“…First, we chose one of the most common AR structures in the behavioral sciences, an AR model of order 1 with two sample size configurations: (1a) T = 100, N = 1, a configuration commonly utilized in the time series literature to evaluate finite-sample performance of statistical approaches; and (1b) T = 14, N = 100, as commonly seen in many ILD studies in psychology (Armstrong et al, 2010;Beidel, Neal, & Lederer, 1991;Roche et al, 2016;Steger & Frazier, 2005). Second, we simulated a complex AR structure with sharp shifts using a seasonal autoregressive integrated moving average (SARIMA; with N = 100, T = 14) (Box & Jenkins, 1976;Box et al, 2013;Brockwell & Davis, 2002;Reinsel, 2003).…”