Hello,

I am trying to estimate the effect of Drug Law (coded 0/1) on recidivism rate using fixed effects modeling (command xtreg) and state-level panel data. My data includes 50 states and covers 1994-2016. I also included a continuous measure year to control for the effects of time. It appears that the Drug Law had an effect on recidivism when I limit my period of time to 1994-2008 (that is, I deleted years 2009-2016 from the sample). However, when I extend the time period under investigation to 1994-2016, the effect of the Drug Law disappears and dilute. Is there any statistical strategy to show in what year the effect of Drug Law is no longer significant? When I computed an interaction Law*year, the interaction term is not significant. Thank you.
Sylwia