Telling what yesterday's news might be tomorrow: Modeling media dynamics1

In this article, we discuss the use of time series models in communication research. More specifically, we consider autoregressive and moving-average processes, which together constitute the autoregressive integrated moving average-framework (ARIMA). This approach provides a comprehensive framework to deal with the essential issue of stationarity and to model the dynamics of any time series by estimating the autocorrelation structure. Underlying the models are questions as to what extent news tends to reproduce itself and how news flows adjust after deviations from the normal news stream. The data illustrating the models consist of visibility-scores of the immigration issue in Dutch national newspapers. The empirical analysis demonstrates that the impact of immigration figures on this visibility is not significant when the ARIMA-framework is applied, while an analysis using OLS suggests a positive influence.