Finally, the overall decline in AW use was only partially offset by substitution of
the post-ban legalized models. Even if the post-ban models are counted as AWs, the
share of crime guns that were AWs still fell 24% to 60% across most jurisdictions. The
exception was Milwaukee where recoveries of a few post-ban models negated the drop in
banned models in a small sample of guns recovered during murder investigations.
58
6.4. Summary
Consistent with predictions derived from the analysis of market indicators in
Chapter 5, analyses of national ATF gun tracing data and local databases on guns
recovered by police in several localities have been largely consistent in showing that
criminal use of AWs, while accounting for no more than 6% of gun crimes even before
the ban, declined after 1994, independently of trends in gun crime. In various places and
times from the late 1990s through 2003, AWs typically fell by one-third or more as a
share of guns used in crime.
59, 60
Some of the most recent, post-2000 data suggest
restrictions that predated the AW ban. It is not yet clear that there has been a decline in the most common
ARs prohibited exclusively by the 1994 ban.
58
This was not true when focusing on just those guns that were used in the incident as opposed to all guns
recovered during the investigations. However, the samples of AWs identified as murder weapons were too
small for valid statistical tests of pre-post changes.
59
These findings are also supported by prior research in which we found that reported thefts of AWs
declined 7% in absolute terms and 14% as a fraction of stolen guns in the early period following the ban
(i.e., late 1994 through early 1996) (Koper and Roth, 2002a, p. 21). We conducted that analysis to account
for the possibility that an increase in thefts of AWs might have offset the effect of rising AW prices on the
availability of AWs to criminals. Because crimes with AWs appear to have declined after the ban, the theft
analysis is not as central to the arguments in this paper.
60
National surveys of state prisoners conducted by the federal Bureau of Justice Statistics show an
increase from 1991 to 1997 in the percentage of prisoners who reported having used an AW (Beck et al.,
1993; Harlow, 2001). The 1991 survey (discussed in Chapter 3) found that 2% of violent gun offenders
had carried or used an AW in the offense for which they were sentenced (calculated from Beck et al. 1993,
pp. 18,33). The comparable figure from the 1997 survey was nearly 7% (Harlow, 2001, pp.3, 7).
Although these figures appear contrary to the patterns shown by gun recovery data, there are
ambiguities in the survey findings that warrant caution in such an interpretation. First, the definition of an
AW (and most likely the respondents’ interpretation of this term) was broader in the 1997 survey. For the
1991 survey, respondents were asked about prior ownership and use of a “…military-type weapon, such as
an Uzi, AK-47, AR-15, or M-16” (Beck et al., 1993, p. 18), all of which are ARs or have AR variations.
The 1997 survey project defined AWs to “…include the Uzi, TEC-9, and the MAC-10 for handguns, the
AR-15 and AK-47 for rifles, and the ‘Street Sweeper’ for shotguns” (Harlow, 2001, p. 2). (Survey
codebooks available from the Inter-University Consortium for Political and Social Research also show that
the 1997 survey provided more detail and elaboration about AWs and their features than did the 1991
survey, including separate definitions of APs, ARs, and assault shotguns.)
A second consideration is that many of the respondents in the 1997 survey were probably
reporting criminal activity pr
ior to or just around the
time of the ban. Violent offenders participating in the survey, for example, had been incarcerated nearly six years on average at the time they were interviewed
(Bureau of Justice Statistics, 2000, p. 55). Consequently, the increase in reported AW use may reflect an
upward trend in the use of AWs from the 1980s through the early to mid 1990s, as well as a growing
recognition of these weapons (and a greater tendency to report owning or using them) stemming from
publicity about the AW issue during the early 1990s.
Finally, we might view the 1997 estimate skeptically because it is somewhat higher than that from
most other sources. Nevertheless, it is within the range of estimates discussed earlier and could reflect a
51
This document is a research report submitted to the U.S.
Department of Justice. This report has not been published by
the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official