.-
help for ^swboot^ (JMGarrett 12/10/99)
.-
Bootstrap stepwise linear or logistic regression models
-------------------------------------------------------
^swboot^ yvar xvars [^if^ exp] [^in^], [^r^eps^(^#^) pe(^#^) pr(^#^) for^ward ^n(^#^)^
^mod^el ^roc gof^]
^swboot^ uses bootstrap samples of size _N (based on number of observations
without missing values) to validate the choice of variables in stepwise
procedures for linear or logistic regression; variables selected are
displayed for each sample drawn; a summary at the end counts the total
number of times each variable is selected; backward stepwise algorithm is
assumed unless "forward" option is specified
Variables required
------------------
yvar -- dependent variable
If yvar is continuous, defaults to linear regression
If yvar is binary (0,1), defaults to logistic regression
xvars -- list of independent variables
Options
-------
^reps(^#^)^ -- number of samples drawn and stepwise models repeated (default=1)
^pe(^#^)^ -- sign. level for a variable to enter the model (default=.05)
^pr(^#^)^ -- sign. level for a variable to remain in the model (default=.10)
^forward^ -- forward (rather than backward) stepwise regression
^n(^#^)^ -- bootstrap sample size; if not specified, defaults to whole data
set; if specified, can't be larger than original (based on
observations with no missing values for the variables listed)
^model^ -- displays the model for each rep (default: selected variable list)
^roc^ -- displays the area under the ROC curve for each rep
^gof^ -- performs the Hosmer-Lemeshow goodness-of-fit test for each rep
Examples
--------
. ^swboot hpt age race sex educ ses smk bmi chl, reps(50)^
Select 50 samples and run stepwise logistic regression to choose sets
of predictors of hypertension; pr defaults to .1 and pe defaults to .05
. ^swboot chol age race sex ses smk, reps(100) forward pr(.05) pe(.01)^
Select 100 samples and run stepwise linear regression to choose sets of
predictors of cholesterol; uses a foward stepwise method