burglary, and motor vehicle theft (see U.S. Department of Justice, 1989b, for an expanded discussion of NCS data and methodology).

In some respects, the undercounting of children, transients, and the homeless represents an important omission because these persons are at greatest risk of victimization. Further, the omission of institutionalized cohorts is troubling because persons who sustained very severe injury due to victimization may be institutionalized, receiving ongoing care for long-term temporary or permanent impairment. The costs for such care can be extremely high and probably are not reflected in this analysis. Finally, although this is not a study of incidence of victim injury, calculations of costs per victimization are affected by the accuracy of incidence estimates because the number of incidents serves as the denominator when costs of injury are distributed across all victims.

Estimates of domestic violence, child abuse, and nonstranger victimizations very considerably. Some victims do not report such incidents for a variety of reasons: for example, nonstranger victimization may not be regarded as a crime; the incident may be viewed as a private matter; or the offender may be present during the survey interview, thus constraining discussion. It may be possible to triangulate estimates of such offenses to improve existing incidence data. Cost estimates could also be refined based on the distribution of injuries associated with adult assault and child abuse captured by hospital trauma registries and data bases.

One measure of the possible underreporting of medical costs by victims is represented by gunshot wounds. The NCS data document an average medical cost per injury that ranges from $816.87 (for robbery—gunshot, conscious) to $13,000 (for rape—gunshot, conscious). These figures are significantly different from the $33,159 average medical cost of firearm injury per hospitalized victim and the $458 average medical cost for injured, nonhospitalized victims reported by Rice et al. (1989:xxviii).

They used data on length of hospital stay from NHDS or NASS, or medical cost data from the National Medical Care Utilization and Expenditure Survey, the National Council on Compensation Insurance's Detailed Claims Information data base, or the Comprehensive Health and Medical Program for the Uniformed Services.

This data base contains data on nearly 500,000 injured people. Once a claim enters the sample, all medical costs are tracked until the workers compensation insurer is able to sell the claim