This paper by Jerry Hausman and Daniel McFadden presents two sets of computationally convenient specification tests for the multinomial logit model, a widely used model in applied econometrics. The first test is based on the Hausman specification test procedure, which involves eliminating one or more alternatives from the choice set to see if the underlying choice behavior from the restricted choice set obeys the independence from irrelevant alternatives (IIA) property. The second set of tests is based on classical test procedures, such as the Wald, likelihood ratio, and Lagrange multiplier tests, applied to a nested logit model, which allows for more flexible patterns of similarities among alternatives. The authors apply these tests to an example of consumer choice of clothes dryers, using data from the 1975 WCMS survey. They find that both sets of specification tests lead to a decisive rejection of the multinomial logit model, indicating that the model is misspecified. The paper also discusses the performance of the classical tests, comparing their size and power characteristics with those of the Hausman test. Surprisingly, the Wald test and Hausman test have approximately equal power, while the likelihood ratio test is less powerful, and the Lagrange multiplier test is the least powerful. These findings suggest that the choice between different specification tests may depend on the specific application and the trade-offs between computational complexity and statistical power.This paper by Jerry Hausman and Daniel McFadden presents two sets of computationally convenient specification tests for the multinomial logit model, a widely used model in applied econometrics. The first test is based on the Hausman specification test procedure, which involves eliminating one or more alternatives from the choice set to see if the underlying choice behavior from the restricted choice set obeys the independence from irrelevant alternatives (IIA) property. The second set of tests is based on classical test procedures, such as the Wald, likelihood ratio, and Lagrange multiplier tests, applied to a nested logit model, which allows for more flexible patterns of similarities among alternatives. The authors apply these tests to an example of consumer choice of clothes dryers, using data from the 1975 WCMS survey. They find that both sets of specification tests lead to a decisive rejection of the multinomial logit model, indicating that the model is misspecified. The paper also discusses the performance of the classical tests, comparing their size and power characteristics with those of the Hausman test. Surprisingly, the Wald test and Hausman test have approximately equal power, while the likelihood ratio test is less powerful, and the Lagrange multiplier test is the least powerful. These findings suggest that the choice between different specification tests may depend on the specific application and the trade-offs between computational complexity and statistical power.