Oneway multivariate analysis of variance (MANOVA) (STB-6: smv4) ------------------------------------------------- ^manova^ grp_var dep_vars, ^m^atrix Description ----------- ^manova^ allows the user to perform a oneway multivariate analysis of variance on two dependent variables and one independent variable. The latter may be over two or more levels or groups. The ^matrix^ option displays W (within-groups sum of squares and cross-products) and T (total SS and CP) matrices together with their respective determinants. Wilk's Lambda is reported with an F statistic and corresponding significance. In the case of non-factorial designs, lambda is the ratio of the determinant of W to the determinant of T. It is a multi- variate test evaluating if the groups significantly differ in their measures on the set of dependent variables. The displayed ANOVA table provides univariate F tests for each dependent var- iable. The F tests show whether the levels or groups differ for each dependent variable. Example ------- The following example consists of a data set comprised of one independent variable with four levels (coded as 1 to 4), and two dependent variables. . ^manova iv dv1 dv2, matrix^ Oneway Multivariate Analysis of Variance Number of obs in model = 33 Number of groups = 4 Model degrees freedom = 3,29 Variable | HypSS ErrSS HypMS ErrMS F Prob>F -------------------------------------------------------------------- dv1 | 7449.2531 32410.9893 2483.0844 1117.6203 2.22 0.1069 dv2 | 23481.0713 43160.9893 7827.0238 1488.3100 5.26 0.0051 ____________________________________________________________________ Wilk's lambda (df:6,56) = 0.3327 6.85 0.0000 W matrix 32410.9893 32756.2393 32756.2393 43160.9893 T matrix 39860.2424 40946.8788 40946.8788 66642.0606 W determinant = 325919149.1571 T determinant = 979721808.9394 Author ------ Joseph Hilbe, Editor, STB FAX 602-860-4331