This is the first-order autocorrelation estimate (shown on
Line 20 of the regression ANOVA tab). Whenever this coefficient is zero (no autocorrelation) the D-W
statistic is about 2.0. When a model is
characterized by autocorrelation this statistic is used to create the Durbin
corrected data file. In ORS the D-W
test is used to determine whether the autocorrelation Rho is significantly
different from zero. The null hypothesis
states, Ho: Rho = 0. The result of the
hypothesis test is reported as: Reject,
Inconclusive, or Do Not Reject. A result of Reject is an indication of a
serious problem. Inconclusive suggests that the test does not not have
sufficient observations to compute a definitive result. Do Not Reject -- autocorrelation is not a
significant problem in this model.