-------------------------------------------------------------------------------
help for rhetplot
-------------------------------------------------------------------------------

Plot checking for residual heteroscedasticity after model fit

rhetplot , by(byvarlist) [residualtype variance lowess_options plot(plot) ]

rhetplot varname [, at(numlist) | group(#)] [residualtype variance lowess_options plot(plot) ]

rhetplot [, at(numlist) | group(#)] residualtype variance lowess_options plot(plot) ]

Description

rhetplot is for use after regress and similar commands; see help on the command of interest.

With the first syntax, rhetplot graphs standard deviations (optionally variances) of residuals for the distinct groups formed by combinations of byvarlist. If byvarlist is a single numeric variable, it is used for the x axis; otherwise byvarlist is used as the argument to egen, group() label to define a grouped variable. See egen.

With the second syntax, rhetplot graphs standard deviations (optionally variances) of residuals against means of groups of varname. Groups are defined by the argument of either at() or cut(), used as the corresponding option for egen, cut(). See egen.

With the third syntax, rhetplot graphs standard deviations (optionally variances) of residuals against means of groups of fitted values from the model. Groups are defined by the argument of either at() or cut(), used as the corresponding option for egen, cut(). See egen.

The residuals are, by default, those calculated by predict, residuals or (if the previous estimation command was glm) by predict, response.

The graph is produced by lowess. The `smooth' curve shown (unless the number of groups specified is very small) is best regarded as an informal indication of the general pattern of variability of residuals.

The graph is restricted to the estimation sample.

Options

by(byvarlist), required with the first syntax, specifies that the plot should be of variability of residuals among the distinct groups defined by combinations of byvarlist.

at(numlist) or group(#), required with the second and third syntaxes, specifies how groups of either a specified variable (second syntax) or fitted values (third syntax) are to be determined. at() and group() are options of egen, cut(): see help on egen. In brief, at() specifies lower limits of user-defined classes and group() specifies the number of quantile-based groups with (approximately) equal frequency to be used. (With at(), supply one high value beyond the maximum of the data for best results.)

residualtype specifies a type of residual other than the default. The following types are currently supported: anscombe, deviance, likelihood, pearson, residuals, response, rstandard, rstudent, score, working.

variance specifies plotting of variances rather than standard deviations. Occasionally, especially with counted variables, the natural question may be how far variance is constant or a function of mean fitted values.

lowess_options are any of the options allowed with lowess.

plot(plot) provides a way to add other plots to the generated graph; see plot_option.

Examples

. webuse systolic . anova systolic drug disease . rhetplot, by(drug disease) xla(1/12, valuelabel)

. sysuse auto . glm mpg weight, link(power -1) . rhetplot weight, g(7)

. reg turn length . regplot . rhetplot, g(7) . rhetplot length, g(7)

. insheet using http://www.kgs.ku.edu/Mathgeo/Books/Stat/ASCII/OCS.TXT, clear . label data "petroleum reservoirs, Outer Continental Shelf, Texas and Louisiana" . label var mmboe "ultimate production, million barrels oil equivalent" . label var area "area of closure, acres" . regress mmboe area . rhetplot area, g(7) . regress log10mmboe log10area . rhetplot log10area, g(7)

Author

Nicholas J. Cox, University of Durham, U.K. n.j.cox@durham.ac.uk

Also see

On-line: help for predict, modeldiag