search

andrewmcaleavey / ReliableTrendIndex

R package for analysis of "reliable" changes
GNU Affero General Public License v3.0
1 stars 0 forks source link

Behavior for 3 time points #13

Open andrewmcaleavey opened 2 years ago

andrewmcaleavey commented 2 years ago

At present, because we estimate the RTI with a free intercept, when there are three time points the RTI recreates the RCI (oversaturated model). The alternative is to fix the intercept, which would effectively treat one observation as observed without error, but model the linear trend with error. This is consistent with the RCI, which treats both observations as fixed and the observed difference score as measured with error.

I did not want to implement this initially, but it would likely be empirically superior in detecting trends with 3 time points, so it should at least be an option if not the default. But default is possible too. Requires some study to make default.

andrewmcaleavey commented 1 year ago

This is being addressed in the branch threeTimePoints.