[SPSS-ANOVA] Should I use a mixed ANOVA design?


New Member
Good evening

I am writing to inquire how I should tackle the data for my dissertation research paper. I have two main problems that I will point out in this post.

It starts off pretty simple. I want to determine wheter there are any significant differences in reaction times between two groups, one with depression and a control group, in respect to a particular task.

The task is called the "Self/ Other Differentiation Task" (Bradford, 2015) and it was developed, briefly, to asses reaction times in response to certain scenarios where a perspective shift (from self to other or from other to self) may occur. Furthermore, this task was sort of based on the theory of mind Smarties task and it also has another condition, for Expected vs. Unexpected contents.
I'll offer an example of an item so it becomes clearer. "You are searching for your keys. Where would you look?" For this question there are three possible answers, one of which is, say, "purse". When the participant chooses the purse, there are two possible scenarios: the purse contains keys (Expected) or the purse contains an egg (Unexpected). After this phase, the perspective shift comes in. The next question can be: "Before you saw what's inside, what did you think was in the container?" (No perspective shift) or "Before John saw what was inside, what did he think was in the container?" (Perspective shift). Alternatively, some items may start with "John is searching for his keys...", followed by Expected vs. Unexpected Contents and the Perspective Shift vs. No perspective shift question.

And now it gets really confusing for me. In the original Bradford paper, they use a 2 x 2 x 2 ANOVA to analyze the RTs, with the factors: Shift: Shift vs. No Shift, Contents: Expected vs. Unexpected and Probe: Self vs. Other (whether the final question involved a Self or an Other perspective).
First problem: I cannot completely grasp why they decided to use these factors. Personally, for my dissertation, I was thinking of using just two factors: Shift: Self/Other vs. Other/Self vs. No shift and Contents: Expected vs. Unexpected. Do you believe that would be appropriate? To me the two seem to mean roughly the same thing but I am not sure if that is indeed the case, at least from a statistical point of view. Long story short, is my solution of using only two factors any good or should I stick to their original three factors? May using only two factors affect the results in any significant way?
Second problem: Among all these, how do I also fit in the two groups (Depression vs. Control). I was thinking of a mixed ANOVA design with group as a between subjects factor and the above-mentioned (Shift and Contents) as within subjects factors. Would this model be appropriate for analyzing my data or do you recommend something else?

Sorry for the long and messy post. I hope I explained my research clear enough so that you may provide an answer. Looking forward to your replies. Thank you for your time.


TS Contributor
Self/Other vs. Other/Self vs. No shift
AFAICS, with "no shift" you would collapse self/self with other/other. Would that be appropriate from your theoretical standpoint?

I was thinking of a mixed ANOVA design with group as a between subjects factor and the above-mentioned (Shift and Contents) as within subjects factors.
That's ok, at least in my opinion (an alternative would be multilevel modeling).

With kind regards