Hi, I'm researching a treatment and using the CMAI short form as a measure of participant agitation. The CMAI short form breaks questions into 3 categories of agitation. I administered the CMAI at pre, mid and post test and have scores for each of the categories at each of the times. What is the best way to analyze this data to know if there have been changes from pre to mid to post?
The best way I think is to calculate the changes between pre to mid and mid to post for each patient and then with that new variable estimate a one group test (Wilcoxon or similar) whom CI must not include 0 to be significant. Same way from mid to post. Otherwise you can use a paired wilcoxon test from pre to mid and from mid to post.
This is an intervention problem. The best thing to do is to fit an arima model to the pre-intervention data. On the basis of this model obtain post-intervention forecasts. Find the difference of these post-intervention data and forecasts. Model these difference to obtain the transfer function. If this transfer function is statistically significant then the impact of the intervention is significant. Otherwise it isn't.
I would need to know more about the treatment. Did all participants receive the same treatment? I take it that this is your IV. So you are testing agitation at 3 points in time during treatment (with 3 measures of agitation at each interval). I looked the the CMAI short form (14 items). It is temporally based. Hopefully you accounted for this in your treatment intervals. The manual discusses scoring strategies for researchers (see page 5):
https://shine-dementia.wikispaces.com/file/view/Cohen-Mansfield+Agitation+Inventory+(CMAI).pdf
I would recommend paired t-tests or repeated measures ANOVA:
https://www.youtube.com/watch?v=lLebzE_ipuc
Good luck with your study...Nick
- Badges
- Science topic