POWER_ITSA: Stata module to compute power for single and multiple-group interrupted time series analysis

power itsa computes power for a specified number of time periods n() in a single-group or multiple-group interrupted time series analysis (ITSA), using simulation. For a single-group ITSA, the process involves (1) generating a time series using itsadgp, based on the user-inputs replicating the coefficients of a single-group ITSA regression model (see Remarks section below)-- which may include AR(k) autocorrelated errors; (2) estimating a single-group itsa model using either regression with Newey-West standard errors or Prais-Winsten AR(k) regression via praisk; (3) testing if the difference in pre- and post-intervention trends = 0, or when level is specified, testing if the change in level in the period immediately following introduction of the intervention (compared to the counterfactual) = 0; and (4) repeating this process the number of times specified in reps(). For a multiple-group group ITSA, the process involves (1) generating one time series for the treated unit and one or more time series for the controls, replicating the coefficients of a multiple-group ITSA regression model (see Remarks section below)-- which may include separate autocorrelations for the treated unit and controls; (2) estimating a multiple-group ITSA model using either regression with Newey-West standard errors or Prais-Winsten regression; (3) testing if the difference in differences of the pre- and post-intervention trends = 0, or when level is specified, testing if the differences in the change in level in the period immediately following introduction of the intervention (compared to the counterfactual) = 0; and (4) repeating this process the number of times specified in reps().