of the subject community to allow the use of the estimates without reservations or adjustments. Unless these conditions are satisfied, the resultant estimates may be of questionable value.

There is no evidence that the Chicago study attempted to consider whether the estimated costs and prevalence rates borrowed from other studies were appropriate to Chicago. In addition, the authors do not appear to have tried to separate real costs from transfer costs, nor did they try to estimate aggregate pathological gambling costs rather than incremental costs due to pathological gambling.

In a study that strays from traditional economic impact analysis, Grinols and Omorov (1995) attempted to determine, using benefit-cost analysis, whether improved access to casino gambling offsets the externality (or spillover) costs associated with pathological gambling. Their study takes a unique approach to the estimation of the net economic effects of gambling. Instead of focusing on a particular geographic area, as most economic impact studies do, they attempted to estimate the effect of increasing gambling accessibility nationwide. They define externality costs as criminal justice system costs, social service costs, and costs due to lost productivity. In order to estimate the per capita social costs due to pathological gambling, they relied on the annual cost estimates per pathological gambler and prevalence rates for pathological gambling computed in earlier studies (Goodman, 1994; Lorenz et al., 1990; Politzer et al., 1981). They do not, however, further the understanding of what constitutes the costs of pathological gambling or the magnitude of these costs. Instead, Grinols and Omorov relied on the work done by others to assign dollar values to the externalities and used these estimates without any attempt to determine whether the estimates were appropriate for the task at hand.

In a study that attempted to identify the benefits and costs associated with gambling, Madden (1991) looked at the socioeco-