Syntax error when programming for ML

[assa martinson]

Hi everyone; I use Stata version 16.

As the title suggests, I am facing syntax issues with programming for maximum likelihood estimation.

I aim to run the following model:

ml model lf our_ML_crraax (r: choice probMit0 probFix0 ra mfMit0total mfMit1total mfFix0total mfFix1total = ) (LNw: ), cluster(accountid) missing maximize technique(nr) difficult constraint(1)

The program for such model is:

program our_ML_crraax

args lnf r LNw muRA

tempvar ra proba1 proba2 probb1 probb2 mA1 mA2 mB1 mB2 yA1 yA2 yB1 yB2 wp1 wp2 euF euM euDiff

tempvar w

syntax varlist [if] [in] [fweight pweight] [, noLOg Robust CLuster(varname) svy EForm *]

marksample touse

if "`svy'" !="" {

svymarkout `touse'

}

quietly {

generate double `proba1' = $ML_y2

generate double `proba2' = 1 -`proba1'

generate double `probb1' = $ML_y3

generate double `probb2' = 1 - `probb1'

generate double `ra' = $ML_y4

generate double `mA1' = $ML_y5

generate double `mA2' = $ML_y6

generate double `mB1' = $ML_y7

generate double `mB2' = $ML_y8

generate double `w' = exp(`LNw')

generate double `yA1' = ((`w' + `mA1')^(1-`r'))/(1-`r') if `ra'== 1

generate double `yA2' = ((`w' + `mA2')^(1-`r'))/(1-`r') if `ra'== 1

generate double `yB1' = ((`w' + `mB1')^(1-`r'))/(1-`r') if `ra'== 1

generate double `yB2' = ((`w' + `mB2')^(1-`r'))/(1-`r') if `ra'== 1

replace `yA1' = ln(`w'+`mA1') if `ra' == 1 & `r'==1

replace `yA2' = ln(`w'+`mA2') if `ra' == 1 & `r'==1

replace `yB1' = ln(`w'+`mB1') if `ra' == 1 & `r'==1

replace `yB2' = ln(`w'+`mB2') if `ra' == 1 & `r'==1

generate double `euA' = (`proba1'*`yA1') + (`proba2'*`yA2') if `ra' == 1

generate double `euB' = (`probb1'*`yB1') + (`probb2'*`yB2') if `ra' == 1

generate double `euDiff' = `euF' - `euM'

replace `lnf' = ln(normal(`euDiff')) if `choice' == 2

replace `lnf' = ln(normal(-`euDiff')) if `choice' == 1

replace `lnf' = ln(normal(`euDiff'))/2 + `lnf' = ln(normal(-`euDiff'))/2 if `$ML_y1' & `ra'==1

}

end

However, when I use ml check to understand what is happening, I receive the following message:

ml check

Test 1:  Calling our_ML_crraay to check if it computes log likelihood and
         does not alter coefficient vector...
         FAILED; our_ML_crraay returned error 198.

Here is a trace of its execution:
------------------------------------------------------------------------------
-> our_ML_crraay __000008 __000009
        - `begin'
        = capture noisily version 16: our_ML_crraay __000008 __000009
          ------------------------------------------------------------------------------------------------------------------------ begin our_ML_crraay ---
          - args lnf r LNw muRA
          - tempvar ra proba1 proba2 probb1 probb2 mA1 mA2 mB1 mB2 yA1 yA2 yB1 yB2 wp1 wp2 euF euM euDiff
          - tempvar w
          - syntax varlist [if] [in] [fweight pweight] [, noLOg Robust CLuster(varname) svy EForm *]
          - marksample touse
          - if "`svy'" !="" {
          = if "" !="" {
            svymarkout `touse'
            }
          - quietly {
          - generate double `proba1' = $ML_y2
          = generate double __00000B = probMit0
          - generate double `proba2' = 1 -`proba1'
          = generate double __00000C = 1 -__00000B
          - generate double `probb1' = $ML_y3
          = generate double __00000D = probFix0
          - generate double `probb2' = 1 - `probb1'
          = generate double __00000E = 1 - __00000D
          - generate double `ra' = $ML_y4
          = generate double __00000A = ra
          - generate double `mA1' = $ML_y5
          = generate double __00000F = mfMit0total
          - generate double `mA2' = $ML_y6
          = generate double __00000G = mfMit1total
          - generate double `mB1' = $ML_y7
          = generate double __00000H = mfFix0total
          - generate double `mB2' = $ML_y8
          = generate double __00000I = mfFix1total
          - generate double `w' = exp(`LNw')
          = generate double __00000S = exp()
invalid syntax
            generate double `yA1' = ((`w' + `mA1')^(1-`r'))/(1-`r') if `ra'== 1
            generate double `yA2' = ((`w' + `mA2')^(1-`r'))/(1-`r') if `ra'== 1
            generate double `yB1' = ((`w' + `mB1')^(1-`r'))/(1-`r') if `ra'== 1
            generate double `yB2' = ((`w' + `mB2')^(1-`r'))/(1-`r') if `ra'== 1
            replace `yA1' = ln(`w'+`mA1') if `ra' == 1 & `r'==1
            replace `yA2' = ln(`w'+`mA2') if `ra' == 1 & `r'==1
            replace `yB1' = ln(`w'+`mB1') if `ra' == 1 & `r'==1
            replace `yB2' = ln(`w'+`mB2') if `ra' == 1 & `r'==1
            generate double `euA' = (`proba1'*`yA1') + (`proba2'*`yA2') if `ra' == 1
            generate double `euB' = (`probb1'*`yB1') + (`probb2'*`yB2') if `ra' == 1
            generate double `euDiff' = `euF' - `euM'
            replace `lnf' = ln(normal(`euDiff')) if `choice' == 2
            replace `lnf' = ln(normal(-`euDiff')) if `choice' == 1
            replace `lnf' = ln(normal(`euDiff'))/2 + `lnf' = ln(normal(-`euDiff'))/2 if `$ML_y1' & `ra'==1
            }
          -------------------------------------------------------------------------------------------------------------------------- end our_ML_crraay ---
        - `end'
        = set trace off

Fix our_ML_crraax.
r(198);

All help is immensely appreciated, thanks in advance

[FernandoRios]

Hi Assa
Welcome to Statalist.

I think there are two possible causes for the error:
1: You do not need this part of the code

syntax varlist [if] [in] [fweight pweight] [, noLOg Robust CLuster(varname) svy EForm *]
marksample touse
if "`svy'" !="" {
svymarkout `touse'
}

2. your program calls for 3 parameters : args lnf r LNw muRA
but is only receiving 2

ml model lf our_ML_crraax
first: (r: choice probMit0 probFix0 ra mfMit0total mfMit1total mfFix0total mfFix1total = )
second: (LNw: )
, cluster(accountid) missing maximize technique(nr) difficult constraint(1)

.
HTH