Repeated ANOVA interpretation

#1
Could you please advise me on the best way to figure out whether my independent variables (low suicide risk, medium & high risk) affects the extent of improvement in depression scores taken at two time points (Pre-intervention and post intervention). I have been using repeated ANOVA on SPSS and I wonder if the "SuicideRisk * time" measure of Within Subjects Measures tells us whether the suicide risk affects the Mean difference in Pre-Post scores? My data has a normal distribution.
Many thanks
 

katxt

Active Member
#2
It should.
Here is an easier to understand method which should give the same answer.
Find the prepost difference for each subject. Each subject now has one difference score.
Do a two sample t test between the two groups using these differences.
 
#3
Thank you Katxt. But the problem is that the Independent Variables are nominal (low, medium or high risk) so the mean can't be calculated, would your method work in this case? I just wondered whether the "SuicideRisk * time" looks at the effect of suicide on Pre-Test scores and Post Test Scores but not specifically at the DIFFERENCE because I'd like to know if someone who is high risk of suicide will do significantly better/worse on this intervention compared to low risk and medium risk.
Thank you!
 

Karabiner

TS Contributor
#4
Suicide * time is an interaction. An interaction is a difference of differeneces.
This particular interaction tells you whether the differences between risk groups
changes over time. Or, to put it the other way around: it tells you whether the
pre-post change is different between groups high, medium, low.

If you have a risk score instead of an 3-level risk categorization, it could
perhaps be useful to use the score instead of high/medium/low, because
categorization destroys information.

With kind regards

Karabiner
 
#5
Thank you all.
I do have a risk score but now which method should I use? Linear Regression with Risk score as my Independent Variable and Pre-Post Differences as my dependent variables/outcome variables?
 

Karabiner

TS Contributor
#6
You can use the difference as dependent variable in a regression.
Or, you peform a "mixed" anova where you use the continuous score
as "covariate" instead of the categorized score as factor. In some
software packages, the covaritae*repeated-measures-factor interaction
is not be provided automatically and must be requested.

With kind regards

Karabiner