ARIMA vs. Spectral Analysis

Revision with unchanged content. This book aims to introduce the reader to the time-domain and frequency-domain techniques of time series analysis. Both methods are compared and their equivalence is discussed. The application in psychological research is demonstrated on an example of mood diaries, similar to the study of Larsen and Kasimatis (1990). Univariate and bivariate time series analysis is done regarding the question whether time series techniques reveal deeper information than traditional methods. Special methodological issues and possible errors are discussed in terms of technical advancement. Empirical results showed that, contrary to the original study, there were no circaseptum rhythms in mood. Methodological results showed that the equivalence of both approaches can be used to crossvalidate the results for an improved model identification. Aggregation of mood raw data appears to be unjustifiable since mood processes differ very much between individuals. Recommendations are made for the user.