Exploring Differences in Estimates of Visits to Emergency Rooms for Injuries from Assaults Using the NCVS and NHAMCS

Date of Original Version



Working Paper

Abstract or Description

Researchers seeking to provide a better understanding of crime statistics tend to compare survey-based statistics such as the NCVS with data from police administrative series like the UCR. Because these two types of data collections systems are so different, simple direct comparisons are of little value regarding limitations inherent to a particular data collection system. This chapter explores the NCVS data using a different perspective that compares data from the national crime survey of population with those from a national survey of establishments-the National Hospital Ambulatory Care Survey (NHAMCS). This comparison provides an understanding of how the design, instrumentation and procedures of the NCVS may influence estimates of interpersonal violence, particularly that component of violence resulting in injuries treated in hospital emergency rooms. The estimates of emergency room visits for injuries due to violence obtained from the NCVS are considerably smaller than those from the NHAMCS. The analyses include a series of adjustments to these estimates that explore the role of features specific to each survey in the observed differences. The household sampling frame employed in the NCVS receives special attention as a potential source of the observed differences. Investigating this source of divergence is particularly important, since many of our major social indicators on the economy and participation in government programs depend upon household surveys. If some population groups are under-represented in the household sampling frame used in Census surveys, and this under-coverage results in underestimates of violence, this finding could have implications for the use of the household frame to estimate the magnitude of other problems that disproportionately affect marginal populations, such as unemployment, poverty, drug abuse and poor health status. The first section of this paper describes the two surveys, but principally the NHAMCS, since the NCVS is described extensively in Chapter 2. The second section presents the unadjusted estimates of the rate of emergency room visits due to violent crime from the two surveys. The third section outlines a series of potential explanations for the observed rate differences and the last section includes a series of adjustments to the rates designed to test the plausibility of the various explanations.