stepwedgepower (Q7375140)

Provides simulation-based power, type I error, and operating- characteristic evaluation for stepped-wedge cluster randomized trials. General binary-outcome designs support arbitrary crossover schedules, unequal clusters per sequence, secular trends, flexible cluster-period sizes, reproducible parallel execution, sensitivity grids, and mixed-model diagnostics. A component-based engine supports the four treatment states Control, A, B, and A+B under arbitrary asynchronous schedules, including structurally unobserved cluster-periods, withdrawal, reintroduction, wash-in delays, user-defined carryover decay, component main effects, an A-by-B interaction, estimability audits, multiplicity adjustment, conditional and failure-aware power, and explicit classification of hard model errors, nonconvergence, non-finite contrasts, singular fits, and successful fits. A batch-aware interface represents classic and batched stepped-wedge designs with delayed cluster initiation, structurally unobserved cluster-periods, and calendar-time, time-on-trial, or separate batch-specific time effects, following Kasza, Bowden, Hooper and Forbes (2022) <doi:10.1002/sim.9438>. The cumulative Control-to-A-to-A+B interface from earlier versions remains available. The design and mixed-effects analysis follow Hussey and Hughes (2007) <doi:10.1016/j.cct.2006.05.007>. Intraclass correlations for binary outcomes are converted to random-intercept standard deviations on the latent scale of the logistic-normal model, as reviewed by Eldridge, Ukoumunne and Carlin (2009) <doi:10.1111/j.1751-5823.2009.00092.x>. Monte Carlo uncertainty is reported using exact intervals from Clopper and Pearson (1934) <doi:10.1093/biomet/26.4.404>.