Type Mismatch

[Nikola Angelov]

Hello,

I need some help:

I am using the data from this https://dataverse.harvard.edu/datase...910/DVN/II2DB6

The following states have banned abortion:

"AL"
"AR"
"ID"
"KY"
"LA"
"MS"
"MO"
"OK"
"SD"
"TN"
"TX"
"WV"
"WI"

I want to create a variable ''BanYes'' = 1 for the above states.

I tried using this code:

gen BanYes = 0
replace BanYes = 1 if st == "AL"
replace BanYes = 1 if st == "AR"
replace BanYes = 1 if st == "ID"
replace BanYes = 1 if st == "KY"
replace BanYes = 1 if st == "LA"
replace BanYes = 1 if st == "MS"
replace BanYes = 1 if st == "MO"
replace BanYes = 1 if st == "OK"
replace BanYes = 1 if st == "SD"
replace BanYes = 1 if st == "TN"
replace BanYes = 1 if st == "TX"
replace BanYes = 1 if st == "WV"
replace BanYes = 1 if st == "WI"

However, stata says ''type mismatch''.

How do I solve this?

[Andrew Musau]

There is a difference between a string variable (in red font while browsing the data) and a numerical variable with value labels (in blue font). The error can occur if the variable "st" is a numerical variable with value labels and you are treating it as a string.

browse st

To see the corresponding numerical values, run

lab list

Also see

help inlist

for a more efficient method to the series of replace commands that you have.

[Nikola Angelov]

Thanks for the help. With lab list I can now see each value given to each state.

Now it works

By the way, I am not using inlist, because stata says the expression is too long.

[Andrew Musau]

See this thread for solutions using -inlist()- and alternatives: https://www.statalist.org/forums/for...ssion-too-long.

[Hemanshu Kumar]

By the way, I am not using inlist, because stata says the expression is too long.

Yes, with strings, inlist can take at most 10 arguments, which falls short for you. However, it is still better perhaps to chain two inlists together, like so:

gen byte BanYes = inlist(st, "AL", "AR", "ID", "KY", "LA", "MS", "MO", "OK", "SD") | inlist(st, "TN", "TX", "WV", "WI")

[Nick Cox]

As the variable in question st s evidently numeric, which is why the thread arises at all IIUC, the limit on the number of string arguments for inlist() will not bite.

That said, if the code ends up looking like

gen BanYes = inlist(st, 1, 3, 12)

or whatever the numbers are, then the code will not be especially transparent.