Hi

I would be grateful for your help. I have run a ITSA using Lindens ITSA command for a survival outcome by the monthly averages:

itsa survivemean, force single trperiod(660 663) lag(12) posttrend figure

The graph from this ITSA is attached. I want to compare the “drop” in the two regressions line in the graph excluding the period between the dashed lines, which was a training period for the intervention. So I want to exclude months between 660 and 663 and only measure the drop in survival starting after the training period. I read Lindens “A comprehensive set of post estimation measures to enrich interrupted time-series analysis”, and according to that paer I should use lincom the following way:

lincom _b[_x_t663] + _b[_x_t660 ]

But this command does not give me a believable answer. However this one does, with a minus instead of a plus suign between the two coefficients:

lincom _b[_x_t663] - _b[_x_t660 ]

How can I calculate this “drop”? Thanks for any advice.

Cheers

Pieter Francsois Fouche