I have a problem I cannot get my head around. I have 22 point estimates and standard errors calculated for the costs associated with a social programs that starts in January (month 1 and goes on for 22 months). Each month new participants joing the program.
The data structure looks as follows (it is not the real data though):
point estimate standard error month observations (N)/participants in the program 148.05 20.08 1 (january) 355 70.98 2.66 2 (febraury) 799 -140.07
Now, my results show that the effect of the program are significant during the first 5 months but then are no longer. And my guess is that it is due to the changing population size. I imagine I would need some kind of weighting to compare a result I obtained with 1500 observations and one with 355 to reach an overall evaluation of whether the effects are globally significant over the 22 months.
Does anyone have an idea with respect to this problem? Any help is appreciated,