simul_confusion/Modules/ado/plus/p/prvalue.hlp

{smcl}
{* 17Feb2007}{...}
{hline}
help for {hi:prvalue}{right:17Feb2007}
{hline}

{title:Predicted values with confidence intervals for regression models}

{p 8 15 2}{cmd:prvalue} [{cmd:if} exp] [{cmd:in} range] [{cmd:,}
{cmd:x(}{it:variables_and_values}{cmd:)}
{cmdab:r:est(}{it:stat}{cmd:)}
{cmd:all}
{cmdab:l:evel(}{it:#}{cmd:)}
{cmdab:s:ave}
{cmdab:d:iff}
{cmdab:lab:el(}{it:string}{cmd:)}
{cmdab:b:rief}
{cmdab:max:cnt(}{it:#}{cmd:)}
{cmdab:noba:se}
{cmdab:nola:bel}
{cmdab:ys:tar}
{cmd:ept}
{cmdab:del:ta}
{cmdab:boot:strap}
{cmdab:rep:s(}{it:#}{cmd:)}
{cmdab:si:ze(}{it:#}{cmd:)}
{cmdab:bias:corrected}|{cmdab:percent:ile}|{cmdab:norm:al}
{cmd:match}
{cmdab:do:ts}
{cmdab:sa:ving(}{it:filename, save_options}{cmd:)}]

{p 4 4 2}
where {it:variables_and_values} is an alternating list of variables
and either numeric values or mean, median, min, max, upper, lower,
previous.

{p 4 4 2}
{it:stat} is either mean, median, min, max, upper, lower,
previous, grmean(group mean), mrmedian, grmin, grmax.


{title:Description}

{p 4 4 2}
After estimating a regression model, {cmd:prvalue} computes the
predicted values at specific values of the independent variables.
Depending on the model and the options chosen, predicted values can
be estimated values of y, y*, probabilities for each outcome, or expected rate.
By default, the predictions are calculated holding all other variables
at their mean. Values for specific independent variables can be set
using the x() option after {cmd:prvalue}. For example, to compute
predicted values when educ is 10 and age is 30, type {cmd:prvalue, x(educ=10 age=30)}.
Values for the unspecified independent variables can be set using the rest()
option, e.g., {cmd:prvalue, x(educ=10 age=30) rest(mean)}. Changes in
predictions as values of the independent variables change can be
computed using the {cmd:save} and {cmd:diff} options. The {cmd:if}
and {cmd:in} conditions specify conditions for computation of means,
min, etc., that are used with rest(). The command works with {help cloglog},
{help cnreg}, {help fit}, {help gologit}, {help intreg}, {help logistic},
{help logit}, {help mlogit}, {help nbreg}, {help ologit}, {help oprobit},
{help poisson}, {help probit}, {help regress}, {help rologit}, {help slogit}, {help tobit}, {help zinb},
{help ztnb}, and {help ztp}. Standard maximum likelihood based confidence intervals
are computed for cnreg, fit, intreg, regress, and tobit. All other models for which
confidence intervals are available use delta method as default except for {help zinb} and {help zip}.
Confidence intervals for {help zinb} and {help zip} can only use bootstrap method.

{title:Options}

{p 4 8 2}
{cmd:save}  saves current values of indepenent variables and predictions
for computing changes using the diff option.

{p 4 8 2}
{cmd:diff}  computes difference between current predictions and those
that were saved.

{p 4 8 2}
{cmd:label()} adds a label for the prvalue associated with a given 
{cmd:save} or {cmd:diff}. Labels are shown when using {cmd:prvalue, diff}.

{p 4 8 2}
{cmd:level()} sets the {help level} of the confidence interval for predicted
values or probabilities for the commands for which these are provided. The
default is 95.

{p 4 8 2}
{cmd:maxcnt()} is the maximum count value for which the probability
is computed in count models. Default is 9.

{p 4 8 2}
{cmd:x()} sets the values of independent variables for calculating
predicted values.  The list must alternate variable names and values.
The values may be either numeric values or can be mean, median, min,
max, previous, upper, or lower.  The latter cannot be used if rest()
specifies a group summary statistic (e.g., grmean).

{p 4 8 2}
{cmd:rest()} sets the independent variables not specified in x()
to their {cmd:mean} (default), {cmd:minimum}, {cmd:maximum}, {cmd:median}
when calculating predicted values. {cmd:grmean} sets these independent
variables to the mean conditional on the variables and values specified
in x(); {cmd:grmedian},{cmd:grmax}, and {cmd:grmin} can also be used.
If {cmd:prvalue} has already been run after the last estimate,
{cmd:previous} will set unspecified variables to their prior values.
For models other than mlogit, {cmd:upper} and {cmd:lower} can be used
to set independent variables to their minimum or maximum depending
on which will yield the upper or lower extreme predicted value.

{p 4 8 2}
{cmd:all} specifies that any calculations of means, medians, etc.,
should use the entire sample instead of the sample used to estimate
the model.

{p 4 8 2}
{cmd:nolabel} uses values rather than value labels in output.

{p 4 8 2}
{cmd:nobase} suppresses inclusion of the base values of x in the output.

{p 4 8 2}
{cmd:brief} prints only limited output.

{p 4 8 2}
{cmd:ystar} prints the predicted values and maximum likelihood based
confidence intervals of ystar for binary, ordinal, ols regression,
or tobit models.

{p 4 8 2}
{cmd:ept} computes confidence intervals for predicted probabilities
for cloglog, logit, and probit by endpoint transformation.
This method cannot be used for changes in predictions.

{p 4 8 2}
{cmd:delta} calculates confidence intervals by the delta method
using analytical derivatives. This method works with cloglog, logistic,
logit, probit, ologit, oprobit, gologit, poisson, and nbreg.

{p 4 8 2}
{cmd:bootstrap} computes confidence intervals using the bootstrap
method. This method takes roughly 1,000 times longer to compute than
other methods. This method works with cloglog, logistic, logit, mlogit,
probit, ologit, oprobit, gologit, poisson, nbreg, zip, and zinb.

{p 4 8 2}
{cmd:dots} is used with bootstrap to write a . at the beginning of
each replication and periodically prints the percent of total
replications that have been completed. If computations appears to
be stuck (i.e., new dots do not appear), it is likely that the
estimation is not converging for the current bootstrap sample.
This is to be most common with zip, zinb and gologit.
When this happens, you can click on the break symbol to stop
computations for the current sample or wait until the maximum
number of iterations have been computed (by default, the maximum
number of iterations is 16,000). When a model does not converge
for a given bootstrap sample, that sample is dropped.

{p 4 8 2}
{cmd:match} specifies that the bootstrap will resample within
each category of the dependent variable in proportion to the
distribution of the outcome categories in the estimation sample.
If match is not specified, the proportions in each category of
the bootstrap sample are determined entirely by the random draw
and it is possible to end up with samples in which no cases are
found in some of the categories. This option does not apply to
regression or count models (cnreg, intreg, nbreg, poisson, regress,
tobit, zinb, and zip). Usually, bootstrapped confidence intervals
using match option tend to be smaller than those without.

{p 4 8 2}
{cmd:percentile} computes the bootstrapped confidence interval using
the percentile method. This is the default method for bootstrap.

{p 4 8 2}
{cmd:biascorrected} computes the bootstrapped confidence interval
using the bias-corrected method.

{p 4 8 2}
{cmd:normal} computes the bootstrapped confidence interval using
the normal approximation method.

{p 4 8 2}
{cmd:saving()} creates a Stata data file (.dta file) containing
the bootstrap distribution for predictions (predicted probabilities
and expected rates) and discrete changes in discrete choice models
that {cmd:prvalue} applies to.


{title:Examples}

{p 4 4 2}
To compute the predicted values and confidence intervals using delta method
for an ordered logit in which all independent variables are held at the mean.

{p 4 8 2}{cmd:.oprobit warm yr89 male white age ed prst}

{p 4 8 2}{cmd:.prvalue, delta}

{p 4 4 2}To compute predicted values and confidence intervals using
bootstrap method where all independent variables are held at their
minimum

{p 4 8 2}{cmd:.prvalue, rest(minimum) boot}

{p 4 4 2}
To compute values for white females, holding all other variables
at their median with default delta method for confidence intervals.

{p 4 8 2}{cmd:.prvalue, x(white=1 male=0) rest(median)}

{p 4 4 2}
To compute values for white females, holding all other variables at
the median for white females with default delta method for confidence
intervals.:

{p 4 8 2}{cmd:.prvalue, x(white=1 male=0) rest(grmedian)}

{p 4 4 2}
To compute values at the minimum of education, holding all other
variables to the mean with default delta method for confidence intervals.:

{p 4 8 2}{cmd:.prvalue, x(ed=min)}

{p 4 4 2}
To compare the predicted values and compute confidence intervals of discrete
changes for males and females using delta method:

{p 4 8 2}{cmd:.prvalue, x(male=0) save delta}

{p 4 8 2}
 :::

{p 4 8 2}{cmd:.prvalue, x(male=1) dif delta}

{hline}

{p 2 4 2}Authors: J. Scott Long & Jun Xu{p_end}
{p 11 4 2}{browse www.indiana.edu/~jslsoc/spost.htm}{p_end}
{p 11 4 2}spostsup@indiana.edu{p_end}