Dear Community,
For my research, I would like to perform an interrupted time series analysis to test the effectiveness of an intervention.
Regarding power, I read different information: on the one hand, it is important not to have too few data points, on the other hand, the amount of data points before and after the intervention should be about the same.
As it seems so far, I will not be able to meet both criteria, because the time period after the intervention will be shorter than before the intervention.
My question now is: would it make more sense to collect as much data as possible, i.e. to make the pre-intervention phase as long as possible, or would it be better to keep both intervention periods shorter, but then have fewer data points overall?
Has anyone ever performed an IST analysis and can help me with my question?
This is a good article discussing many aspects of ITSA:
Article Using intervention time series analyses to assess the effect...
From your description you may also want to consider longitudinal models. A super set of notes by Prof. Marie Davidian
is attached. Notice that the intervention does not have to be the second observation in the series and that multiple interventions can take place. Best wishes, David Booth
https://www4.stat.ncsu.edu/~davidian/st732/notes.html
- Badges
- Science topic