command for a dynamic threshold model …
Quantile regressions-panel data, Pre-reqsHi Joao Santos Silva I am worried about two things in quantile regressions for panel data. 1. My d…
Subscribe to:
Post Comments (Atom)
0 Response to Recoding rules based on string variable
Post a Comment