Mixed-Design ('Split-Plot') ANOVA - SPSS (Part 2)

[how2stats]

Regarding whether the effect is due the intervention or the control group. Statistically, this means the same thing. Whether the intervention increases/decreases or the control group increases/decreases is not a statistical consideration; it's an interpretation consideration. Sometimes the control group decreases but the intervention group doesn't change from time 1 to time 2. That's still an effect. Whether it is supportive of your hypothesis is a qualitative consideration.

[arif978]

@how2stats Thank you for the explanation.

[how2stats]

If you have a 2*2 design, then you know where the interaction effect is taking place, as it can only be taking place from time 1 to time 2 (and between the two groups). However, if you have more than two levels in your design, then you'll have to decompose the interaction effect by doing a series of 2*2 split-plot ANOVAs to see which ones are significant.

[jonrrobinson]

Great video! Can I ask a question though.

When you have a significant interaction, this tells you that there was a significant difference (over-time/between pre- and post measures) between (only) one of the groups, but not both of them?

If this is correct, and when the interaction is significant, how do you know for which groups the significance is applicable? (For which group was the change over time significant).

I hope that makes sense - thanks again for the great video.