------------------------------------------------------------------------------- help forcnbreg(Joseph Hilbe) -------------------------------------------------------------------------------

Negative binomial regression - canonical parameterization

cnbregdepvar[varlist] [ifexp] [inrange] [,offset(varname)exposure(varname)cluster(varname)level(#)from(asis)irrrobustnologmaximize_optionssurvey_options]

aweights,fweights,iweights, andpweights are allowed; see help weights.

cnbregprovides access to allmaximizeoptions; see help maximize.

cnbregprovides access to allsurveyoptions; see help svy.

Description

cnbregfits a maximum-likelihood negative binomial regression model, with canonical parameterization, ofdepvaronindepvars, wheredepvaris a non-negative count variable.

cnbregacccepts all of thehelp maximizeoptions, theconstraint()option, and all survey options and capabilities documented in[SVY]; including multi-level surveys; poststratification; and BRR, jackknife, and linearization VCE estimators.This program uses

ml lfmethod.

Options+-------+ ----+ Model +------------------------------------------------------------

offset(varname)specifies avarnamein model with coefficient constrained to 1.

exposure(varname)specifies aln(varname)in model with coefficient constrained to 1.

constraints(constraints)apply specified linear constraints.

+-----------+ ----+ SE/Robust +--------------------------------------------------------

cluster(varname)

robustspecifies that the Huber/White/sandwich estimator of variance is to be used in place of the traditional calculation.robustcombined withcluster()allows observations which are not independent within cluster (although they must be independent between clusters). If you specifypweights,robustis implied.

vce(options)allowed.vce()supportsrobust,opg, andnative.vcedoes not support optionsbootstraporjacknife, However,cnbregdoes support thebootstrapandjacknifecommands, so these modeling capabilities are allowed.

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

level(#)specifies the confidence level, in percent, for confidence intervals of the coefficients; see help level.

nologsuppresses the iteration log.

+-------------+ ----+ max options +------------------------------------------------------

maximize_options: technique(algorithm_spec), [no]log, trace, hessian, gradient, showstep, shownrtolerance, difficult, iterate(#), tolerance(#), ltolerance(#), gtolerance(#), nrtolerance(#), nonrtolerance, from(init_specs); see maximize.

+-------------+ ----+ svy options +------------------------------------------------------

survey_optionsare all available. See help svy

Author and support

Joseph Hilbe,Arizona State University:jhilbe@aol.com

Remarks

cnbregis a user authored program. Support is by author.NOTE: The canonnical parameterization views the negative binomial as a GLM-type probability function; not a Poisson-gamma mixture model. It includes the negative binomial ancillary parameter in the variance function.

cnbregrequires a nonnegative response.

Examples

. cnbreg los hmo white type2 type3, nolog irr

. cnbreg los hmo white type2 type3, nolog exposure(pop) cluster(state)

. bootstrap: cnbreg los hmo white type2 type3, nolog irr

Also seeReference:

Hardin, J. & Hilbe J. (2001), Generalized Linear Models andExtensions, Stata Press.Online:

helpglmnbreg