In this paper we propose an alternative method to the Durbin-Watson(DW) test for the fitness of a regression model (see, for example, Makridakis [5, pp. 267-258; 303-304; 630-631] and Wilson & Keating [8, pp. 182-184; 234-236] for DW statistic). The proposed method tests whether the residual terms (i.e., actual – model) display any sign of non-randomness by comparing two variance estimators of the residuals. This is a transformation of the DW statistic into a standardized normal statistic, N(0,1) which is readily interpreted; and unlike the DW test does not require a special table. A numerical example is provided.

Date of creation, presentation, or exhibit



Presented at the 2008 Northeast Decision Sciences Institute Annual Meeting Proceedings, March 28-30, 2008.

Copyright 2008 The Authors

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works in February 2014.

Document Type

Conference Paper

Department, Program, or Center

Accounting (SCB)


RIT – Main Campus