/LOGREG <data> / K.Vehkalahti
performs logistic regression analysis of <data> by using the GENREG operation.
The logistic regression model is a special case of Generalized Linear Model
(GLM) concept, where the probability distribution is assumed to be binomial.
Most often its canonical link function, the logit function, is used.
The dependent variate is activated by Y. Its values can be 1's and 0's
only (for success and failure, respectively). The covariates are activated
by X's. Also the constant term must be specified by X (i.e. there must be a
variable activated by X, in which all values are 1). Another constant variable
is needed for the number of trials, which is always 1 in a dichotomous case.
This variable is activated by N.
The GENREG operation is applied for the estimation of the parameters
and their standard errors. In addition, Wald's Chi-square tests and their
p-values are given. The Odds Ratios (OR's) of the parameters are saved in
an output matrix OR.M with their 95% confidence intervals.
The residuals and the predicted values may be computed by activating
more variables, see GENREG.
IND, CASES and SELECT specifications can not be used.
M = More information on generalized linear models
Help System (web edition)