1. Select one or more numeric variables and move them onto the Variables field.
  2. Under Display, select Autocorrelations, Partial autocorrelations, or both.

    Displaying partial autocorrelations requires the solution of a system of equations whose size grows with the number of lags. Be careful about requesting partial autocorrelations to a high number of lags (over 24). Even on a fast machine, this will take much longer than the autocorrelations. If you have a series with seasonal effects and need to look at high lags, look at the autocorrelations alone until you are sure the series is stationary. Then ask for the partials as well.

    You can transform your data by selecting Natural log transform, Difference, or Seasonally difference from the Transform group.

    If you select either Difference or Seasonally difference, enter the degree of differencing in the text box next to the selection. This number must be a positive integer.

    The current periodicity is displayed beneath these options.

  3. Click Options to set the maximum number of lags or the method for calculating standard errors or to display autocorrelations only at seasonal lags.