Future Research

No single research project can hope to fill all of the gaps in our knowledge of the incidence, consequences, and costs of crime. This research has taken a comprehensive look at both incidence and victim costs, summarized the best estimates currently available, and, in some instances, provided new estimates where major gaps existed. Along the way, this project identified a host of future research issues. These fall into three general categories: improving estimated incidence, improving cost estimates, and applying the costs in benefit-cost and demographic analyses.

Improving incidence estimates. For many Americans, crime is a secret. Consequently, crime statistics are prone to undercounting. That is especially true of rape and domestic violence. Recent surveys show that Americans are increasingly willing to talk about these problems. Partly from a lack of data, various researchers and organizations have undertaken surveys to estimate the incidence of crimes such as rape, domestic violence, and child abuse. Unfortunately, these surveys suffer from lack of consistent definitions and often are discredited for including incidents that are not criminal or not deemed criminal in the minds of the victims. It is particularly difficult to compare various estimates because of methodological differences. For example, one survey might ask for all victimizations that occurred during the past year, while another asks for the "most serious" that occurred during the past 6 months. Even if they had consistent victimization definitions, these two surveys would differ due to response error (e.g., "telescoping" the timeframe so that an event that happened 7 months ago is included in a study measuring the crime over a 6-month period). They would also differ because of victims' subjective answers to the "most serious" question. A new version of the NCVS incorporates a more probing approach to these issues and begins to address this undercounting problem in a systematic way that can be analyzed over time.

We still know very little about the true extent and consequences of violence, abuse, and neglect of children. This study's preliminary estimates of nondomestic assaults against children under age 12, for example, indicate that more than 500,000 children under age 12 may be assaulted or raped annually. Although these estimates are based on reasonable and conservative assumptions, they are not based on direct survey or other direct measurement techniques. We know even less about the severity of injury for these child victims and about social service utilization. Thus, further research in this area should be a priority.

Improving cost estimates. A particular research priority seems to be the long-term impact on earnings potential for victims of domestic violence and child maltreatment. The frequency of physical injury due to different types of child maltreatment, as well as recent data on social/victim service intensity and cost, also is lacking.

The cost estimates for domestic assault and especially for child abuse and neglect are probably incomplete. They do not fully capture the effects of these crimes. Repeated victimization can shatter lives, reducing the earning capacity of victims who lack the self-confidence to pursue educational opportunities they might have in the absence of the abuse. Child abuse may also lead to intergenerational violent abuse. For example, Widom (1992) finds abused and neglected children are 1.38 times as likely as other children to commit violent acts. Using this number, the study's estimated rate of child maltreatment would imply that 13 percent of all violence can be linked to earlier child maltreatment. In theory, the costs of these induced crimes, including the incarceration costs, could properly be added to costs of the original crime. Doing so would substantially raise the costs per maltreated child. In practice, however, adequate data on the causal connection between child abuse and subsequent violence is not yet available.

A serious gap in this research effort stemmed from the lack of nationally representative contextual and demographic data about murder victims. For example, lack of data about income levels of murder victims forced this study to use national average earnings by age and sex to estimate earnings lost to murder. The first fledgling attempt to fill the murder data gap was the 1986 National Mortality Follow-Back study. That study covered all deaths. A more focused study is needed that captures NCVS-like data for murder victims.

This study's work on mental health care costs is preliminary in nature and merits followup with a larger sample and more extensive pretesting. A victim service agency survey should be coordinated with this survey. Challenges in that related