ORS computes and reports the Durbin-Watson (D-W) statistic
at the 95% confidence level. The D-W
d statistic is meaningful for time
series data only.
The D-W statistic tests for first-order autocorrelation in
the residual error terms. One of the
regression assumptions of the OLS technique requires successive values of the
residual error term to show NO discernible pattern. A violation of this assumption occurs when successive values of
the residual error term show either postive or negative autocorrelation. When the D-W statistic is about 2.0, there
is confidence that there is an absence of autocorrelation. The hypothesis test reported by ORS is
automatically computed at the 95% confidence level.