I Just find out a not so obvious assumption in SPSS.
When you need to estimate several ANOVA, ROC curve, even maybe with T TEST, and you replace in the depedent variable section for more than one variable, like this:
UNIANOVA
VARD1 VARD2 BY group
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/POSTHOC=group(BONFERRONI)
/PLOT=PROFILE(group)
/EMMEANS=TABLES(OVERALL)
/EMMEANS=TABLES(group) COMPARE ADJ(BONFERRONI)
/PRINT=ETASQ HOMOGENEITY DESCRIPTIVE
/CRITERIA=ALPHA(.05)
/DESIGN=group.
What SPSS would do regarding the listwise deletion cases is to restrict your analizable sample data, to the common cases which have valid values for the 3 variables as a whole. All in all, this means all the estimates are restricted to this maybe smaller data sample of the overall data available to estimate all the parameters.
If you have full data (no missing cases), or just a few missing data, is not going to cause too much of trouble. Maybe just a small deviation for the parameters which would not affect decisions over scale selection, for example. Nevertheless, if you got at least one variable of them with a considerable loss of data (lets say 50% valid cases for the overall data set), all the estimates will be calculated using this restricted list wise cases instead of the full available data for each pair of variables (vard1 with group & vard2 with group) biasing seriously the parameter estimates (F, n2p, for this case).
So be careful.
But if you still need to get the n2p for each pair of variables for a set of 36 dependent variables for example, employing cross reference 'email list' from WORD, you can automatize the production of a proper list of SPSS syntax for each pair, avoiding to write down 36 sets of code line.
UNIANOVA
<<insert field here>> BY group
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/POSTHOC=group(BONFERRONI)
/PLOT=PROFILE(group)
/EMMEANS=TABLES(OVERALL)
/EMMEANS=TABLES(group) COMPARE ADJ(BONFERRONI)
/PRINT=ETASQ HOMOGENEITY DESCRIPTIVE
/CRITERIA=ALPHA(.05)
/DESIGN=group.
===> edit documents:
UNIANOVA
vard1 BY group
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/POSTHOC=group(BONFERRONI)
/PLOT=PROFILE(group)
/EMMEANS=TABLES(OVERALL)
/EMMEANS=TABLES(group) COMPARE ADJ(BONFERRONI)
/PRINT=ETASQ HOMOGENEITY DESCRIPTIVE
/CRITERIA=ALPHA(.05)
/DESIGN=group.
UNIANOVA
vard2 BY group
/METHOD=SSTYPE(3)
/INTERCEPT=INCLUDE
/POSTHOC=group(BONFERRONI)
/PLOT=PROFILE(group)
/EMMEANS=TABLES(OVERALL)
/EMMEANS=TABLES(group) COMPARE ADJ(BONFERRONI)
/PRINT=ETASQ HOMOGENEITY DESCRIPTIVE
/CRITERIA=ALPHA(.05)
/DESIGN=group.
I'll try to re-write this post with a proper example, in a few days.
