## Stata 15 help for eteffects_postestimation

```
[TE] eteffects postestimation -- Postestimation tools for eteffects

Postestimation commands

The following postestimation command is of special interest after
eteffects:

Command            Description
-------------------------------------------------------------------------
estat endogenous   perform tests of endogeneity
-------------------------------------------------------------------------

The following standard postestimation commands are available after
eteffects:

Command            Description
-------------------------------------------------------------------------
estat summarize    summary statistics for the estimation sample
estat vce          variance-covariance matrix of the estimators (VCE)
hausman            Hausman's specification test
lincom             point estimates, standard errors, testing, and
inference for linear combinations of coefficients
nlcom              point estimates, standard errors, testing, and
inference for nonlinear combinations of coefficients
predict            predictions, residuals, influence statistics, and
other diagnostic measures
predictnl          point estimates, standard errors, testing, and
inference for generalized predictions
test               Wald tests of simple and composite linear hypotheses
testnl             Wald tests of nonlinear hypotheses
-------------------------------------------------------------------------

Syntax for predict

predict [type] {stub* | newvar | newvarlist} [if] [in] [, statistic
tlevel]

predict [type] {stub* | newvarlist} [if] [in] , scores

statistic        Description
-------------------------------------------------------------------------
Main
te             treatment effect; the default
cmean          conditional mean at treatment level
ps             propensity score
xb             linear prediction
psxb           linear prediction for propensity score
xbtotal        linear prediction, using residuals from treatment model
-------------------------------------------------------------------------
Specify one new variable with te; specify one or two new variables with
cmean, ps, and xb.

Statistics > Postestimation

Description for predict

predict creates a new variable containing predictions such as treatment
effects, conditional means, propensity scores, and linear predictions.

Options for predict

+------+
----+ Main +-------------------------------------------------------------

te, the default, calculates the treatment effect.

cmean calculates the conditional mean for the control group.  To also
obtain the conditional mean for the treatment group, specify two
variables. If you want the conditional mean for only the treatment
group, specify the tlevel option.

ps calculates the probability of being in the control group.  To also
obtain the probability of being in the treatment group, specify two
variables.  If you want the probability of being in the treatment
group only, specify the tlevel option.

xb calculates the linear prediction for the control group.  To also
obtain the linear prediction for the treatment group, specify two
variables.  If you want the linear prediction for only the treatment
group, specify the tlevel option.

psxb calculates the linear prediction for the propensity score.

xbtotal calculates the linear prediction for the control group, including
the residuals from the treatment model as regressors.  To also obtain
the linear prediction for the treatment group, specify two variables.
If you want the linear prediction, including the residuals from the
treatment model as regressors, only for the treatment group, specify
the tlevel option.

tlevel specifies that the statistic be calculated for the treatment
group; the default is to calculate the statistic for the control
group.

scores calculates the score variables. For eteffects, this is the same as
the residuals in the moment conditions used by the generalized method
of moments (see [G] gmm). For the average treatment effect, the
average treatment effect on the treated, and the potential-outcome
means, parameter-level scores are computed.  For the auxiliary
equations, equation-level scores are computed.

Syntax for estat

estat endogenous

Statistics > Postestimation

Description for estat

estat endogenous performs a Wald test to determine whether the estimated
correlations between the treatment-assignment and potential-outcome
models are different from zero.  The null hypothesis is that the
correlations are jointly zero.  Rejection of the null hypothesis suggests
endogeneity.

Examples

Setup
. webuse cattaneo2
. eteffects (bweight i.prenatal1 i.mmarried mage i.fbaby) (mbsmoke
i.mmarried mage i.fbaby medu fedu)

Test for endogeneity
. estat endogenous

Compute the estimated treatment probabilities
. predict prob1 prob2, ps

Summarize the estimated treatment probabilities
. summarize prob1 if mbsmoke==1, detail

Summarize the estimated control probabilities
. summarize prob2 if mbsmoke==0, detail

```