Stata local macros not displaying

[Kathleen Broussard]

Hello,
I am trying to rescale weights within a single NSFG survey wave to have a mean of 1, however Stata is not returning anything when I use the 'display' command. Stata also says invalid syntax when I attempt to multiply the weight by the new scaling factor, but I can't find anything wrong with it. Is there something small here that is wrong, or am I misunderstanding something about how local macros and scalars work?

summarize finalwgt
return list
local rescalefactor = `r(N)'/`r(sum)'
display `rescalefactor' /// This is the line that returns nothing to me
gen weight2 = .
replace weight2 = finalwgt*`rescalefactor' // Stata error says invalid syntax
summarize weight2

Thank you in advance!

[Clyde Schechter]

Well, I loaded in a data set that contains a variable named finalwgt and ran your code. I cannot reproduce your problem. Instead, I find a different problem. Instead of returning nothing, that line throws an error message:

. display `rescalefactor' /// This is the line that returns nothing to me
> gen weight2 = .
.00008835gen not found

This is happening because of the ///. /// declares the material that follows on that line to be comment, but it also tells Stata that the command continues on the next line. Of course, -genweight2 = .- is a separate command and is not a valid continuation of -display `rescalefactor'.

When I change that /// to // it runs just fine. // declares the material that follows on that line to be comment, but also tells Stata that the command is complete.

[Kathleen Broussard]

Ah, I apologize- I didn't run the code with those comments- I only added them into the blog post to indicate where I was having trouble.

[Clyde Schechter]

Well, there's nothing wrong with the code. The other likely explanation is that you are trying to run this code one line a time, or in blocks of some kind. You can't do that with code involving local macros. When the line or group of lines you run ends, any local macros defined in that block of code disappear. So if you try to refer to them later, they are read as empty strings. If you run the code in one fell swoop from beginning to end, you will overcome that.

[Kathleen Broussard]

Yep, that was the important detail I was missing! I had always heard that locals exist until I exit Stata (exist "within" the program or do-file)- I didn't realize that they only existed within a command or single code block. Thanks so much Clyde!

[Bjarte Aagnes]

I didn't realize that they only existed within a command or single code block

Any selection of lines is not a code block in the Stata ado language. A local macro defined in a code block is available outside the code block:

if (1==1) { // start first block 
    
    local one = 1
} 

if (2==2) { // start second block
    
    local two = 2
} 

macro dir _one _two

_one:           1
_two:           2

[Clyde Schechter]

Bjarte Aagnes is right, I should have chosen my words more carefully. What I meant to say is that if you select a group of lines in the do-file editor and run them, any local macros defined therein go out of existence immediately aftereward. But a block has a different meaning from selected group of lines.