When data are collected over time for each subject, serial correlation is often present. When data are serially correlated, observartions that are closer together tend to have higher correlations than observations which are farther apart. One implication of serially correlated data is that if an observation is above the subject’s mean level, the next observation taken a short time later will also tend to be above the subject’s mean level. A set of serially correlated observations is often referred to as a time series. In fact, any set of observations collected over time is a time series. Most of the studies of time series analysis have concentrated on the analysis of a single long series of observations taken at equally space time points (Box and Jenkins, 1976). The purpose of a time series analysis may be to determine the correlation structure of the data in order to make forecasts.