Re: st:glm with bin family and link probit VS. probit

[Richard Williams <richardwilliams.ndu@gmail.com>]

My guess is that -glm- lacks the specialized code that detects the 
"predicts failure perfectly" variables. So, instead of vars and 
observations getting dropped, f15 and m15 stay in but their estimated 
standard errors are really large. I would trust the results from 
-probit- more than the results from -glm- (why do you need both 
anyway?). You could try -exlogistic- but your sample may be too big 
for that to be feasible.

At 11:13 PM 5/30/2011, Judy You wrote:
> Dear Stata Experts:
>
>
>
> I have a question regards to comparisons of the two models: glm with
> bin family and link probit VS. probit.
>
>
>
> The data is number of people who died from infection disease by age
> group as follows.
>
>
>
> agegp  0          1
>
>
>
> 0          154573            2
>
> 15        97581  0
>
> 25        159888            9
>
> 40        191122            35
>
> 65        29329  20
>
>
>
> I got the different results by using glm with bin family and link
> probit and probit. The main difference is that glm dropped the last
> dummy variable "f65", while keep all the coefficient even with the zeo
> values (eg., f15 and m15). The Probit dropped not only f65, but also
> the zeo values eg., f15 and m15. The different results lead to
> different estimation of marginal effects followed by the two models.
> Any idea and advice how to control the two models using the same
> independent dummy variables?
>
>
>
> Your help will be much appreciated!
>
>
>
> . glm  AB   m0- f65  [fw= ABfreq], f(b) l(probit) iterate(10)
>
> note: f65 omitted because of collinearity
>
>
>
> Iteration 0:   log likelihood =  -47450.19
>
> Iteration 1:   log likelihood = -825.64878
>
> Iteration 2:   log likelihood = -629.32298
>
> Iteration 3:   log likelihood =  -622.2711
>
> Iteration 4:   log likelihood = -621.33495
>
> Iteration 5:   log likelihood = -621.31077
>
> Iteration 6:   log likelihood = -621.30724
>
> Iteration 7:   log likelihood = -621.30711
>
> Iteration 8:   log likelihood = -621.30711
>
> Iteration 9:   log likelihood = -621.30711
>
> Iteration 10:  log likelihood = -621.30711
>
> convergence not achieved
>
>
>
> Generalized linear models                          No. of obs      =   632559
>
> Optimization     : ML                              Residual df     =
>      632549
>
> Scale parameter =      1
>
> Deviance         =  1242.614219                    (1/df) Deviance =
>         .0019645
>
> Pearson          =   534978.001                    (1/df) Pearson  = .8457495
>
>
>
> Variance function: V(u) = u*(1-u)                  [Bernoulli]
>
> Link function    : g(u) = invnorm(u)               [Probit]
>
>
>
> AIC             =   .001996
>
> Log likelihood   = -621.3071096                    BIC             =
>   -8448049
>
>
>
>
>
> OIM
>
> AB       Coef.   Std. Err.      z    P>z     [95% Conf.     Interval]
>
>
>
> m0   -.9250873   .2495697    -3.71   0.000    -1.414235         -.4359398
>
> m15   -2.901722   7.830577    -0.37   0.711    -18.24937       12.44593
>
> m25   -.5044691    .146919    -3.43   0.001     -.792425         -.2165132
>
> m40   -.2180508   .1199777    -1.82   0.069    -.4532027       .0171012
>
> m65    .1468668    .133816     1.10   0.272    -.1154077         .4091414
>
> f0   -.9122453   .2501379    -3.65   0.000    -1.402507           -.4219841
>
> f15   -2.901722   8.073848    -0.36   0.719    -18.72617         12.92273
>
> f25   -.6696904   .1741904    -3.84   0.000    -1.011097         -.3282836
>
> f40   -.3572294   .1297122    -2.75   0.006    -.6114606         -.1029981
>
> f65   (omitted)
>
> _cons   -3.288246   .1063749   -30.91   0.000    -3.496737   -3.079755
>
>
>
> . probit  AB   m0- f65  [fw= ABfreq], iterate(10)
>
>
>
> note: m15 != 0 predicts failure perfectly
>
> m15 dropped and 190 obs not used
>
>
>
> note: f15 != 0 predicts failure perfectly
>
> f15 dropped and 190 obs not used
>
>
>
> note: f65 omitted because of collinearity
>
> Iteration 0:   log likelihood = -660.01746
>
> Iteration 1:   log likelihood =  -629.8237
>
> Iteration 2:   log likelihood =  -621.8218
>
> Iteration 3:   log likelihood = -621.31032
>
> Iteration 4:   log likelihood = -621.30614
>
> Iteration 5:   log likelihood = -621.30613
>
>
>
> Probit regression                                 Number of 
> obs   =      534978
>
> LR chi2(7)      =           77.42
>
> Prob > chi2     =          0.0000
>
> Log likelihood = -621.30613                       Pseudo R2       = 0.0587
>
>
>
>
>
> AB       Coef.   Std. Err.      z    P>z     [95% Conf.     Interval]
>
>
>
> m0   -.9250871   .2495696    -3.71   0.000    -1.414235         -.4359397
>
> m15   (omitted)
>
> m25   -.5044691    .146919    -3.43   0.001     -.792425         -.2165132
>
> m40   -.2180508   .1199777    -1.82   0.069    -.4532027       .0171012
>
> m65    .1468668    .133816     1.10   0.272    -.1154077         .4091414
>
> f0   -.9122452   .2501378    -3.65   0.000    -1.402506           -.4219841
>
> f15   (omitted)
>
> f25   -.6696904   .1741904    -3.84   0.000    -1.011097         -.3282836
>
> f40   -.3572294   .1297122    -2.75   0.006    -.6114606         -.1029981
>
> f65   (omitted)
>
> _cons   -3.288246   .1063749   -30.91   0.000    -3.496737   -3.079755
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

-------------------------------------------
Richard Williams, Notre Dame Dept of Sociology
OFFICE: (574)631-6668, (574)631-6463
HOME:   (574)289-5227
EMAIL:  Richard.A.Williams.5@ND.Edu
WWW:    http://www.nd.edu/~rwilliam

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/