Hi everyone,

I have a problem with an analysis that's part of a statistics course. We were handed out a dataset (total n=456) that's supposed to measure the impact of background graphics on player experience in a specific video game genre. Background graphics were either high or low detail as well as animated or not animated. This leads to four treatments.

But there are five treatments, since they added one with no background graphics at all (thus, no animation either), considered as being some kind of a control group. Also it's a between subjects design with independent groups. The number of observations per group is around 90.

I'm interested in main effects as well as the interaction of detail and animation on player experience. Plus I wanted to include the control group to see, if the presence of background graphics does matter at all. Since there are three continous dependent variables representing the construct of player experience, I figured I could conduct a two-way MANOVA.

The treatment variable consists of five numbers representing each treatment. So I recoded it into two new nominal variables, namely graphics (high, low, none) and animation (animated, not animated). The data met all the necessary assumptions (normality, homogenity, etc.) so I ran the MANOVA in SPSS.

My problem is: the way I created the two independent variables leads to six cells, but I only have 5 groups (since there obviously can't be a "no background graphics / animated" group). SPSS only reports on those 5 groups, as I expected. There were no significant differences.

Would it still be valid to report the results even though there's one cell missing?

Any suggestions are greatly appreciated! I couldn't find anything helpful using google. Sorry for any mistake concerning grammar, I'm not a native speaker