**[ERM] erm options** -- Extended regression model options

__Syntax__

*erm_cmd* ... [**,** *extensions* *options*]

*erm_cmd* is one of **eregress**, **eprobit**, **eoprobit**, or **eintreg**.

*extensions* Description
-------------------------------------------------------------------------
Model
__endog__**enous(***enspec***)** model for endogenous covariates; may be
repeated
__entr__**eat(***entrspec***)** model for endogenous treatment assignment
__extr__**eat(***extrspec***)** exogenous treatment
__sel__**ect(***selspec***)** probit model for selection
__tobitsel__**ect(***tselspec***)** tobit model for selection
-------------------------------------------------------------------------

*options* Description
-------------------------------------------------------------------------
Model
__nocons__**tant** suppress constant term
__off__**set(***varname*_o**)** include *varname*_o in model with coefficient
constrained to 1
__const__**raints(***numlist***)** apply specified linear constraints
__col__**linear** keep collinear variables

SE/Robust
**vce(***vcetype***)** *vcetype* may be **oim**, __r__**obust**, __cl__**uster** *clustvar*,
**opg**, __boot__**strap**, or __jack__**knife**

Reporting
__l__**evel(***#***)** set confidence level; default is **level(95)**
__nocnsr__**eport** do not display constraints
*display_options* control columns and column formats, row
spacing, line width, display of omitted
variables and base and empty cells, and
factor-variable labeling

Integration
__intp__**oints(***#***)** set the number of integration (quadrature)
points for integration over four or more
dimensions; default is **intpoints(128)**
__triint__**points(***#***)** set the number of integration (quadrature)
points for integration over three
dimensions; default is **triintpoints(10)**

Maximization
*maximize_options* control the maximization process; seldom used

__coefl__**egend** display legend instead of statistics
-------------------------------------------------------------------------

*enspec* is *depvars*_en **=** *varlist*_en [**,** *enopts*]

where *depvars*_en is a list of endogenous covariates. Each variable
in *depvars*_en specifies an endogenous covariate model using the
common *varlist*_en and options.

*entrspec* is *depvar*_tr [**=** *varlist*_tr] [**,** *entropts*]

where *depvar*_tr is a variable indicating treatment assignment.
*varlist*_tr is a list of covariates predicting treatment assignment.

*extrspec* is *tvar* [**,** *extropts*]

where *tvar* is a variable indicating treatment assignment.

*selspec* is *depvar*_s **=** *varlist*_s [**,** __nocons__**tant** __off__**set(***varname*_o**)**]

where *depvar*_s is a variable indicating selection status. *depvar*_s
must be coded as 0, indicating that the observation was not selected,
or 1, indicating that the observation was selected. *varlist*_s is a
list of covariates predicting selection.

*tselspec* is *depvar*_s **=** *varlist*_s [**,** *tselopts*]

where *depvar*_s is a continuous variable. *varlist*_s is a list of
covariates predicting *depvar*_s. The censoring status of *depvar*_s
indicates selection, where a censored *depvar*_s indicates that the
observation was not selected and a noncensored *depvar*_s indicates
that the observation was selected.

*enopts* Description
-------------------------------------------------------------------------
Model
__prob__**it** treat endogenous covariate as binary
__oprob__**it** treat endogenous covariate as ordinal
__povar__**iance** estimate a different variance for each level
of a binary or an ordinal endogenous
covariate
__pocorr__**elation** estimate different correlations for each
level of a binary or an ordinal endogenous
covariate
__nom__**ain** do not add endogenous covariate to main
equation
__nocons__**tant** suppress constant term
-------------------------------------------------------------------------
**povariance** is available only with **eregress** and **eintreg**.

*entropts* Description
-------------------------------------------------------------------------
Model
__povar__**iance** estimate a different variance for each
potential outcome
__pocorr__**elation** estimate different correlations for each
potential outcome
__nom__**ain** do not add treatment indicator to main
equation
__nocutsint__**eract** do not interact treatment with cutpoints
__noint__**eract** do not interact treatment with covariates in
main equation
__nocons__**tant** suppress constant term
__off__**set(***varname_o***)** include *varname_o* in model with coefficient
constrained to 1
-------------------------------------------------------------------------
**povariance** is available only with **eregress** and **eintreg**.
**nocutsinteract** is available only with **eoprobit**.

*extropts* Description
-------------------------------------------------------------------------
Model
__povar__**iance** estimate a different variance for each
potential outcome
__pocorr__**elation** estimate different correlations for each
potential outcome
__nom__**ain** do not add treatment indicator to main
equation
__nocutsint__**eract** do not interact treatment with cutpoints
__noint__**eract** do not interact treatment with covariates in
main equation
-------------------------------------------------------------------------
**povariance** is available only with **eregress** and **eintreg**.
**nocutsinteract** is available only with **eoprobit**.

*tselopts* Description
-------------------------------------------------------------------------
Model
**ll(***varname*|*#***)** left-censoring variable or limit
**ul(***varname*|*#***)** right-censoring variable or limit
**main** add selection indicator to main equation
__nocons__**tant** suppress constant term
__off__**set(***varname_o***)** include *varname_o* in model with
coefficient constrained to 1
-------------------------------------------------------------------------

__Description__

This entry describes the options that are common to the extended
regression commands; see **[ERM] eregress**, **[ERM] eprobit**, **[ERM] eoprobit**,
and **[ERM] eintreg**.

__Options__

+-------+
----+ Model +------------------------------------------------------------

**endogenous(***depvars*_en **=** *varlist*_en [**,** *enopts*]**)** specifies the model for
endogenous covariates. *depvars*_en is a list of one or more
endogenous covariates modeled with *varlist*_en. This option may be
repeated to allow a different model specification for each endogenous
covariate. By default, the endogenous covariates are assumed to be
continuous, and a linear Gaussian model is used. Unless the **nomain**
suboption is specified, the variables specified in *depvars*_en are
automatically included in the main equation. The following *enopts*
are available:

**probit** specifies to use a probit model for the endogenous covariates.
**probit** may not be specified with **oprobit**; however, you may
specify **endogenous(**...**, probit)** and **endogenous(**...**, oprobit)**.

**oprobit** specifies to use an ordered probit model for the endogenous
covariates. **oprobit** may not be specified with **probit**; however,
you may specify **endogenous(**...**, probit)** and **endogenous(**...**,**
**oprobit)**.

**povariance** specifies that different variance parameters be estimated
for each level of the endogenous covariates. In a
treatment-effects framework, we refer to levels of endogenous
covariates as potential outcomes, and **povariance** specifies that
the variance be estimated separately for each potential outcome.
**povariance** may be specified only with **eregress** and **eintreg** and
with a binary or an ordinal endogenous covariate.

**pocorrelation** specifies that different correlation parameters be
estimated for each level of the endogenous covariates. In a
treatment-effects framework, we refer to levels of endogenous
covariates as potential outcomes, and **pocorrelation** specifies
that correlations be estimated separately for each potential
outcome. **pocorrelation** may be specified only with a binary or an
ordinal endogenous covariate.

**nomain** specifies that the endogenous covariate of covariates be
excluded from the main model, thus removing the effect. This
option is for those who intend to manually construct the effect
by adding it to the main model in their own way.

**noconstant** suppresses the constant term (intercept) in the model for
the endogenous covariates.

**entreat()** and **extreat()** specify a model for treatment assignment. You
may specify only one.

**entreat(***depvar*_tr [**=** *varlist*_tr] [**,** *trtopts modopts*]**)** specifies a
model for endogenous treatment assignment with *depvar*_tr = 1
indicating treatment and *depvar*_tr 0 indicating no treatment.
*varlist*_tr are the covariates for the treatment model; they are
optional.

**extreat(***depvar*_tr [**,** *trtopts*]**)** specifies a variable that signals
exogenous treatment. *depvar*_tr = 1 indicates treatment and
*depvar*_tr = 0 indicates no treatment.

*trtopts* are

**povariance** specifies that different variance parameters be
estimated for each potential outcome (for each treatment
level). **povariance** may be specified only with **eregress** and
**eintreg**.

**pocorrelation** specifies that different correlation parameters be
estimated for each potential outcome (for each treatment
level).

**nomain**, **nocutsinteract**, and **nointeract** affect the way the
treatment enters the main equation.

**nomain** specifies that the main effect of treatment be
excluded from the main equation. Thus, a separate
intercept is not estimated for each treatment level. In
the case of **eoprobit**, this means separate cutpoints are
not added.

**nocutsinteract** specifies that instead of the default of
having separate cutpoints for each treatment level, you
get one set of cutpoints that are shifted by a constant
value for each treatment level. This is implemented by
placing a separate constant in the main equation for each
treatment level. **nocutsinteract** is available only with
**eoprobit**.

**nointeract** specifies that the treatment variable not be
interacted with the other covariates in the main
equation.

These options allow you to customize how the treatment enters the
main equation. When **nomain** and **nointeract** are specified
together, they remove the effect entirely, and you will need to
explicitly reintroduce the treatment effect.

*modopts* are

**noconstant** suppresses the constant term (intercept) in the
treatment model.

**offset(***varname*_o**)** specifies that *varname*_o be included in the
treatment model with the coefficient constrained to 1.

**select()** and **tobitselect()** specify a model for endogenous sample
selection. You may specify only one.

**select(***depvar*_s **=** *varlist*_s [**,** *modopts*]**)** specifies a probit model for
sample selection with *varlist*_s as the covariates for the
selection model. When *depvar*_s = 1, the model's dependent
variable is treated as observed (selected); when *depvar*_s = 0, it
is treated as unobserved (not selected).

**tobitselect(***depvar*_s **=** *varlist*_s [**,** **ll(***varname*|*#***)** **ul(***varname*|*#***)** **main**
*modopts*]**)** specifies a tobit model for sample selection with
*depvar*_s as a censored selection variable and *varlist*_s as the
covariates for the selection model.

**ll(***arg***)** specifies that when *depvar*_s __<__ *arg*, the selection
variable is treated as censored and the model's dependent
variable is unobserved (not selected).

**ul(***arg***)** specifies that when *depvar*_s __>__ *arg*, the selection
variable is treated as censored and the model's dependent
variable is unobserved (not selected).

**main** specifies that the censored selection variable be included
as a covariate in the main equation. By default, it is
excluded from the main equation.

Only the uncensored values of the selection variable
contribute to the likelihood through the main equation.
Thus, the selection variable participates as though it were
uncensored.

*modopts* are

**noconstant** suppresses the constant term (intercept) in the
selection model.

**offset(***varname*_o**)** specifies that *varname*_o be included in the
selection model with the coefficient constrained to 1.

**noconstant**, **offset(***varname*_o**)**, **constraints(***numlist***)**, and **collinear**; see
**[R] estimation options**.

+-----------+
----+ 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 use
bootstrap or jackknife methods (**bootstrap**, **jackknife**); see **[R]**
*vce_option*.

+-----------+
----+ Reporting +--------------------------------------------------------

**level(***#***)** and **nocnsreport**; see **[R] estimation options**.

*display_options*: **noci**, __nopv__**alues**, __noomit__**ted**, **vsquish**, __noempty__**cells**,
__base__**levels**, __allbase__**levels**, __nofvlab__**el**, **fvwrap(***#***)**, **fvwrapon(***style***)**,
**cformat(***%fmt***)**, **pformat(%***fmt***)**, **sformat(%***fmt***)**, and **nolstretch**; see **[R]**
**estimation options**.

+-------------+
----+ Integration +------------------------------------------------------

**intpoints(***#***)** and **triintpoints(***#***)** control the number of integration
(quadrature) points used to approximate multivariate normal
probabilities in the likelihood and scores.

**intpoints()** sets the number of integration (quadrature) points for
integration over four or more dimensions. The number of
integration points must be between 3 and 5,000. The default is
**intpoints(128)**.

**triintpoints()** sets the number of integration (quadrature) points for
integration over three dimensions. The number of integration
points must be between 3 and 5,000. The default is
**triintpoints(10)**.

When four dimensions of integration are used in the likelihood, three
will be used in the scores. The algorithm for integration over four
or more dimensions differs from the algorithm for integration over
three dimensions.

+--------------+
----+ Maximization +-----------------------------------------------------

*maximize_options*: __dif__**ficult**, __tech__**nique(***algorithm_spec***)**, __iter__**ate(***#***)**,
[__no__]__lo__**g**, __tr__**ace**, __grad__**ient**, **showstep**, __hess__**ian**, __showtol__**erance**,
__tol__**erance(***#***)**, __ltol__**erance(***#***)**, __nrtol__**erance(***#***)**, __nonrtol__**erance**, and
**from(***init_specs***)**; see **[R] maximize**.

Setting the optimization type to **technique(bhhh)** resets the default
*vcetype* to **vce(opg)**.

The following option is available with *erm_cmd* but is not shown in the
dialog box:

**coeflegend**; see **[R] estimation options**.