Conventionally, time series have been studied either in the time domain
or the frequency domain. The representation of a signal in the time
domain is localized in time, i.e . the value of the signal at each
instant in time is well defined . However, the time representation of a
signal is poorly localized in frequency, i.e. little information about
the frequency content of the signal at a certain frequency can be known
by looking at the signal in the time domain . On the other hand, the
representation of a signal in the frequency domain is well localized in
frequency, but is poorly localized in time, and as a consequence it is
impossible to tell when certain events occurred in time. In studying
stationary or conditionally stationary processes with mixed spectra, the
separate use of time domain and frequency domain analyses is sufficient
to reveal the structure of the process . Results discussed in the
previous chapters suggest that the time series analyzed in this book are
conditionally stationary processes with mixed spectra. Additionally,
there is some indication of nonstationarity, especially in longer time
series.