COMPARISM OF THE POWER OF SOME SPECIFICATION ERROR TESTS

Because there are a number of tests for specification error in detecting the errors of omitted variables or incorrect functional form, one rarely knows the best test to use. This paper compares the power of the test RESET (regression specification error test) to that of Durbin-Watson in detecting the errors of omitted variables or incorrect functional form in a regression analysis using the Bootstrap method of simulation to see which test is better. The overall results show that the RESET is more powerful at all sample sizes in detecting a non zero disturbance mean (i.e in detecting specification error) as a result of incorrect functional forms or omitted variables in a regression model.