error r(3102) after using fracreg (and fracglm)

[Robert Schemmer]

Hi,
I need to run a regression. As i use fractional data I thought about using the fracreg (or similar Richard Williams' fracglm) command. But I alway get an error code from stata.
In his PDF "Analyzing Proportions: Fractional Response and Zero One Inflated Beta Models" (https://www3.nd.edu/~rwilliam/stats3...onseModels.pdf) he shows an example with Wooldridge's (2011) data from 4,075 companies in 1987. Using exactly the same codes as he does, mainly

use https://www3.nd.edu/~rwilliam/statafiles/401kpart, clear
fracglm prate mrate ltotemp age i.sole, het(mrate ltotemp age i.sole) link(p)

I get the same error from stata as with my own data:

. use https://www3.nd.edu/~rwilliam/statafiles/401kpart, clear
. fracglm prate mrate ltotemp age i.sole, het(mrate ltotemp age i.sole) link(p)
_robust_work(): 3102 function found where matrix required
rob__compute(): - function returned error
robust(): - function returned error
moptimize_result_V_robust(): - function returned error
mopt__result_V(): - function returned error
moptimize_result_post(): - function returned error
Mopt_maxmin(): - function returned error
<istmt>: - function returned error
r(3102);



The same happens after using the fracreg command which he also describes. It is the same error code.

Can someone help me solving this? Did I (and therefore he as I used what he told me to use) miss something out?

Thanks a lot
Robert

[William Lisowski]

Welcome to Statalist, Robert.

I'm sorry to say that I cannot reproduce your problem on my installation.

. about

Stata/SE 15.1 for Mac (64-bit Intel)
Revision 20 Dec 2017
...

. net install fracglm, from(https://www3.nd.edu/~rwilliam/stata)
checking fracglm consistency and verifying not already installed...
installing into /Users/lisowskiw/Library/Application Support/Stata/ado/plus/...
installation complete.

.
. which fracglm
/Users/lisowskiw/Library/Application Support/Stata/ado/plus/f/fracglm.ado
*! version 0.3 beta April 16, 2016   Richard Williams, [email protected]

. 
. do "/var/folders/xr/lm5ccr996k7dspxs35yqzyt80000gp/T//SD18557.000000"

. use https://www3.nd.edu/~rwilliam/statafiles/401kpart, clear

. fracglm prate mrate ltotemp age i.sole, het(mrate ltotemp age i.sole) link(p)

Heteroskedastic Fractional Probit Regression    Number of obs     =      4,075
                                                Wald chi2(4)      =     152.30
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -1674.5212               Pseudo R2         =     0.0632

------------------------------------------------------------------------------
             |               Robust
       prate |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
prate        |
       mrate |   1.384624   .2238384     6.19   0.000     .9459091    1.823339
     ltotemp |   -.149509   .0139654   -10.71   0.000    -.1768806   -.1221374
         age |   .0670689   .0100628     6.67   0.000     .0473461    .0867917
      1.sole |  -.1182798   .0932275    -1.27   0.205    -.3010024    .0644428
       _cons |   1.679357   .1058955    15.86   0.000     1.471806    1.886908
-------------+----------------------------------------------------------------
lnsigma      |
       mrate |   .2403517    .053781     4.47   0.000     .1349428    .3457606
     ltotemp |   .0375134   .0144204     2.60   0.009     .0092499     .065777
         age |   .0171713   .0027289     6.29   0.000     .0118228    .0225198
      1.sole |  -.1627654   .0631052    -2.58   0.010    -.2864493   -.0390814
------------------------------------------------------------------------------

. 
end of do-file

What version of Stata are you using?

[Robert Schemmer]

Hey William,
thanks a lot for your answer! I use the same version as you do, namely 15.1 but for Windows, so I guess that should not be a problem as at least 14 is required.
After reading your post I tried it again and failed again and have no idea why...


Thanks again,
Robert

[William Lisowski]

Perhaps something is wrong with your Stata installation. See the following topic for two good ideas to try.

https://www.statalist.org/forums/for...tata-15-r-3499

[Abdul Adam]

Like @William Lisowski, I can replicate the code without a trouble. So, I think you should follow @William Lisowski's advice in #4.

. . about

Stata/MP 15.0 for Windows
...


. . net install fracglm, from(https://www3.nd.edu/~rwilliam/stata)
checking fracglm consistency and verifying not already installed...
installing into c:\ado\plus\...
installation complete.

. . which fracglm
c:\ado\plus\f\fracglm.ado
*! version 0.3 beta April 16, 2016   Richard Williams, [email protected]


. . use https://www3.nd.edu/~rwilliam/statafiles/401kpart, clear

. . fracglm prate mrate ltotemp age i.sole, het(mrate ltotemp age i.sole) link(p)


Heteroskedastic Fractional Probit Regression    Number of obs     =      4,075
                                                Wald chi2(4)      =     152.30
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -1674.5212               Pseudo R2         =     0.0632

------------------------------------------------------------------------------
             |               Robust
       prate |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
prate        |
       mrate |   1.384633   .2238377     6.19   0.000     .9459188    1.823346
     ltotemp |  -.1495092   .0139654   -10.71   0.000     -.176881   -.1221375
         age |   .0670695   .0100629     6.67   0.000     .0473465    .0867925
      1.sole |  -.1182803   .0932273    -1.27   0.205    -.3010024    .0644419
       _cons |    1.67936   .1058958    15.86   0.000     1.471809    1.886912
-------------+----------------------------------------------------------------
lnsigma      |
       mrate |   .2403522    .053781     4.47   0.000     .1349434    .3457609
     ltotemp |   .0375141   .0144203     2.60   0.009     .0092508    .0657774
         age |   .0171714   .0027289     6.29   0.000     .0118229    .0225198
      1.sole |  -.1627647   .0631046    -2.58   0.010    -.2864475   -.0390819
------------------------------------------------------------------------------

[Richard Williams]

An update to Stata 15.1 was just released yesterday, so you might as well update anyway. Given that itis my example, I am glad it works for most people!

[Robert Schemmer]

Sorry for the late answer.. I am very happy that with the advice it is working now. Thank you very much!