Problem when running loops with forvalues command: error message: `i' invalid name

[Nico Ochmann]

Dear Statalist,

I got a problem with running a loop.
I give you my whole code, so you can see how variables/matrices are defined.
The problem is: forvalues i=......{ } gives me an error message with `i' invalid name.

I would highly appreciate your help/suggestions.

Best wishes and thank you very much in advance!

Nico

set more off 
preserve 
drop if aage>0 & aage<=16 
drop if STATR==2   

reg  lfp   edu pexp pexp2 pexp3 pexp4 i.year i.yuk  i.cohort uratem married fd  if  (age>=25 & age<65) & female==0 & im==1 & yuk<=60, robust   

matrix beta=e(b) // to display matrix values write `matrix list beta' 
predict yhat if e(sample)
su yhat if e(sample)
scalar ybar=r(mean)
scalar nobs=e(N)
scalar beta0=beta[1,109] //  _constant is at 109.
matrix yuk=beta[1,"0.yuk".."60.yuk"]   
tabulate yuk if e(sample), matcell(yukmeans) // this gives me the absolute frequencies in a matrix or vector
matrix yukmeans=yukmeans/nobs // now generate the relative frequencies 
matrix i1=yuk*yukmeans // this is a 1x1 matrix, I checked. 
scalar i1=trace(i1) //  single value of the matrix and define a scalar 
scalar i2=ybar-i1  
                

generate yhat3=0 in 1/61    

forvalues i=0(1)60 {                                             // here is where the error message comes up!!!!!!!!!!!!!! `i' invalid name
    local j = `i’+1
    local alpha = el(yuk,1,`i'.yuk) 
    replace yhat3= beta0 + `alpha'  in `j'     
}     
         

         
generate x=_n   in 1/45     
    
    
scatter yhat3  x   
restore

[Mike Lacy]

First, I would note that at least part of the relevant code you have here that precedes the problem could not have executed. I note particularly

matrix yuk=beta[1,"0.yuk".."60.yuk"]

which is not legal syntax. (It's unusual enough that I don't know what you were trying to do.)

Regardless of that, there's a problem with

local alpha = el(yuk,1,`i'.yuk)

The -el()- function expects its last two argument to be numbers, but your `i'.yuk is not a number. One way to show this to yourself is to use -set trace on- at the beginning of your forvalues loop.

I don't understand what you wanted that statement to do, so I don't have a positive suggestion.. What, for example, did you want 0.yuk, 1.yuk, ..., 60.yuk to contain?

[Scott Merryman]

You appear to using the acute accent mark rather than the right single quote. Compare:

  local j = `i’+1

local j = `i' +1

[Nick Cox]

What you show in #1 also is not consistent about quotation marks. Local macro foo is evaluated like this

`foo'

Although fonts vary, in very many fonts the closing quotation mark appears to be vertical. That is sometimes but not always true of #1. So

 
 local j = `i’+1

should look like

 
 local j = `i' + 1

[Nico Ochmann]

Dear Mike, Scott and Nick
thanks for your kind help and careful reading of my code. I apologize about overlooking such a thing. Now this should take care of the error message. Two questions remain. I have 61 estimates of 61 yuk indicators where the reference category for years since migration or years in UK is year 0 with “estimate” equal to 0. So i.yuk generates 61 values. They all appear in matrix beta = e(b). Basically what I want are two things. First define a new matrix with these 61 regression estimates. This it is what I am trying to do with matrix yuk=beta[1,........] where #2 says not a Stata syntax. When I do the loop with el() I want to extract each of these 61 estimates individually one at a time from the matrix yuk.
This is it gentlemen!
Any suggestions I am very grateful for.

Best wishes

Nico