heckprobit—Probit model with sample lection
Syntax Menu Description Options
Remarks and examples Stored results Methods and formulas References四个意识是指
Also e
Syntax
heckprobit depvar indepvars
if
in
weight
,
lect(微信加好友设置
depvar s=
varlist s
,noconstant offt(varname o)
)
options
options Description
Model
∗lect()specify lection equation:dependent and independent
variables;whether to have constant term and offt variable noconstant suppress constant term
茶具套装offt(varname)include varname in model with coefficient constrained to1 constraints(constraints)apply specified linear constraints
collinear keep collinear variables
SE/Robust
vce(vcetype)vcetype may be oim,robust,cluster clustvar,opg,bootstrap,
or jackknife
Reporting
level(#)t confidence level;default is level(95)
first reportfirst-step probit estimates
noskip perform likelihood-ratio test
nocnsreport do not display constraints
display options control column formats,row spacing,line width,display of omitted
variables and ba and empty cells,and factor-variable labeling
Maximization
maximize options control the maximization process;ldom ud
coeflegend display legend instead of statistics
∗lect()is required.
The full specification is lect(
depvar s=
varlist s
,noconstant offt(varname o)
).
indepvars and varlist s may contain factor variables;e[U]11.4.3Factor variables.
depvar,indepvars,depvar s,and varlist s may contain time-ries operators;e[U]11.4.4Time-ries varlists. bootstrap,by,fp,jackknife,rolling,statsby,and svy are allowed;e[U]11.1.10Prefix commands. Weights are not allowed with the bootstrap prefix;e[R]bootstrap.
vce(),first,noskip,and weights are not allowed with the svy prefix;e[SVY]svy.
pweight s,fweight s,and iweight s are allowed;e[U]11.1.6weight.
coeflegend does not appear in the dialog box.
See[U]20Estimation and postestimation commands for more capabilities of estimation commands.
1
2heckprobit—Probit model with sample lection
Menu
Statistics>Sample-lection models>Probit model with lection
Description
heckprobitfits maximum-likelihood probit models with sample lection.
heckprob is a synonym for heckprobit.
Options
£
£
Model lect(
depvar s=
varlist s
,noconstant offt(varname o)
)specifies the variables and options for the lection equation.It is an integral part of specifying a lection model and is required.The lection equation should contain at least one variable that is not in the outcome equation.
If depvar s is specified,it should be coded as0or1,0indicating an obrvation not lected and1 indicating a lected obrvation.If depvar s is not specified,obrvations for which depvar is not missing are assumed lected,and tho for which depvar is missing are assumed not lected.
noconstant suppress the lection constant term(intercept).
offt(varname o)specifies that lection offt varname o be included in the model with the coefficient constrained to be1.
noconstant,offt(varname),constraints(constraints),collinear;e[R]estimation op-tions.
£
£
SE/Robust vce(vcetype)specifies the type of standard error reported,which includes types that are derived from asymptotic theory(oim,opg),that are robust to some kinds of misspecification(robust),that allow for intragroup correlation(cluster clustvar),and that u bootstrap or jackknife methods (bootstrap,jackknife);e[R]vce option.
£
£
Reporting level(#);e[R]estimation options.
first specifies that thefirst-step probit estimates of the lection equation be displayed before estimation.
noskip specifies that a full maximum-likelihood model with only a constant for the regression equation befit.This model is not displayed but is ud as the ba model to compute a likelihood-ratio test for the model test statistic displayed in the estimation header.By default,the overall model test statistic is an asymptotically equivalent Wald test that all the parameters in the regression equation are zero(except the constant).For many models,this option can substantially increa estimation time.
nocnsreport;e[R]estimation options.
display options:noomitted,vsquish,noemptycells,balevels,allbalevels,nofvla-bel,fvwrap(#),fvwrapon(style),cformat(%fmt),pformat(%fmt),sformat(%fmt),and nolstretch;e[R]estimation options.
heckprobit—Probit model with sample lection3
£
£
Maximization maximize options:difficult,technique(algorithm spec),iterate(#),
no
log,trace, gradient,showstep,hessian,showtolerance,tolerance(#),ltolerance(#),
nrtolerance(#),nonrtolerance,and from(init specs);e[R]maximize.The options are ldom ud.
Setting the optimization type to technique(bhhh)rets the default vcetype to vce(opg).
The following option is available with heckprobit but is not shown in the dialog box: coeflegend;e[R]estimation options.
Remarks and The probit model with sample lection(Van de Ven and Van Pragg1981)assumes that there exists an underlying relationship
y∗j=x jβ+u1j latent equation
such that we obrve only the binary outcome
y probit
j
=(y∗j>0)probit equation The dependent variable,however,is not always obrved.Rather,the dependent variable for obrvation j is obrved if
y lect
j
=(z jγ+u2j>0)lection equation where
u1∼N(0,1)
u2∼N(0,1)
corr(u1,u2)=ρ英语书评
Whenρ=0,standard probit techniques applied to thefirst equation yield biad results.heckprobit provides consistent,asymptotically efficient estimates for all the parameters in such models.
For the model to be well identified,the lection equation should have at least one variable that is not in the probit equation.Otherwi,the model is identified only by functional form,and the coefficients have no structural interpretation.
Example1
We u the data from Pindyck and Rubinfeld(1998).In this datat,the variables are whether children attend private school(private),number of years the family has been at the prent residence (years),log of property tax(logptax),log of income(loginc),and whether one voted for an increa in property taxes(vote).
In this example,we alter the meaning of the data.Here we assume that we obrve whether children attend private school only if the family votes for increasing the property taxes.This assumption is not true in the datat,and we make it only to illustrate the u of this command.
We obrve whether children attend private school only if the head of houhold voted for an increa in property taxes.We assume that the vote is affected by the number of years in residence,
the current property taxes paid,and the houhold income.We wish to model whether children are nt to private school on the basis of the number of years spent in the current residence and the current property taxes paid.
4heckprobit—Probit model with sample lection
.u /data/r13/school
.heckprob private years logptax,lect(vote=years loginc logptax)
Fitting probit model:
Iteration0:log likelihood=-17.122381
Iteration1:log likelihood=-16.243974
(output omitted)
Iteration5:log likelihood=-15.883655
Fitting lection model:
Iteration0:log likelihood=-63.036914
设置页码
Iteration1:log likelihood=-58.534843
Iteration2:log likelihood=-58.497292
Iteration3:log likelihood=-58.497288
Comparison:log likelihood=-74.380943
Fitting starting values:
Iteration0:log likelihood=-40.895684
微信网Iteration1:log likelihood=-16.654497
(output omitted)
Iteration6:log likelihood=-15.753765
Fitting full model:
Iteration0:log likelihood=-75.010619(not concave)
Iteration1:log likelihood=-74.287786
Iteration2:log likelihood=-74.250137
Iteration3:log likelihood=-74.245088
Iteration4:log likelihood=-74.244973
Iteration5:log likelihood=-74.244973
Probit model with sample lection Number of obs=95
Censored obs=36
Uncensored obs=59
Wald chi2(2)= 1.04 Log likelihood=-74.24497Prob>chi2=0.5935
Coef.Std.Err.z P>|z|[95%Conf.Interval] private
土家族摆手舞
years-.1142597.1461717-0.780.434-.400751.1722317
logptax.3516098 1.0164850.350.729-1.640665 2.343884
_cons-2.780665 6.905838-0.400.687-16.3158610.75453 vote
years-.0167511.0147735-1.130.257-.0457067.0122045
loginc.9923024.4430009 2.240.025.1240366 1.860568
logptax-1.278783.5717545-2.240.025-2.399401-.1581647
_cons-.545821 4.070418-0.130.893-8.5236947.432052
/athrho-.8663156 1.450028-0.600.550-3.708318 1.975687
rho-.6994973.7405343-.9987984.962269 LR test of indep.eqns.(rho=0):chi2(1)=0.27Prob>chi2=0.6020
The output shows veral iteration logs.Thefirst iteration log corresponds to running the probit model for tho obrvations in the sample where we have obrved the outcome.The cond iteration log
corresponds to running the lection probit model,which models whether we obrve our outcome of interest.Ifρ=0,the sum of the log likelihoods from the two models will equal the log likelihood of the probit model with sample lection;this sum is printed in the iteration log as the comparison log likelihood.The third iteration log shows starting values for the iterations.
heckprobit—Probit model with sample lection5 Thefinal iteration log is forfitting the full probit model with sample lection.A likelihood-ratio test of the log likelihood for this model and the comparison log likelihood is prented at the end of the output.If we had specified the vce(robust)option,this test would be prented as a Wald test instead of as a likelihood-ratio test.
Example2
In example1,we could have obtained robust standard errors by specifying the vce(robust) option.We do this here and also eliminate the iteration logs by using the nolog option: .heckprob private years logptax,l(vote=years loginc logptax)vce(robust)nolog
Probit model with sample lection Number of obs=95
Censored obs=36
Uncensored obs=59
Wald chi2(2)= 2.55 Log pudolikelihood=-74.24497Prob>chi2=0.2798
Robust
尿姓Coef.Std.Err.z P>|z|[95%Conf.Interval] private
years-.1142597.1113977-1.030.305-.3325951.1040758
logptax.3516098.73582650.480.633-1.090584 1.793803
_cons-2.780665 4.786678-0.580.561-12.16238 6.601051 vote
years-.0167511.0173344-0.970.334-.0507259.0172237
loginc.9923024.4228044 2.350.019.1636209 1.820984
logptax-1.278783.5095156-2.510.012-2.277415-.2801508
_cons-.545821 4.543892-0.120.904-9.4516868.360044
/athrho-.8663156 1.630643-0.530.595-4.062318 2.329687
rho-.6994973.8327753-.9994079.981233 Wald test of indep.eqns.(rho=0):chi2(1)=0.28Prob>chi2=0.5952 Regardless of whether we specify the vce(robust)option,the outcome is not significantly different from the outcome obtained byfitting the probit and lection models parately.This result is not surprising becau the lection mechanism estimated was invented for the example rather than borne from any economic theory.
