Help (ANOVA?) - Dissertation due on Wednesday!

#1
Hi,
I'm currently writing up my final year year psychology dissertation and I'm having some stats problems. My experiment involved exposing participants to two types of violent media (violent video game, violent film), with measurements taken at baseline, after exposure, and while watching footage of real violence. I was looking to see which had to the greatest desensitising effect while watching the footage of real violence. There were 12 participants, 6 per condition (game or film). I also took into consideration gender, and average weekly exposure to violent media.

I'm pretty sure I need to apply an ANOVA here but I'm not sure which one. I want to take the interacting effect of gender and average exposure to violent media into consideration. Would I use a series of two-way anova's? Or would it be a two-way anova with 3 experimental factors (media type, gender, exposure)? Or maybe a 3 way anova with repetition? Does the low sample size limit what tests I can apply?

Any help would be greatly appreciated!
 
Last edited:
#2
Not very professional answer, maybe wrong.
Just think this is a very interesting problem. I think it is kind of split-plot. (not 100% sure)

The participants nested in the media effect and kind of error for main effect.

I need more information about how do you setting the gender and weekly disposure.
 
#3
Well yes it's definately a mixed design - between in the sense that there are differences in the groups (video game and film), and within in the sense that there are a number of observations. I don't know much about split-plot designs other than relating to mixed designs however.

I didn't set gender or weekly exposure in advance of the study. The groups were equated as best as possible, with 2 males and 4 females in one group, and 3 males and 3 females in the other. Average weekly exposure was recorded via a questionnaire whereby participants said either >1 Hour, or 1-3 hours. So I suppose I'm treating gender and average weekly exposure as covariates.

However, having suggested this to my stats lecturer she informed me that my sample size was to small for an Ancova? Unfortunately, I'm still waiting to here back from her. What else could I apply?

Grateful for any input at all!
 
#4
Spss

Okay I've figured out that using SPSS its definately a General Linear Model with repeated measurses. The only problem is I'm having trouble interpreteing it. Basically I need to out whether there is a significant difference between video games and films between the post measure and the measure during footage of real violence (levels 2 & 3). Where should I look for this? I have also entered in gender and average weekly exposure as covariates. Should I assume these have been taken into consideration?

Just to recap -
Within-Subjects Factor - GSR_Measures -(1) Baseline, (2) Post, (3) During Footage

Between-Subjects Factor - Media Type (Video game/Film)

Covariates - (1) Gender, (2) Average Weekly Exposure (<1 Hour/1-3 Hours)

Any ideas?