If in foreach

[Joshua Tievoor]

Good day all,

I have three files named A B C and four variables named time, timeA, timeB and timeC.

I regress variables X Y and Z for file A with time variable timeA, X Y and Z for file B with time variable timeB and X Y and Z for file C with time variable timeC

Stata is indicating that my if command is not correct, that 'A' is not found.

Does anyone know what I'm doing wrong in this syntax?

foreach period in A B C{
    clear all
    use "C:\`period'.dta"
    replace time=timeA if `period'=="A"
    replace time=timeB if `period'=="B"
    replace time=timeC if `period'=="C"
    foreach var in X Y Z{
        reg diff `var' time growth size
    }
}

[Gao Liu]

In you current syntax, A is treated as a variable in
replace time=timeA if `period'=="A" So certainly, variable A is not found. You can change the command to
if "`period'"=="A" { replace time=timeA } Do the same for the next three commands. Hope it helps.

[Gao Liu]

My answer above can only solve your issue with "if". You next looping still seems problematic.

[Nick Cox]

The main issue I guess is covered by #6 in http://www.statalist.org/forums/foru...n-a-new-folder See its references.

Backslashes followed by local macro references won't work as you want.

You can avoid the backslash any way. Isn't this equivalent?

cd C:
foreach period in A B C {
    use `period'.dta
    foreach var in X Y Z {
        reg diff `var' time`period' growth size
    }
}

[Joshua Tievoor]

Thanks both!

@Nick, indeed I will correct the backward slash with a forward slash. I mentioned it incorrectly in my example.

Gao Liu, the quotes around `period' indeed did the trick.

[Nick Cox]

Why do you (think you) need to use slashes or if at all? See my code.

[Joshua Tievoor]

Because variable time should be different for each file A B or C that is loaded. At this moment your code excluded this condition. Therefore I needed the correct if-commands, which was resolved by adding quotation marks to the `period' after opening the if-statement.

[Nick Cox]

I don't see that. My code automatically uses whichever of timeA timeB timeC fits the chosen period A B C.

Note the use of the local macro period on the line

 
 reg diff `var' time`period' growth size

[Joshua Tievoor]

oh : D , I follow. Yes, that is even better. Thanks! Gao Liu's solution is also handy to know, but this works as well and is shorter. Thanks again, both.