Statscraft 0.7.0, release v0.7.0

GLM

General linear model: one dependent variable as in UNIANOVA, or repeated measures with WSFACTOR.

Syntax

GLM dependent [BY factors] [WITH covariates]
   [/WSFACTOR=name levels [POLYNOMIAL | contrast] ...] [/WSDESIGN=effects] (GLM)
   [/METHOD=SSTYPE(1 | 2 | 3 | 4)] [/INTERCEPT=INCLUDE | EXCLUDE] [/MISSING=EXCLUDE | INCLUDE]
   [/CRITERIA=[EPS(1E-8)] [ALPHA(0.05)]] [/REGWGT=var]
   [/PRINT=[DESCRIPTIVE] [HOMOGENEITY] [PARAMETER] [ETASQ] [OPOWER] [LOF]]
   [/EMMEANS=TABLES(OVERALL | factor[*factor...]) [WITH(covariate=value | MEAN ...)]
             [COMPARE[(factor)] [ADJ(LSD | BONFERRONI | SIDAK)]]] ...
   [/POSTHOC=factor ... (tests)] ... [/CONTRAST(factor)=DEVIATION[(ref)] | SIMPLE[(ref)] |
   DIFFERENCE | HELMERT | REPEATED | POLYNOMIAL[(metric)]] ...
   [/LMATRIX=['label'] effect coefficients ... [; ...] | ALL coefficients] ... [/KMATRIX=values]
   [/SAVE=tempvar[(name)] ...] [/PLOT=...] [/DESIGN=effects].
effects: INTERCEPT, a factor or covariate, a*b or a BY b, a(b) or a WITHIN b, a(b(c)), x*x, x*a.
tempvar: PRED WPRED RESID WRESID DRESID ZRESID SRESID SEPRED COOK LEVER. tests: as in ONEWAY
/POSTHOC (LSD BONFERRONI SIDAK TUKEY SCHEFFE GH T2 SNK BTUKEY DUNCAN GT2 GABRIEL FREGW QREGW T3 C
DUNNETT DUNNETTL DUNNETTR). Without DESIGN the model is the intercept, the covariates and the full
factorial of the factors. A case missing on any variable is left out; the weight is rounded to a
whole number. The design matrix has one column per effect level (only combinations that occur);
sweeping it gives a generalized inverse, and a column that adds nothing (EPS) is redundant, its
parameter 0. Tables: Between-Subjects Factors, Descriptive Statistics, Levene's Test of Equality
of Error Variances (absolute residuals by cell; based on the mean, the median, the median with
adjusted df and the trimmed mean for a full factorial without covariates), Tests of
Between-Subjects Effects (Type I sequential, II, III or IV sums of squares, with partial eta
squared, noncentrality and observed power), Parameter Estimates, Lack of Fit Tests, Contrast
Results and Test Results for each CONTRAST and LMATRIX, Estimated Marginal Means (cells weighted
equally, covariates at their means or the WITH values) with Pairwise Comparisons and Univariate
Tests, and the post hoc tables of ONEWAY on the observed means with the model's error. T2, T3,
GH and C need a model with one factor. SAVE adds PRE_1, RES_1 and so on, or the names given.
GLM with WSFACTOR: the dependents are the levels of the within-subjects factors (the first
factor's levels change slowest); Multivariate Tests, Mauchly's Test of Sphericity, Tests of
Within-Subjects Effects (sphericity assumed, Greenhouse-Geisser, Huynh-Feldt, lower bound), Tests
of Within-Subjects Contrasts and Tests of Between-Subjects Effects on the average.
PLOT is accepted with a warning that charts are not drawn yet. Not supported yet: GLM with several
dependents without WSFACTOR (the multivariate model), RANDOM, TEST, ROBUST, OUTFILE, MMATRIX,
MEASURE, POSTHOC VS, CONTRAST SPECIAL, PRINT GEF, TEST(LMATRIX), MBP, BP, F and WHITE; with
WSFACTOR also EMMEANS (left out with a warning), POSTHOC, CONTRAST, LMATRIX, SAVE and PRINT
PARAMETER, HOMOGENEITY and LOF.

Example

GLM t1 t2 t3 BY group /WSFACTOR=time 3 POLYNOMIAL /WSDESIGN=time /DESIGN=group.

statscraft help GLM prints the same text in the terminal. All commands