A New Class of Autoregressive Models for Time Series of Binomial Counts

The binomial AR(1) model of McKenzie (1985) for time series of binomial counts has a well-interpretable structure and applies well to several real-world problems. After a brief review of important properties of this model, we propose and investigate a new class of pth order autoregressive models, which coincide with the binomial AR(1) model for p=1. Special cases of this new model family are discussed, each having a different autocorrelation structure. A real-data example demonstrates that these higher-order models have a great potential to be applied in practice.