Invalid name error while using variable names in a loop

[Ubaidullah Jamal]

Hello, I am trying to replace the missing values for all variables for an observation using the values from a subsequent observation. Below is the code that I am using.

ds
local allvar `r(varlist)'
foreach cc of local allvar {
bys id: replace "`cc'"[_n] = "`cc'"[_n+1] if dup == 1 & missing("`cc'"[_n]) & !missing("`cc'"[_n+1]) & check == 1 & max_dup == 2
}

When I run this code, however, I encounter an invalid name error shown here as: "date invalid name - date here is the first variable in my list of variables. Any and all help is highly appreciated.

[Joseph Coveney]

Omit the double-quotation marks surrounding your variable name's local macro.

quietly ds
foreach var of varlist `r(varlist)' {
    bysort id: replace `var'[_n] = `var'[_n+1] . . .
}

[Ubaidullah Jamal]

Omit the double-quotation marks surrounding your variable name's local macro.

quietly ds
foreach var of varlist `r(varlist)' {
bysort id: replace `var'[_n] = `var'[_n+1] . . .
}

Thanks a ton! This worked.

[Joseph Coveney]

You can refactor the passage a little more, by omitting ds (unless you're selecting a subset of variables) and using the special constant _all instead. (I think an asterisk also works.)

And although it probably isn't so great an inefficiency (because sort I think checks whether the dataset is already sorted), you can put the sort outside the loop and do it just once.

sort id <within-ID sequence variable>
foreach var of varlist _all {
    by id: replace ...