{smcl}
{* 23 aug 2002}{...}
{hline}
help for {hi:gennbreg}
{hline}

{title:Generate data from Stata's {cmd:nbreg} model}

{p 8 14}{cmd:gennbreg} {it:newvar} 
	[{cmd:if} {it:exp}] [{cmd:in} {it:range}] [{cmd:,}
	{cmdab:d:ispersion(}{cmdab:m:ean}|{cmdab:c:onstant}{cmd:)}
	{cmd:mu(}{it:#}|{it:varname}{cmd:)}
	{cmd:xbeta(}{it:#}|{it:varname}{cmd:)}
	{cmd:alpha(}{it:#}{cmd:)} {cmd:delta(}{it:#}{cmd:)}
	{cmdab:ado:only}

{title:Description}

{p}{cmd:gennbreg} generates data {it:newvar} 
appropriate for Stata's {cmd:nbreg}
estimation command.

{title:Options}

{p 0 4}{cmd:dispersion(mean}|{cmd:constant)} specifies the parameterization of
the model.  The default is {cmd:dispersion(mean)}, which yields a model with
dispersion equal to 1 + alpha*exp(x_i*b + offset); i.e., the dispersion is a
function of the expected mean: exp(x_i*b + offset).
{cmd:dispersion(constant)}
has dispersion equal to 1 + delta; i.e., it is a constant for all
observations.

{p 0 4}{cmd:mu(}{it:#}|{it:varname}{cmd:)} specifies the mean of the 
negative binomial distribution, i.e. the exponentiated linear predictor in a 
regression model, {it:varname}, or a constant {it:#}.

{p 0 4}{cmd:xbeta(}{it:#}|{it:varname}{cmd:)} specifies the linear 
predictor in a negative binomial
regression model, {it:varname}, or a constant {it:#} in 
which case negative binomial
random deviates are generated with mean exp({it:#}).

{p 4 4}Only one of {cmd:mu()} or {cmd:xbeta()} should be specified.  If 
neither is specified, then {it:newvar} will contain random deviates from 
a negative binomial distribution with mean one.

{p 0 4}{cmd:alpha(}{it:#}{cmd:)} specifies the dispersion parameter in 
mean dispersion models.  The default is one.

{p 0 4}{cmd:delta(}{it:#}{cmd:)} specifies the dispersion parameter in 
constant dispersion models.  The default is one.

{p 4 4}Only one of {cmd:alpha()} or {cmd:delta()} should be specified.

{p 0 4}{cmd:adoonly} specifies that only ado code be used, whereas by default
a Stata plugin is used.  That is, {cmd:gennbreg} calls {cmd:genpoisson}, and
the file {cmd:genpoisson.ado} contains a subroutine written in the ado
language, and this subroutine has also been implemented as a plugin.  Both
produce equivalent results, but the plugin is faster.  The {cmd: adoonly}
option provides endless amusement to those performing speed comparisons.

{title:Examples}

{p 8 12}{inp:. gen x1 = invnorm(uniform())}{p_end}
{p 8 12}{inp:. gen x2 = invnorm(uniform())}{p_end}
{p 8 12}{inp:. gen xb = -1 + 0.5*x1 + 1.5*x2}{p_end}
{p 8 12}{inp:. gennbreg y1, xbeta(xb) alpha(0.3)}{p_end}
{p 8 12}{inp:. nbreg y1 x1 x2, dispersion(mean)}{p_end}
{p 8 12}{inp:. gen exb = exp(xb)}{p_end}
{p 8 12}{inp:. gennbreg y2, mu(exb) delta(0.3) dispersion(constant)}{p_end}
{p 8 12}{inp:. nbreg y2 x1 x2, dispersion(cons)}{p_end}