1- What is the statistical Model of your study? Two-way ANOVA?

2- Obviously your statement is wrong

I would have thought if between subjects suggests no sig difference between groups for factor two, that none of the parameter estimates should be statistically sig.

3- Upload some images of SPSS outputs and tables. In particular Table Pairwise Comparisons for Interactions.

It's an assignment, I probably shouldn't post online. You may email me at [email protected] and I can send you SPSS output.

Honestly, this is probably the most confused I have been with stats in my experience. I mean I've had my hurdles before, but this one I feel certainly has been the most challenging.

The trouble is , I have chosen a factorial model, but the question asks for GLM. In the information we have been given, there is no clarity as to whether glm must or must not include a covariate (Andy Fields text suggests not, pg 920)... I don't believe either of the factors (gender and year level) to be covariates (were looking at coping scores of gender x year level (7,8&9). I expressed this to the teacher, who in a round about way agreed.

The question also states to take special care to interpret marginal means... but again, this pushes me toward thinking that there should be a covariate.

My feeling is, either I am entirely wrong, or the question is a dodge copy paste from a previous class' assignment that's been altered without consideration for an over thinker ....

Hello Lauren,

0. Yes, it walks like and quacks like a 2 x 3 ANOVA. SPSS GLM (univariate) will certainly handle that.

1. As the main effect for the "B" factor was non-significant, I don't see the rationale for having chosen to run post hoc tests on that factor.

2. GLM/ANOVA may or may not include covariates.

3. If your design is unbalanced, be sure to ask for estimated marginal (EM) means.

4. You can always ask the instructor to elaborate on the assignment if in doubt.

Good luck with your work.