I have a data set that would be best classified as a group randomized
design. In this data set, no attempt was made to measure the same subjects
at two consecutive time points. The thought was that randomly selected
subjects were equally representative of the population. Between the
baseline and follow-up measurement, an intervention was administered to a
subset of the population. In the post measurement, half the sample drawn
came from that subset of the population.
The research question is did the intervention change behavior.
Given the nature of the design, I believe some adjustment must be made to
the degrees of freedom (e.g., Murray & Blitstein, 2003) for the test of
intervention effect. Can this be accomplished using either lme or glme?
Respectfully,
Frank Lawrence
|