Statscraft 0.7.0, release v0.7.0

LOGISTIC REGRESSION

Binary logistic regression with categorical covariates, stepwise selection and case diagnostics.

Syntax

LOGISTIC REGRESSION [VARIABLES=] dependent [WITH var | var BY var ...]
   [/CATEGORICAL=varlist] [/CONTRAST(var)=INDICATOR[(ref)] | DEVIATION[(ref)] | SIMPLE[(ref)] |
   DIFFERENCE | HELMERT | REPEATED | POLYNOMIAL[(metric)] | SPECIAL(matrix)]
   [/METHOD=ENTER | FSTEP[(COND | LR | WALD)] | BSTEP[(COND | LR | WALD)] [ALL | terms]] ...
   [/ORIGIN | NOORIGIN] [/ID=var] [/PRINT=[DEFAULT] [SUMMARY] [CORR] [ITER[(n)]] [GOODFIT] [CI(95)] [ALL]]
   [/CRITERIA=[BCON(0.001)] [ITERATE(20)] [LCON(0)] [PIN(0.05)] [POUT(0.10)] [EPS(1E-8)] [CUT(0.5)]]
   [/MISSING=EXCLUDE | INCLUDE] [/CASEWISE=[tempvars] [OUTLIER(2)]]
   [/SAVE=tempvar[(name)] ...] [/CLASSPLOT] [/EXTERNAL].
tempvar: PRED PGROUP RESID DEV LRESID SRESID ZRESID LEVER COOK DFBETA. The dependent variable has
exactly two values among the cases used: the lower is coded 0 and the higher 1 (strings in
alphabetical order). A case missing on any variable is left out; one missing only on the dependent
still gets PRED and PGROUP. String covariates are categorical; a categorical variable with k
values becomes k - 1 coded variables that enter and leave together (INDICATOR, last category as
reference, by default). The fit is Newton-Raphson from all coefficients 0 and stops when no
coefficient changes by BCON or more, the -2 log likelihood changes by less than LCON percent, or
after ITERATE iterations. Each METHOD starts from the model the one before it left: ENTER enters
its terms, FSTEP enters by the score test (PIN) and removes by the chosen statistic (POUT), BSTEP
enters its terms first and then removes and enters the same way. Tables: Case Processing Summary,
Dependent Variable Encoding, Categorical Variables Codings, then for block 0 and each block the
Classification Table, Variables in the Equation, Variables not in the Equation, Omnibus Tests of
Model Coefficients, Model Summary (-2LL, Cox and Snell and Nagelkerke R square), Model if Term
Removed (LR), and with PRINT the Iteration History, Hosmer and Lemeshow Test with its contingency
table, the Correlation Matrix and the intervals of Exp(B). SUMMARY shows the last step of each
block only. CASEWISE lists every case, or those whose studentized residual is OUTLIER or more
away from 0. SAVE adds PRE_1, PGR_1, RES_1, DEV_1, LRE_1, SRE_1, ZRE_1, LEV_1, COO_1 and DFB0_1,
DFB1_1, ... (one per coefficient), or the names given (DFBETA(name) gives name0, name1, ...).
Weighted when a weight variable is set. CLASSPLOT is accepted with a warning that charts are not
drawn yet; EXTERNAL changes nothing. Not supported yet: SELECT, OUTFILE.

Example

LOGISTIC REGRESSION VARIABLES=bought WITH age income region /CATEGORICAL=region /METHOD=FSTEP(LR) /PRINT=GOODFIT CI(95) /SAVE=PRED.

statscraft help LOGISTIC REGRESSION prints the same text in the terminal. All commands