I really need help with an analysis I am doing. The study involves a comparison of two different treatments for bulimia (weight concern treatment and perfectionism treatment) and a combination of the 2 treatments, as well as a control group over time.
In other words, participants were randomly allocated to one of four groups:
(a) an active control group which involved group counselling sessions but no targeting of perfectionism or weight concern; (b) a perfectionism intervention group who received a treatment aimed at reducing perfectionism; (c) a weight concern treatment group who received a treatment aimed at reducing perceived weight concern; and (d) a combined treatment program where both perfectionism and weight concern were targeted.
I therefore have a factorial design with two independent variables: perfectionism treatment (yes, no) and weight concern treatment (yes, no).
The within-subjects factor is Time (i.e. T1 before treatment and T2 post treatment).
The dependent variable is disordered eating symptoms (continuous).
I initially ran a 3-way mixed-model ANOVA, and found a significant 3-way interaction.
To further test this interaction, I then ran a factorial ANOVA. To do this, I computed a new dependent variable which was the change in disordered eating symptoms between T1 and T2 (I did T1 – T2, so a higher positive change score indicates greater improvement in disordered eating symptoms).
Here are the results from my factorial ANOVA (using change in disordered eating scores as my DV):
My problems are as follows:
I’m not sure how best to present the results (i.e. graphically, table??). I think it would make sense to present the change over time in a graph, a bit like the one I have created, above. I’m not sure if my graph is correct, and if it shows the interaction? The difficulty I’m finding is dealing with the fact that the treatment conditions are not part of one single independent variable with several levels, but rather are separate independent variables. I am used ot seeing them as IV of “treatment group” (perfectionism treatment, weigh concern treatment, combination treatment, control), but instead here we have “perfectionism treatment” (yes, no) and “weight concern treatment” (yes, no)
I don’t know whether I need to do further tests, i.e. simple effects analysis following this factorial ANOVA or if that’s it? I tried running 2 t-test but just got more confused
I'm also confused by the fact that in the initial mixed-model ANOVA I got a barely significant interaction between the 2 treatments in the Test of Between Subjects Effects box (p = 0.55), however the effect was significant in the factorial ANOVA. Also, the main effect of weight concern treatment was not significant in the Test of Between Subjects Effects box (p = 0.085) in the mixed-model ANOVA, however it was significant in the factorial ANOVA (p < 0.001). Can I conclude that the weight concern treatment IV has a main effect as p is significant in the factorial?? What about the fact that it showed as non-significant in the 3-way?
Am I correct in thinking that there wasn’t much difference between the effect of the perfectionism treatment and the combination treatment and the difference in means is small (and the lines on the graph are nearly on top of each other and parallel)? Where can I get the significance measure for this and for the other treatment groups?
Any help would be greatly appreciated!