I'm predicting college enrollment based on prior enrollment using time series modeling. When beginning the analysis, SPSS asks how to label the data -- yearly, quarterly, etc. There is no periodicity of 6 wherein I could enter August and January enrollment data, so up until now I've acted as though the data were yearly. The ARIMA (2,1,0) did pick up the relationship of data two time periods/lags apart, meaning Fall enrollment was related to prior Fall enrollment, and Winter was related to the Winter prior, but ultimately it seems as though there would be a way to enter semi-annual data and use seasonal models.
Could you use monthly but let the autoregressive order be 6?