CONFIDENCE-INTERVALS FOR AUTOCORRELATIONS BASED ON CYCLIC SAMPLES

Clinger and Van Ness have shown how to sample a stochastic process in a periodic manner such that the autocorrelation function for that proc ess can be estimated at all lags. We derive approximate confidence int ervals for these estimates, focusing on low-order autoregressive movin g average processes. Our methods are illustrated with an example from plant disease epidemiology.