The Rise and Decline of Homicide and Why

P1: FQP
March 31, 2000 10:55 Annual Reviews CHAP-21
Annu. Rev. Public Health. 2000. 21:505–41

Copyright
c 2000 by Annual Reviews. All rights reserved
THE RISE AND DECLINE OF HOMICIDE—

AND WHY
?
Alfred Blumstein1,4, Frederick P. Rivara2,4, and
Access provided by 2804:14d:1289:9125:c4aa:47fd:4ac5:282c on 06/06/19. For personal use only.
Richard Rosenfeld3,4
Annu. Rev. Public Health 2000.21:505-541. Downloaded from www.annualreviews.org
1Carnegie Mellon University, Pittsburgh, Pennsylvania

2University of Washington, Seattle, Washington
3University of Missouri-St. Louis, St. Louis, Missouri
4National Consortium on Violence Research
Key Words violence, firearms, drugs, cocaine, youth, crime

■ Abstract A dramatic rise in homicide in the latter half of the 1980s peaked during
the 1990s and then declined at an equally dramatic rate. Such trends in homicide rates
can be understood only by examining rates in specific age, sex, and racial groups. The
increase primarily involved young males, especially black males, occurred first in the
big cities, and was related to the sudden appearance of crack cocaine in the drug markets
of the big cities around 1985. This development led to an increased need for and use of
guns and was accompanied by a general diffusion of guns into the larger community.
The decline in homicide since the early 1990s has been caused by changes in the drug
markets, police response to gun carrying by young males, especially those under 18
years old, the economic expansion, and efforts to decrease general access to guns, as
well as an increase in the prison population and a continued decline in homicide among
those over age 24. The lessons learned from the recent homicide trends and the factors
associated with them have important implications for public health and the criminal
justice system.
INTRODUCTION
Injuries have long been viewed as a public health problem. Injury research as a
discipline is based on the public health model of surveillance: problem identifica-
tion, understanding of risk and protective factors, development of interventions,
and evaluation of their effectiveness. Injury research has traditionally relied on
the basic science of epidemiology to elucidate the etiology of injury and evaluate
the effectiveness of interventions. A 1985 report from the Institute of Medicine
promoted the premier public health agency in the United States, the Centers for
Disease Control and Prevention, as the lead agency for a national injury research
and control agenda (19).
505
P1: FQP
506 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD
Soon after the publication of this report, the nation began to experience a dra-
matic rise in violent crime and its resultant injuries and deaths. Crime has tradition-
ally been considered the bailiwick of the criminal justice system, backed primarily
by the research expertise of the disciplines of criminology and sociology. The
criminal justice system typically focused on the perpetrators of crime. However,
the sheer magnitude of the rise in violent injuries and deaths in the 1980s caused
violence to be viewed as a public health problem as well. Homicide became the
third leading cause of trauma death and the leading cause among 18- to 14-year-old
?
African-American males. It was projected that, without a change in the trends,
firearms would overtake motor vehicles as the leading cause of injury and death
in the United States by the year 2000 (17).

The reasons for the rise in homicide in the latter half of the 1980s, peaking in
the early 1990s, and the subsequent equally dramatic decline span both criminol-
ogy and public health (10). In this review, we first describe trends in homicide
victimization and offenses over the last 20 years and then explore the reasons for
the increase in the late 1980s and the subsequent decrease in the 1990s. We limit
our attention to homicide because more detailed and reliable data are available
for this offense than for other violent crimes and because, as a cause of death,
homicide is more in the purview of public health than crimes such as robbery.
Nonetheless, with few exceptions, the patterns we observe for homicide also char-
acterize robbery and serious assault (11). We hope that the lessons learned from
this review will lead to an integration of public health and criminal justice in-
terventions that further decreases violent injuries and death in the United States
(49).
TRENDS IN HOMICIDE
Despite many indications of a growing fear and concern over the possibility of
victimization, the trend in the U.S. homicide rate for the past 25 years has been
impressively flat. From the early 1970s until 1995, the rate oscillated between 8
and 10 homicides per 100,000 population (see Figure 1). These rates are more
than double those that prevailed in the 1950s and early 1960s. The homicide rate
reached a peak of 9.8/100,000 in 1991 and has been in a steady decline since then;
the more recent declines are approaching rates of <6.5/100,000, which have not
been seen since the late 1960s. [Data are from the Federal Bureau of Investigation
(FBI) Uniform Crime Reports (UCR) (27).]
Although these aggregate rates and their trends are certainly encouraging, it is
important to disaggregate them to be able to identify how each population group
has been involved. In particular, we find that the trends are very different for
younger compared to older victims and offenders (victims and offenders are very
similar in terms of age and other demographic features), and those differences have
stimulated the search for influences on these different age groups. In pursuing
these issues, we first examine trends in victimization, followed by examination of
offender patterns as reflected in arrests.
P1: FQP
THE RISE AND DECLINE OF HOMICIDE 507
?
Figure 1 U.S. homicide victimization rates reported by the Federal Bureau of Investigation’s
Uniform Crime Reports (UCR), 1972–1997 (27).
Victimization Trends
Age Factors Sharp age differences characterize the trends in homicide victim-
ization since 1980. All of the increase in U.S. homicide rates that began in the
mid-1980s resulted from large upswings in the victimization rates of those <25
years old. The victimization trends for younger children <12 years old and for
adults ≥25 years old have been flat or have shown modest declines for nearly
20 years. We document these victimization trends with data from the FBI’s Sup-
plementary Homicide Reports (SHR). Those reports, filed by individual police
departments, provide considerable detail on individual homicide incidents. Each
report contains information on victim and (when known) offender characteristics
and their relationship, the weapon involved, and the circumstances leading up to
the homicide. Unfortunately, only a single circumstance may be designated, and
so trends in the ways police designate these single circumstances may limit the
reliability of that aspect. (The number of incidents reported by an agency to the
FBI’s SHR is close but often not identical to those reported via the FBI’s UCR.
The variations can result from differences in reporting procedures when different
segments of the police department handle the two reporting tasks.)
Figure 2 depicts trends from 1980 through 1995 in the numbers of homicide
victims per 100,000 persons under age 12 (∗’s), between ages 12 and 17 (open
P1: FQP
?
Figure 2 U.S. homicide victimization rates for five age groups, from the Federal Bureau of
Investigation’s Supplementary Homicide Reports, 1980–1995.
boxes), between ages 18 and 24 (closed boxes), between ages 25 and 49 (+’s), and
at or above age 50 (x’s) (changes for more recent years continue the trends shown
in the figure). The SHR data are from Snyder & Finnegan (54). The population
data are from the U.S. Bureau of the Census (12). After decreasing somewhat
during the first half of the 1980s, the victimization rates for juveniles ages 12–17
and young adults ages 18–24 began a steep increase, which lasted until the early
1990s. The juvenile rate roughly doubled, and the rate for young adults increased
by two-thirds over that period. By 1993, the homicide rate for young adults was
2.5-fold that for the population as a whole (cf. Figures 1 and 2). Since the early
1990s, the victimization rates for both juveniles and young adults have declined,
although not to the levels of the early 1980s (29). Meanwhile, we observe no
comparable rise in the rates for children under 12 and adults over 25 (Figure 2).
Whatever factors are responsible for the increase in the nation’s homicide rate
in the late 1980s, then, clearly had a disproportionate influence on teenagers and
young adults. Older adults and children <12 were either shielded from those
influences, or other factors were operating in those groups that diminished their
risk. A similar story can be told about racial differences in homicide victimization
rates and trends.
Racial Factors African Americans are exposed to a risk of homicide many times
that of whites. The African-American victimization rate was 32/100,000 popu-
lation in 1995, compared with 5/100,000 among whites. The trends in homicide
P1: FQP
?
Figure 3 U.S. homicide victimization rates by race from the Federal Bureau of Investigation’s
Supplementary Homicide Reports, 1980–1995. Note that the white rate (open boxes) has been
multiplied by 6 for comparison with the black rate ( filled boxes).
victimization also show some notable differences by race. Figure 3 presents the
victimization trends for whites and African Americans between 1980 and 1995.
People of other races have been excluded from this comparison. The SHR recorded
579 homicide victims among persons of other races in 1995 (<3% of the total of
21,597 victims). Hispanics are not separately identified in this analysis and may
be of any race. To analyze trends, the large difference in homicide rates between
whites and African Americans was made more comparable in the graph by first
multiplying the white rate by six.
Although we observe a rough correspondence between the race-specific trends,
with homicide rates for both blacks and whites declining during the first half of the
1980s and rising to a peak in the early 1990s, the rise for blacks began earlier and
was considerably sharper than that for whites, as shown in Figure 3. The black rate
increased by 44% from a trough of 27/100,000 in 1984 to a peak of 39/100,000
in 1991. The rate for whites did not begin to rise until 1989 and increased only
12%, to a peak of 5.6/100,000 in 1991. The rate of decline in victimization that
was observed in the 1990s was correspondingly sharper for African Americans. It
appears that the factors responsible for the trends in homicide victimization rates in
recent decades have exerted a stronger influence on African Americans than whites.
Gender Factors Without displaying the trends, we observe that sex differences
in the aggregate rates of homicide victimization have not changed appreciably
P1: FQP
since 1980. With some year-to-year fluctuation, women have composed about
one-quarter of all homicide victims over the last 2 decades in the United States.
However, an important interaction between the age, race, and sex of victims char-
acterized the period of increased homicide rates during the late 1980s and early
1990s; the increased levels of risk were concentrated disproportionately among
young black men (28, 29) and were strongly associated with firearms.
?
Weapons Factors The role of weapons must figure centrally in any credible
explanation of U.S. homicide trends over the past 2 decades. All of the increase
in the aggregate rate of victimization occurred in the category of homicides by

firearms. The number of persons killed with other weapons or no weapons has
actually dropped somewhat since 1980. The number of whites killed with firearms
rose by 25% from a trough of 5875 in 1988 to a peak in 1993 of 7315 (see Figure 4).
The number of African-American firearm homicide victims, meanwhile, nearly
doubled from a trough of 4786 in 1984 to a 1993 peak of 9394 (see Figure 5).
Although these common increases do not reveal the precise role of guns in
the so-called homicide epidemic of the late 1980s, they do suggest that factors
affecting access to and carrying of guns very likely contributed to the increase
in risk. We explore these factors in some detail in the following discussion of
offending patterns.
Figure 4 Trends in weapons [firearms (closed boxes) and weapons other than firearms (open
boxes)] involved in homicides of white victims. (Source: Supplementary Homicide Reports data.)
P1: FQP
?
Figure 5 Trends in weapons [firearms (closed boxes) and weapons other than firearms (open
boxes)] involved in the homicides of black victims. (Source: Supplementary Homicide Reports
data.)
Relationship Factors A final component of homicide with important implica-

tions for our assessment of the factors responsible for recent trends in risk is the
relationship between the victim and suspected offender. Because the SHR data
do not contain information about the legal disposition of homicide cases, the “of-
fenders” in the cases are technically “suspects.” The same is true of the arrest data
used to examine offending patterns below. The great majority of persons arrested
for murder are eventually convicted in court (13).
Figure 6 displays the percentages of victims killed by family members, other
acquaintances, and strangers between 1980 and 1995. The figure also shows the
trend for incidents in which the relationship between the victim and offender was
unknown. The percentage of people killed by family members decreased steadily
over this period, from ∼17% of all homicides in the early 1980s to 11% by the
mid-1990s. Most of this decrease is attributable to a sizeable decline in “intimate-
partner” homicides involving spouses and ex-spouses. (We explore this trend in
more detail in the following discussion of offending patterns.) Between 35% and
40% of homicide victims are killed by nonfamily acquaintances, and another 15%
are killed by strangers.
In the remaining incidents, the victim’s relationship to the offender is “un-
known.” The fraction of such incidents increased from about one-quarter of all
P1: FQP
?
Figure 6 Trends in the percent distribution of homicide victim-offender relationships. Closed

box, family member; open box, acquaintance; *, stranger; +, unknown. (Source: Supplementary
Homicide Reports data.)
cases in the mid-1980s to ∼40% in 1995, generating speculation about the in-
creasingly random character of murder in the United States. The FBI lent official
credence to such concerns by announcing in its 1993 UCR: “Every American now
has a realistic chance of murder victimization in view of the random nature the
crime has assumed” (27:287).
The FBI’s striking characterization of the upturn in homicides in which the
relationship between the victim and offender is “unknown” implies that they are
strangers and not simply unknown to the police. Although it is difficult to determine
precisely whether the fraction of homicides involving victims and offenders who
are strangers has risen over the past 20 years, two facts are reasonably certain. First,
the SHR under counts homicides by strangers, which probably account for nearly
a quarter of all homicides rather than the 15% recorded in the SHR during the
1980s (58). Second, it is incorrect to assume that all of the incidents in which the
relationship between the victim and offender is unknown involve strangers. Many
of them very likely are acquaintance cases, such as drug-related killings, in which
the police are unable to secure cooperation from witnesses to make an arrest (47).
Certainly, from what we have observed about the nature of criminal homicide in
the United States, a statement that suggests a uniform risk of victimization across
sex, race, and age groups seems far-fetched at the very least.
P1: FQP
Offender Trends
One of the striking features of the rise in homicide victimization between 1985 and
the early 1990s was the bifurcation in trends by age, with a sharp rise for teenagers
and young adults at the same time that there was a decline for those aged 25 and
above. We observe the same difference in the age-specific trends for homicide
offenses. We examine offender patterns by examining arrest rates based on arrest
data from the FBI’s annual UCRs (27).
?
Figure 7a presents the age-specific arrest rate (the ratio of arrests of any age
to the base population of that age, known as the “age-crime curve”) for murder in
1985, which was the last year of a 15-year period of very stable age-specific rates,
and in 1993, which was the peak year of juvenile age-specific rates. The pattern in
1985 was fairly typical, with a rather flat peak for offenders between the ages of 18
and 24 and steadily lower rates for offenders under age 18 and over age 24. Figure
7a illustrates that there was a dramatic change by 1993, with a very sharp increase
at the younger ages and a very sharp peak at age 18. We also note that the rates
for those over 30 had indeed declined. Figure 7b depicts the same 1993 age-crime
curve along with its counterpart for 1997. Here we see that the rates for all ages
declined from 1993 to 1997 and that the steepest decline occurred around age 18.
Growth in offender rates between 1985 and 1993 reached its highest level—∼60
homicide arrests per 100,000—among 18 year olds.
It is instructive to look at the trends for individual ages. Figure 8a depicts the
trend for the traditional peak homicide arrest rate, ages 18 through 24. Those rates
were similar from 1970 through 1985 and then diverged beginning in about 1986.
The rate for the 18 year olds more than doubled by 1991 (for an annual growth
rate of 16% during this period), dropped in 1992, reached a new peak in 1993,
and then continued down for the next 3 years. The pattern is similar for the other
ages depicted in Figure 8a, although the steepness of the rise in the late 1980s
decreases with increasing age, and the decline after 1993 is correspondingly less
for the older ages.
For offenders 18 and under, depicted in Figure 8b, the pattern is very similar to
that at age 18, although the stable base rate in the 1970–1985 period was lower.
In all of the under-18 groups, the rate more than doubled by 1993. The pattern
for ages above 24 was similarly flat through the mid-1980s, followed by a steady
decline for most of these age groups (data not shown).
These changes for the growth period 1985–1993 and for the decline period
1993–1996 are reflected in Figure 9, which depicts for each age the ratio of the
age-specific arrest rate for murder to the rates that prevailed in 1985. Points above
the heavy line (at the ratio of 1) represent an increase in the rates, and points below
that line represent a decrease. The upper graph portrays the ratio reached in the
peak year 1993, and the lower graph portrays the degree to which the ratio had
declined by 1997.
The arrest rate for 15 year olds in 1993 was triple the rate for 1985. The growth
until 1993 then declined with age, but it was more than double the 1985 rate for
P1: FQP
?
Figure 7 (a) Age-specific arrest rate for homicide, 1985 and 1993, the peak year. (b) Age-
specific arrest rate for homicide, 1993 and 1997. (Sources: Uniform Crime Reports data and
census population data.)
P1: FQP
?
Figure 8 (a) Trends in age-specific arrest rate for homicide for traditional peak ages of 18, 20,
22, and 24. (b) Trends in age-specific arrest rate for homicide at young ages of 18, 17, 16, 15, and
13–14. (Sources: Uniform Crime Reports data and census population data.)
P1: FQP
?
Figure 9 Ratio of age-specific arrest rates for homicide in 1993 and 1997, compared with 1985.
(Sources: Uniform Crime Reports data and census population data.)
all ages of 20 and below. In contrast, for ages 30 and above, the 1993 rates were
actually ∼20% lower than the 1985 rates.
The graph of the 1997-to-1985 ratio is clearly below that for 1993, and the
greatest decline occurred in the teenage years. It is clear that the teenage rates are
still ∼50% above the 1985 rates that had prevailed since 1970, and so there is still
considerable room for improvement to get back down to the 1985 rates.
Also, we note the continuing decline in the homicide rates for older ages. By
1997, the 25- to 30-year-old group had declined ∼20% from the 1985 rates, and
the rates among older groups had declined by ∼40%.
These figures underscore the central importance of examining the factors asso-
ciated with the different age groups to explain the trends in the aggregate homicide
rate since 1985. The aggregate rate shown in Figure 1 grew to the 1991 peak solely
because the rates for people <25 years old were increasing faster than the rates for
older people were declining. Between 1991 and 1993, the rates for younger peo-
ple were generally flat (as reflected in the pattern for 18 year olds in Figure 8a),
and so the decline by those in the older age groups dominated the aggregate,
leading to the downturn that began in 1992. Because the rates for both young
and old age groups were decreasing after 1993, the aggregate rate continued to
fall.
P1: FQP
In summary, all of the increase in the aggregate level of homicide in the United
States during the growth period of the late 1980s and early 1990s can be traced to
trends in the younger age groups <25 years, because homicide rates for those ≥25
years old did not increase. However, the aggregate decrease since 1993 is a product
of both the recent sharp drop in offending among young people and the continuing
decline in offending among older persons. Even though they commit homicides
at much lower rates, the contribution of people ≥25 years old to the recent decline
in the aggregate homicide rate may be appreciable, given their large numbers.
?
An important contributor to the long-term drop in adult homicide is the dec-
line in homicides involving spouses, ex-spouses, and other intimate partners. There
are notable sex and race differences in that decline. The rate at which husbands are
killed by their wives has decreased more sharply than the rate at which wives are
killed by their husbands (24, 33), and the decrease in rates for blacks has been
greater than that for whites.
Figure 10 displays the sex- and race-specific trends in intimate-partner homicide
rates from 1976 to 1996 for adults aged 20 to 44. The values for whites have been
scaled up by a factor of 10 to align them with the black rates. Even though the
rates for all of the age groups have fallen over the past 20 years, the reduction in
Figure 10 Trends in the rate of intimate-partner homicide victimization by race and gender,
1976–1996. (Sources: Uniform Crime Reports data and census population data.) (Note that the
white rates have been multiplied by 10 for comparison with the black rates.)
P1: FQP
the rate at which black women kill their partners is particularly striking. The black
male victimization rate fell by >80%, to 2.8/100,000 in 1996 from 16.5/100,000
in 1976. By contrast, the rate for white females exhibited a more modest de-
crease of 24%, from 1.7/100,000 to 1.3/100,000 (the declines for black females
and white males are 63% and 60%, respectively). By any reckoning, these adult
mortality trends, especially for black males, must be seen as good news, but the
social changes responsible for them remain poorly understood. We offer some
speculation below about these changes in our discussion of the factors associated
?
with the rise and fall in U.S. homicide rates.
FACTORS CONTRIBUTING TO THE INCREASE

IN THE HOMICIDE RATE
Changing Demographic Composition

Much of the speculation about the recent decline in homicide rates attributes the
decline to changing demographics. For example, writing in the New York Times,
David Kocieniewski (41) states that “. . . some (unnamed) criminologists attribute
the decline to demographic factors like a smaller number of teenagers.” In an
article on the “mystery” of the drop in crime, David Anderson (4) notes that some
(also unnamed) analysts explain the drop as resulting from “random demographic
changes” (p. 49). This opinion may be a holdover from the realization that much of
the decline that began in 1980 was attributable to a demographic shift, as the baby
boom generation matured out of the high-crime ages. But those same demographic
effects are no longer at work in the early 1990s, because demographic effects do
not always have to work in the same direction.
The decline after 1980 was significantly affected by the shrinking size of the
cohorts in the high-crime ages, but the United States in the 1990s is in a period of
growing cohort sizes in the late-teens and early-20s age groups. Figure 11 depicts
the age distribution of the U.S. population in 1999. It is evident that the smallest
age cohort >40 is ∼23, the cohort born in 1976. Each of the younger cohorts is
larger than its predecessor until the peak at about age 7. Thus, if teenager-specific
crime rates were to remain constant, then the aggregate crime rate would increase
as a result of the larger cohort sizes. This possibility spurred the warnings of a
demographic “crime bomb” set to go off during the 1990s (23).
Yet, it is important to recognize that these age composition changes are relatively
small, with cohort sizes growing at ∼1%/year. In the face of much larger swings
in the age-specific crime rates—∼10%–20% increases in the 1980s (16% for 18
year olds from 1985 to 1991), as well as decreases in the 1990s—the 1% change
in demographic composition is a minor effect.
It is possible, finally, that changes in relative cohort size could alter the age-
specific rates through mechanisms described by Easterlin (25) and Smith (53).
However, the evidence suggests that, if changes in the relative size of age cohorts
P1: FQP
?
Figure 11 Age composition of the U.S. population in 1999 in number of persons of each age.
(Source: Census population data smoothed by 3-point smoothing.)
influence homicide rates, the cohort effects are minor compared with age and
period effects (44, 45, 55).
The Changing Role of Handguns, Especially Among Juveniles

and Young Adults <25
There is widespread recognition of the changing role of weaponry in young peo-
ple’s hands (34). Over the last decade the weapons involved in settling juveniles’
disputes have changed dramatically from fists or knives to handguns, with their
much greater lethality. That growth in lethal weaponry is reflected in the changes
in the weapons involved in homicides in different race and age groups. Blumstein
& Cork (9) examined those changes in mortality statistics and found a striking
divergence in the trends for gun and nongun homicides, with some significant
differences based on age and race.
As seen in Figures 4 and 5, African-American homicide victimization rates
with guns declined to a trough in 1984, followed by a rather sharp increase into
the early 1990s. The rate of increase is largest for the youngest age group. The
situation for whites is very similar (9), except that the trough occurred somewhat
P1: FQP
later (about 1988); the growth rate after the trough was strongest in the youngest
group, with no growth in ages >25. In stark contrast, there were no comparable
turning points in the nongun homicide victimization rates, which have been flat or
have shown modest declines since 1980.
The pattern in gun suicide rates was very similar to that for homicide rates for
African Americans—a decline through the early 1980s and an increase that varied
inversely with age after a turning point in about 1985. There was no such consistent
pattern for whites. Again, there was no meaningful change in nongun suicides over
?
this period. The consistency between the suicide and homicide patterns, at least
for young African Americans who normally have a lower suicide rate than do
whites, certainly suggests an important role of the increased availability of guns.

The availability of guns after 1985 converted attempts that might not have been
fatal into fatalities.
We can examine similar issues regarding the weaponry used by offenders in
homicides, with data from the SHR (54). Figures 12a, b, and c illustrate the trends
in weaponry used in homicides by offenders in three age categories: older adults,
25–45 years old (Figure 12a); young adults, 18–24 (Figure 12b); and children
and adolescents, 17 and under (Figure 12c). The weapons are classified into three
Figure 12 (a) Trend in types of weapons used by adults (ages 25–45) in homicides. (b) Trends
in types of weapons used by youths (ages 18–24) in homicides. (c) Trends in types of weapons
used by children and adolescents (under 18) in homicides. (Source: Supplementary Homicide
Reports data.)
P1: FQP
March 31, 2000
10:55
Figure 12 (Continued)
Annual Reviews
CHAP-21
?
THE RISE AND DECLINE OF HOMICIDE
521
P1: FQP
groups: handguns, other guns, and nonguns (which includes no weapon). We can
see that, from 1977 through 1995, little meaningful change occurred in the use
of handguns by older adults. The situation for young adults and for children and
adolscents is quite different, however. For both of these groups, there was no clear
trend until 1986, and then a significant growth in handgun use began. With 1985 as
the base year, handgun homicide among young adults increased >100% by 1994,
and use of handguns by those <18 years old increased >300%. In both of these
groups, there is a leveling out from 1993 to 1994, and then we see a sharp decline
?
in 1995 in these groups as well as in older adults. There should be similarly sharp
declines in 1996 and 1997, consistent with the decline in homicide arrest rates
shown in Figure 9a. Even the use of handguns in homicides by older adults shows
a sharp decline in 1995.
In all of these figures, no appreciable increase has occurred in either the long-
gun or the nongun categories. There has been some decline in the nongun category
for young adults, but this decline (a drop of 28% from 1985 to 1994) is small
compared with the >100% growth in their use of handguns. Thus, we observe
that the growth in homicides by juveniles and young adults, which accounted for
all of the growth in homicides in the post-1985 period, was driven totally by the
growth in homicides committed with handguns. Clearly, the sharply increasing
presence of handguns in young-adult and juvenile homicides must be considered
of fundamental importance in any explanation of the homicide increase of the late
1980s and early 1990s.
We also observe some important racial differences in the growth of handgun
homicides, with the dominant growth being among African-American juveniles
and young adults, both as offenders and as victims. Figure 13 presents the number
of homicides committed by blacks, ages 18–24, by type of weapon involved. Here
we see an even sharper growth in handgun use than for young adults generally
(Figure 12b); the number of handgun homicides by young black adults more than
tripled from their low in 1984 to their peak in 1993. There was no comparable
growth in the role of the other weapon types.
Although some growth also occurred in handgun homicides by white young
adults 18–24 years old, that growth was far less than among black youths. The
difference is depicted in Figure 14, which compares the two racial groups. This
figure focuses on all cities of >100,000 population. Here we see the strong growth
in handgun prevalence for black young adults 18–24 years old, from a low in 1984
to a tripling by 1993. The rise for whites does not start until 1989, but does display
a doubling by the 1993 peak. Finally, the post-1993 decline is much sharper for
black young adults than white young adults. These patterns parallel closely those
observed for the victimization trends observed earlier.
How Did the Young People Get the Guns:

Links to Illicit-Drug Markets
It is widely recognized that an important feature of the late 1980s was that drugs,
and especially crack cocaine, became an important part of life in many American
P1: FQP
?
Figure 13 Trends in types of weapons used by black youths (ages 18–24) in homicides. (Source:
Supplementary Homicide Reports data.)
inner cities. This trend is reflected in the drug arrest rates displayed in Figure 15a
(for adults aged 18 and older) and Figure 15b for juveniles (under 18). For non-
white adults, there was a rise that began in the early 1980s and accelerated after
1985. For juveniles of both racial groups, there was a clear upward trend from a
negligible level of arrests in 1965 to a peak in 1974, representing predominantly
arrests for marijuana. The peak in 1974 was very likely a result of widespread
decriminalization of marijuana, which led to a decline in drug arrests for both
whites and nonwhites. For whites, that decline continued until the early 1990s,
when there was a new rise in marijuana use and arrests. For nonwhite juveniles,
the decline was somewhat slower, leveling off in the early 1980s (while the non-
white adult rate increased) and then increasing very sharply, more than doubling
between 1984 and 1989.
We have thus identified three major changes that have occurred in the short
period between 1985 and 1993: (a) homicide arrest rates for youth ≤20 more
than doubled, whereas there was a decline in homicide rates for adults ≥30
(Figure 9); (b) the number of homicides with guns committed by juveniles more
than quadrupled, whereas there was no change in nongun homicides (Figure 12c);
and (c) the arrest rate of nonwhite juveniles on drug charges more than dou-
bled, whereas there has been no growth in the drug arrest rate for white juveniles
(Figure 15b).
P1: FQP
?
Figure 14 Trend in number of handgun homicides in cities with populations over 100,000,
by race of offenders ages 18–24 (black and combined white and Hispanic). (Source: Sup-
plementary Homicide Reports data.)
The important feature of these rapid changes is that they occurred after a prior
15-year period of relative stability, and so we need some explanation of what
changes brought about those sharp transitions from a stable pre-1985 period to a
rapidly deteriorating post-1985 period.
One possible explanation (6) for this array of changes derives from examining
the changes in the illegal-drug markets associated with the introduction of crack
cocaine. That introduction occurred at different times in different parts of the
country, but the leading edge occurred in about 1985, at least in larger cities like
New York and Los Angeles. An important feature of crack is its low price, and
that brought into the cocaine market many low-income people who could only buy
it one “hit” at a time. The number of transactions in those markets significantly
increased, owing both to the numbers of new buyers brought in and the number of
transactions each buyer engaged in per week.1
1 Contrast this with the more typical middle-class consumers of powder cocaine. Those
buyers had an interest in minimizing the number of illegal transactions they engaged in, and
they had the assets to buy a reasonable supply in any transaction and the storage space in
which they could keep their drugs safely. (We thank Jonathan Caulkins for calling attention
to this feature of the emerging crack markets.)
P1: FQP
?
Figure 15 Trends in arrest rates for drug offenses. (a) Trend in arrest rate of adults (ages 18
and above) for drug offenses by race. (b) Trend in arrest rate of juveniles (under age 18) for drug
offenses by race. (Sources: Uniform Crime Report and census population data.)
P1: FQP
To accommodate that increased demand, the drug dealers had to recruit a large
numbers of new sellers. Juveniles were the natural source of supply for that labor
market. They were probably willing to work more cheaply than adults, partly
because they are less vulnerable to the punishments imposed by the adult criminal
justice system. But also they tend to be daring and willing to take risks that more
mature adults would eschew. The economic plight of young urban black juveniles
at the time, many of whom saw no other comparably satisfactory route to economic
success or even sustenance, made them particularly amenable to the lure of the
?
drug markets.
These juveniles, like many other participants in the illicit-drug industry, are
likely to carry guns for self-protection, largely because that industry uses guns
as an important instrument for dispute resolution. Also, the participants in the
industry are likely to be carrying a considerable amount of valuable product—
drugs or money derived from selling drugs—and are not likely to be able to call
on the police if someone tries to rob them. Thus, they are forced to provide their
own defense, and a gun is a natural instrument for use in that function (38).
Because the drug markets are pervasive in many inner-city neighborhoods and
the young people recruited into them are fairly tightly networked with other young
people in their neighborhoods, it has become easy for the guns to be “diffused” to
other teenagers who go to the same school or who walk the same streets. These
other young people are also likely to arm themselves, primarily for their own pro-
tection, but also because possession of a weapon may become a part of status seek-
ing in the community. This initiates an escalating process: as more guns appear
in the community, the incentive for any single individual to be armed increases.
Then, given this availability of guns, the recklessness and bravado that are
often characteristic of teenagers, as well as their lower levels of skill in settling
disputes other than by physical force, many fights that would otherwise have
resulted in nothing more than a bloody nose turn into shootings.2 This can be
exacerbated by the problems of socialization associated with high levels of poverty,
high rates of single-parent households, educational failures, and a widespread sense
of economic hopelessness. But those factors have been changing gradually over
the years, and so they cannot readily provide the explanation for the very sharp
changes that began to take place in the mid-1980s.
By the time people mature beyond their early 20s, it appears that they do
develop some prudence, are more cautious even if they are armed, and display
greater restraint in the use of guns. (That inference seems reasonable, given
evidence of the decline in homicide rates among people in their mid-20s and
2 Tarr has called our attention to a passage in Jane Addams’ The Spirit of Youth and City
Streets (1), in which some 13-year-old boys were taunting each other, and one of them went
into the house to get a gun and shot the taunter through the head and killed him. Addams
notes that “This tale could be duplicated almost every morning; what might be merely a
boyish scrap is turned into a tragedy because some boy has a revolver.” The original edition
was published in 1909.
P1: FQP
beyond.) Also, because adults in their mid-20s are not as tightly networked as
younger people, the gun diffusion process is likely to be much slower and less
widespread. Alternatively, we might be witnessing a cohort effect, and the 18 year
olds involved in the higher homicide rates may possibly continue their recklessness
as they mature. That issue still needs to be watched and explored.
The validity of these hypothesized processes can be tested with city-level data
on drugs, guns, and homicides. Such an analysis would take advantage of the
fact that drug markets flourished at different times in different cities, for example,
?
reportedly earlier in New York and Los Angeles but later in Washington and
Pittsburgh. Thus, the sharp change that took place in the national statistics in
about 1985 should have been displayed in the city-level data shortly after the crack
markets began functioning in these cities, with the change occurring earlier in
some cities than in others. But if the above hypothesized process is correct, that
would suggest a relationship between the rise in drug arrests and youth homicides
in the same city, with the rise in homicides presumably coming after a rise in drug
arrests.
Cork (22) has used a diffusion model to examine city-level patterns in arrests
of juveniles for drug offenses and for homicide. He finds that the crack arrests
rose sharply beginning in about 1984 in cities in the Northeast and the West and in
about 1987 in the center of the country. He also finds a similar geographic pattern
in the rapid rise in juvenile homicide arrests, but with a median lag of ∼2 years
after the rise in crack arrests. These observations are consistent with the diffusion
hypothesis.
There is some further evidence that supports the hypothesis of the diffusion
of guns from drug markets throughout the larger community of juveniles. Figures
16a and 16b present the arrest rates for murder among adults and juveniles by race.
We see from Figure 16a that, since 1980, both white and nonwhite adults have
followed the same negative trend, even though there has been a large difference in
their involvement in drug markets (see Figures 15a and b). In contrast, the arrest
rates for juveniles—whites and nonwhites—have grown markedly since 1985,
when the drug arrest rate for nonwhites began to climb. The murder arrest rate
for nonwhites increased by 123%, from 7.1/100,000 in 1985 to 15.8/100,000 in
1992. The rate for whites also increased markedly, but by a lesser amount—80%,
from 1.5/100,000 to 2.7/100,000. Also, the first rise in the nonwhite juvenile
rate began in 1985, whereas the first rise in the white juvenile rate did not occur
until 1988, a 3-year lag that is consistent with the diffusion hypothesis that the
major arrival of handguns began in the African-American community and then
took 3 years to diffuse to the white community. The fact that there seems to have
been no significant involvement of white juveniles in the drug markets during
this time (Figure 15b) has not insulated them from the growth in involvement in
homicide, and one possibility for that growth is through the hypothesized gun
diffusion process.
This hypothesis of the diffusion of guns into the community is the contagion
model, familiar to most workers in public health. As applied to firearms, the
P1: FQP
?
Figure 16 Trends in arrest rates for homicide. (a) Trend in arrest rate of adults (ages 18 and
above) for homicide by race. (b) Trend in arrest rate of juveniles (under age 18) for homicide by
race. (Sources: Uniform Crime Report and census population data.)
P1: FQP
presence of an armed group in the population increases the likelihood that others
will become armed to protect themselves (43). Surveys among teens lend some
credence to the contagion model. Many inner-city teenagers and young adults in
the late 1980s had guns or had easy access to guns. For example, 25% of junior
high school males in Washington, DC, reported carrying a gun (57). The 1993
Youth Risk Behavior Survey found that 14% of males in grades 9–12 reported
carrying a gun in the previous 30 days (18). In Seattle, 40% of male 11th graders
had easy access to a gun, and 11% owned guns (16). A survey in four inner-city
?
areas with high rates of gun violence found that 86% of incarcerated youths and
37% of male high school students owned guns (51).

Most adolescents feel less safe as their classmates acquire guns (35). Likewise,
living in a community with frequent shootings increases the likelihood of youths
carrying guns. Among those who have carried a gun in the past, 34% are more
likely to bring a gun to school if others do (35). Sheley & Wright (51) report that
two-thirds of the inner-city high school students they surveyed who carried guns
cited self-protection as the main reason for gun ownership.
ROLE OF THE BIG CITIES
The largest cities contribute disproportionately to patterns of serious violence for

the nation as a whole. The prominent role of the large cities is clearly evident in the
trends in homicide. Based on UCR data for 1991, for example, the United States
experienced 24,700 homicides. New York City alone provided 2,154 of them (9%
of the total). Because New York City’s homicide rate has declined faster than the
national rate, its percentage contribution to the total has dropped to <5%.
Although no other city has as large an effect as New York, the importance
of the large cities is reflected in the relative contribution they make to the total
homicide picture. In 1996, ten cities (New York, Chicago, Los Angeles, Detroit,
Philadelphia, Washington, New Orleans, Baltimore, Houston, and Dallas, in de-
creasing order of numbers of homicides) accounted for fully one-quarter of all of
the nation’s homicides. In contrast, in 1991, when New York alone accounted for
9% of all U.S. homicides, only seven cities (New York, Los Angeles, Chicago,
Detroit, Houston, Dallas, and Washington) were needed to account for a quarter
of U.S. homicides (based on city-specific data from UCR 1991 and 1996).
New York City has been a major contributor to the national decline since the
early 1990s. In the national net decline in homicides from 1993 to 1994 (a re-
duction of 1,200 homicides), New York City’s drop of 385 accounted for 32% of
that change. In the net change from 1994 to 1995 (a national net drop of 1,720
homicides), New York City’s drop of 384 accounted for 22% of the total decrease.
New York City’s contribution to the drop since 1995 has been closer to 10%, still
very large, but smaller than in the earlier years, in part because the smaller cities
are beginning to catch up. It is thus clear that what goes on in New York City or the
largest cities more generally can have a very powerful effect on national statistics.
P1: FQP
Examination of the trends over time offers a compelling picture of the saliency
of the large cities, both in the rise of homicide in the 1980s and the decline during
the 1990s. Figures 17a (for homicides with other than handguns) and 17b (for
homicides with handguns) use SHR data to estimate the number of homicides in
each of four groups of cities by population (those of ≥1 million and those in the
ranges of 500,000–1 million, 250,000–500,000, and 100,000–250,000).
The numbers of homicides associated with each of the city-sized groups other
than the largest are roughly the same in each year (partly because, as the popu-
?
lation size approximately halves between groups, the number of cities approxi-
mately doubles, thereby keeping the number of homicides roughly stable); there-
fore, we can contrast the larger and smaller cities. There were six cities in the
million-plus group: New York, Detroit, Philadelphia, Los Angeles, San Diego,
and Dallas (some cities, including Chicago and Houston, were not included here
because their reporting to the SHR was sporadic, and we did not want to at-
tribute these fluctuations in reporting to changes in the homicide patterns being
observed).
Figure 17a shows the limited variation associated with the nonhandgun homi-
cides. There was very little change in the smaller cities and a rather gradual decline
of 18% in the large cities, from a peak in 1986 through 1993, with a comparable
drop of 21% from 1993 to 1994. These changes were much smaller than those in
the handgun homicides. Figure 17b shows that the large cities had major growth
beginning in 1986, increasing 85% from 1985 to the flat 1991–1993 peak, and then
declining 37% to the low in 1995, with indications that that decline will continue,
at least for some years.
We note that the smaller cities also had a distinct upturn in handgun homicides,
but not until 1988, 2 years later than in the large cities. That upturn was even larger
in percentage terms, collectively increasing 116% from the trough in 1987 to the
peak in 1993. The more recent downturn also began later than in the large cities,
in 1994 in the 250,000- to 500,000-population cities and not until 1995 in the other
two groups. The drop from the collective 1993 peak was still only ∼16% in 1995.
The differences in the timing of the increase in handgun homicides for the large
and smaller cities could be associated with crack markets. Crack markets generally
emerged first in the largest cities and may have diffused to smaller cities at a later
time, which could possibly account for these lag effects.
It is also true that the peak occurred later in smaller cities. Although peak
homicide rates were reached in 1991 in the largest cities, the lags before peak rates
in smaller cities were progressively larger as the city sizes decreased. We also note
that the decline in the largest cities was quite sharp after the flat 1991–1993 peak.
In the smaller cities, however, a comparable sharp decline was not yet observed.
Indeed, recent news reports chronicle the escalating homicide rates in some mid-
sized cities and speculate that these increases could be associated with the later
emergence of crack markets and associated drug-related violence (39).
To the extent that both the increase and downturn in handgun homicides in the
largest cities were associated with corresponding changes in crack markets, we
P1: FQP
?
Figure 17 Trends in weapons used for homicides in cities with populations >100,000. (a) Trend
in homicides with weapons other than handguns. (b) Trend in homicide with handguns. (Source:
Supplementary Homicide Reports data.)
P1: FQP
would anticipate that homicide rates in the smaller cities should be reaching a peak
and in some instances beginning to decline. The timing of these changes should
correspond to city size; the declines should begin first in larger cities and later in
smaller cities. SHR data beyond 1995 will be needed to confirm these speculations
about the role of drug markets in the increase and the decline of U.S. homicide rates
and in the differences in timing of these changes observed across cities of different
sizes. However, the observed patterns are highly consistent with explanations of
homicide trends that assign central importance to the rise and decline of crack in
?
the United States.
SOME OBSERVATIONS ABOUT CAUSES

OF THE DECLINE
The sharp rise in violence by young people during the late 1980s and the corre-
spondingly sharp decline in the 1990s are striking. The increase in the aggregate
homicide rate was caused by escalating rates among juveniles and young adults,
predominantly (although not exclusively) by and against black males, particularly
in the larger cities and predominantly involving handguns. Until 1997, the de-
cline among those <20 years old was still less than halfway to the stable rate that
prevailed for the 15 years from 1970 through 1985, and so we are not necessarily
at the end of the downturn; there is some reason to expect that the decline will
continue, at least until it reaches the 1985 level and perhaps even beyond.
One aspect of that continued decline in homicide rates could be that smaller
cities lag behind larger cities by 1–3 years. Subsequent data from the smaller cities
will become available over the next few years to test this speculation. If this specu-
lation is in fact confirmed, that opens the questions, first, of what forces are driving
this process, both up and down, and, second, what factors are contributing to the
lag between the larger and smaller cities. The evidence available so far, although
short of providing unambiguous confirmation, is largely consistent with the earlier
hypothesis of the sequence that created the rise phase: introduction of crack in
the mid-1980s; recruitment of young minority males to sell the drugs; arming of
the drug sellers with handguns; diffusion of guns to peers; and irresponsible and
excessively casual use of guns by young people, leading to a “contagious” growth
in homicide (6, 43) and also robbery (11).
The subsequent decline is undoubtedly caused by a much more complex pro-
cess. Some of the drop is attributable to reactions, primarily by police and by
other groups in the community, to the factors that contributed to the rise—most
importantly focused on efforts to remove guns from the possession of young peo-
ple who were carrying them illegally. There are also various independent circum-
stances that have contributed to the great reduction in violence by young people,
particularly changes in crack markets; the strength of the economy, providing op-
portunities in the legitimate labor market; and general efforts to control guns that
also contribute to reduced access to handguns. The steady declines among older
P1: FQP
populations undoubtedly were assisted by some of these same factors, but also had
some of their own distinctive aspects, most particularly reductions in intimate-
partner violence and increased incarceration. These aspects are addressed in the
next sections.
Reactive Response to Factors Contributing to the Rise

in Violence Among Juveniles and Young Adults
?
Police Response to Gun Carrying by Juveniles Notable among the reactive
forces are police efforts to remove guns from juveniles. These tactics include a
mixture of aggressive stop-and-frisk methods, especially in high-violence neigh-

borhoods (reputed to have been an important part of the response in New York
City), introduction of programs offering a bounty to confidential informants for
reports of illegal guns that lead to confiscation (undertaken in Charleston, SC,
with apparent success), and “voluntary” searches of homes with suspected illegal
weapons (with an agreement to confiscate the weapons but with a commitment not
to press criminal charges for the possession) carried out in St. Louis (50).
The theory behind the confiscation strategies lies not only in the benefits of
the confiscation itself, but also in the broader deterrent effect that the risk of
confiscation has on carrying weapons or even brandishing a gun; these are the
behaviors that contribute to the diffusion of guns, especially among young people.
To the extent that gun carrying is reduced, the concern over self-protection will
also be reduced, which will diminish the incentive for others to carry their own
guns. Thus, the contagious escalation characteristic of the rise period can display
a similar contagion process of disarmament during the decline period.
Some of the policing efforts to crack down on juvenile weapon carrying seem
to have had dramatic impacts. In Boston, a program to decrease gang use of
guns has nearly eliminated youth gun homicide (40). This project involved a clear
message to gangs to control their level of violence or else face extremely tight
police surveillance and control.
The Kansas City Weed and Seed program demonstrates the effect of concerted
efforts by police to get guns out of the hands of juveniles. Police used every
encounter with youth to search for illegal weapons. In a high-risk neighborhood
with a homicide rate 20-fold the national average, the program reduced crime
by ≥50% during a 6-month period, without displacement to other neighborhoods
(52).
Enforcement activity and related community-based reactive forces almost cer-
tainly have contributed to the drop in violence in specific localities. However, the
magnitude of this effect is difficult to gauge, largely because levels of violence
have also decreased in places with no discernible change in enforcement and be-
cause the effects of enforcement tend to interact with other influences over which
the police and community leaders have little control.
The patterns of growth and decline in handgun use at various ages are reflected
in Figures 18a and 18b, which depict the trends in rates of weapons arrests at
P1: FQP
?
Figure 18 Trends in arrest rate for weapons offenses by age. (a) Arrests for weapons offenses
for ages 18, 20, 22, and 24. (b) Arrests for weapons offenses for ages 18, 17, 16, 15, and 13–14.
(Sources: Uniform Crime Report and census population data.)
P1: FQP
various ages. The pattern is very similar to the homicide patterns depicted in
Figures 8a and 8b, but there is a much more distinct peaking in 1993, with a clear
decline thereafter. Changes in the rate of weapons arrests result from a combina-
tion of changes in the presence of weapons in the population and changes in police
aggressiveness in pursuing illegal weapons. It is clear from other data (e.g. Figures
12b and 12c) that there was considerable growth in weapon prevalence during the
late 1980s and also that police became more concerned about weapons, especially
in the hands of young people. That combination is reflected in the rise in weapons
?
arrests until the peak in 1993. There is no indication that there was any abatement
by police in their concern about young people’s guns after 1993, and so it seems
likely that the decline after 1993 owes much more to a reduction in the carrying
of guns than to a slackening of police efforts to capture the guns. The reduction
in carrying seems to have contributed to the decrease in homicide after 1993 and
the decrease in robberies after 1994 by young people (11).
Independent Effects Especially on Juveniles and Young Adults

Changing Drug Markets The independent forces affecting homicide rates are
of many forms. One involves changes in the nature of the demand for drugs, as
identified by Golub & Johnson (32). The decline in the number of new crack users
in the early 1990s could well have brought some stability to drug markets, allow-
ing them to function in a more secretive and surreptitious manner off the streets
and reducing the need to keep recruiting young people who are irresponsible in
their use of violence. Those alterations could well have been amplified by the
maturity of those managing the supply side of the market, many of whom were the
survivors of the earlier violent period and had opportunity and growing incentives
to develop dispute resolution mechanisms other than violence, much as their Mafia
predecessors had done a generation earlier.
Explanations of the homicide decline that emphasize the central role of changes
in drug markets are a promising point of departure for subsequent research. For
one thing, they are causally symmetrical; the drug markets were associated with
both the increase and the decline in violence. Rates of serious violence, including
homicide and robbery, went up during the rise of the crack epidemic and have been
dropping during the decline (5, 15, 42).
The focus on changes in drug markets also helps to account for the variable
timing of the peaks and declines in violence across cities. A large coastal city
such as New York, for example, where crack took hold earlier and where it peaked
sooner than in other cities (22, 32), should have experienced a drop in its rate
of homicide sooner than in other cities—and it did. An additional advantage of
the drug market hypothesis is that it directs attention to the population groups in
which the changes in homicide were concentrated: youths, not necessarily as drug
users but as attractive sellers because of their reduced legal liability, and African-
American youths in particular, who disproportionately participated as sellers in
inner-city crack markets.
P1: FQP
Finally, the focus on the changing dynamics of the inner-city drug markets
helps to explain recent trends in other predatory crimes and the risks associated
with them. The decline in the drug markets contributed to the decline in robbery
rates in a very direct way. Robberies by drug users had to be an important part
of the growth in robbery rates in the late 1980s. Indeed, it has been suggested by
Baumer et al (5) that the simultaneous decline in burglaries was associated with the
substitution of robberies for burglaries by drug users in a hurry to get the money to
buy more drugs. As the number of drug users declined, the demand for robberies
?
to feed drug habits diminished. Also, as the carrying of guns by younger people
diminished, casual robberies that capitalized on their newfound power could well
have diminished.
Economic Expansion Some of the recent decline in homicide rates is almost

certainly related to the economic expansion of the past 6 years. Unemployment
rates have dropped to levels not seen since the early 1970s, and consumer confi-
dence is higher than in nearly 3 decades (36, 37, 56). It is important that economic
gains have been shared by racial minorities, teenagers, and high school dropouts,
groups at disproportionate risk for serious criminal violence (46).
The role of opportunities in the legitimate labor market interacts in complex
ways with changes in the illicit-opportunity structure of distressed urban com-
munities. The availability of low-wage jobs in the secondary labor market is
particularly relevant when illicit markets and the employment opportunities they
offer are shrinking. The conventional view of the connection between employment
and crime portrays individuals, especially teenagers and young adults, as turning
to criminal activity when their legitimate employment opportunities are restricted.
The relationship is likely to operate in the other direction also. Young people can
also turn to legitimate jobs in response to dwindling opportunities for illegitimate
work. Evidence from Freeman (30, 31) suggests that low-income teenagers will
substitute illegitimate for legitimate work when the perceived rewards of doing so
outweigh the costs and that a sizable fraction of inner-city young men engage in
both legal and illegal activity at the same time, moving from one to the other as
opportunity permits: “Someone may need help selling stolen goods; a car with a
stereo may be parked on a deserted street; the local fast food franchise or super-
market may be hiring. If the opportunity is there, and if the likely gain exceeds
the reservation wage, someone will act on it” (31:17).
These observations imply that the effects on criminal involvement of legitimate
and illegitimate opportunities are fundamentally interactive; when the supply of
illegitimate opportunities drops, the demand for legitimate work increases. If
the assessments of the decline in crack markets in large cities in recent years are
correct, then the movement to legitimate employment should have a particularly
pronounced effect on the level of criminal involvement of low-skilled teenagers
and especially on their willingness to risk the serious violence associated with
drug markets. There may be much to criticize about the low-end “go nowhere”
jobs produced during the economic expansion of the 1990s, but they do employ
teenagers—they are the only kind of jobs for which the great majority of teenagers
P1: FQP
are qualified—and these jobs do reduce the risk of teenagers becoming victims or
offenders.
Whatever the other drawbacks of such work, juveniles are far less likely to kill or
be killed when working in a fast-food restaurant or supermarket than when selling
crack on the street corner outside. These effects, however, are inherently short run,
not only because of the cyclical character of legitimate employment opportunities,
but also because jobs in the secondary labor market are not, by themselves, a strong
foundation for the kind of long-term integration into the mainstream economic and
?
social life of a community that is necessary to reduce the economic attractions of
crime in a permanent way.

General Decrease in Access to Guns It is clear from the earlier analyses, par-
ticularly those shown in Figures 12b and 12c, that the post-1993 decline in homi-
cides was associated with a major reduction in the use of handguns in homicides,
even among adults (Figure 12a). The general efforts to limit the opportunity for
high-risk individuals to purchase handguns, exemplified by the Brady Bill and
implemented in 1994, could well have contributed to those effects. The decline
in weapons arrests for both young adults (Figure 18a) and juveniles (Figure 18b)
began in 1994, the first year the Brady Bill was in effect. The Brady Bill required a
5-day waiting period for the purchase of handguns from licensed dealers, thereby
discouraging buying by felons. A recent study in California found that individuals
with misdemeanor records who were able to buy handguns were more than seven
times as likely to be charged with new offenses as those who had no prior criminal
record (59). A recent study found that denial of handgun purchase to high-risk
individuals reduces the subsequent risk of criminal activity (60).
A Federal Firearm License (FFL) has been traditionally easy to get and allows an
individual to purchase guns directly from a wholesaler. FFLs are rarely inspected
and operate virtually without scrutiny (21). The Brady Bill raised the dealer license
fee from $10 to $200; in addition, the Clinton administration in 1993 instructed
the Bureau of Alcohol, Tobacco, and Firearms (ATF) to be more vigorous in its
regulation of FFLs and guns. As a result, the number of federal firearm dealers
decreased from 284,000 in 1993 to 100,000 in 1998 (20).
The ATF has increased its effort to trace guns and made this information avail-
able to local jurisdictions. The Youth Crime Gun Interdiction Initiative provided
crime gun trace analysis for 37,000 guns from 17 cities in 1997 and 76,000
trace requests from 27 cities in 1998 (2, 3). Of all illegal-gun-trafficking investig-
ations conducted by the ATF, 40% involved youths. In 1998, 11.3% of crime guns
recovered in 27 cities were from juveniles and 32.4% were from young adults (2, 3).
Juvenile and young-adult offenders are obtaining both new and used guns through
the illegal-gun market. Most guns used by juveniles in crime have been recently
acquired (51), and over three-fourths of the gun traces involved new guns (2, 3). In
Boston, 26% of all traceable firearms and >40% of semiautomatic weapons recov-
ered from youths had been involved in crime within 2 years of their first legal sale.
This has important implications for the potential effectiveness of interventions. In
12 of the 17 cities in the 1997 study, the majority of the traced guns originated in
P1: FQP
the state itself. Thus, efforts to intervene on illegal-firearm traffickers, especially

when combined with aggressive confiscation efforts, can have a potential effect on
juvenile and youth violence.
Factors Affecting Older Populations

Increase in Prison Population and Incapacitation Effects The incarceration
rate has increased steeply in the last 25 years. Since the early 1970s, the incar-
?
ceration rate in the United States has climbed from 110/100,000, a rate that had
prevailed from the early 1920s to the early 1970s, to 452/100,000 (14), a com-
pounded growth rate of about 6.5% annually (7). Drug offenses alone account for
29% of this growth (8).

Although in one sense changes in sentencing policy and resulting changes in the
levels of imprisonment are clearly “reactive” concerning the problem of criminal
violence, we include them as independent factors in our discussion because the
dramatic growth in incarceration began a decade before violence rates went up
in the mid-1980s. Incarceration effects are undoubtedly important contributors
to the continuing decline of homicide rates among older people, especially for
those >30, who displayed a 40% drop in homicide rates between 1985 and 1997
(see Figure 9). This connection is particularly close because the median age of
prisoners is ∼32.
Incarceration effects are far less likely to have been a significant factor in the
more recent decline in violence rates among teenagers and youth, because so few
of them are incarcerated. In addition, levels of violence have fallen in the younger
age groups in recent years, even as their incarceration risk has increased. It is
possible, of course, that the decline might have been less steep in the absence of
the “get tough on kids” sentencing policies enacted in recent years.
Declining Domesticity and Intimate-Partner Violence In addition to the inca-

pacitation effects of increasing incarceration on adults, we have noted a marked
drop in homicides involving spouses, ex-spouses, and other “intimate partners”
over the past 2 decades. This decrease, which is especially pronounced among
African-American adults, results in part from a corresponding drop in “domestic-
ity,” that is, declining marriage rates, increasing age at marriage, and high divorce
rates (48). Some preliminary evidence suggests that the increasing availability of
legal advocacy and other domestic-violence services also may have played a role
(24, 26).
CONCLUSIONS
Violence is both a public health and a criminal justice problem. Neither dis-
cipline has exclusive jurisdiction; both are necessary to address the continuing
problem of violent injuries in our society. Public health and criminal justice can
P1: FQP
be complementary and integrated into an overall program that will be more effec-
tive than many efforts in the past (49). Primary, secondary, and tertiary strategies
aimed at the human, environmental, and instrumental elements of violence will
ultimately be necessary. The experience of the last decade and the available evi-
dence indicate that a number of interventions are necessary and can be successful
in further decreasing violent crime. Future interventions should focus on the high-
risk groups (i.e. juveniles and young adults) and be guided by the success of the
last 5 years.
?
Finally, the current downturn in rates of violent offenses by and victimization
of juveniles and young adults, although these rates are still higher than those in the
early 1980s, provides an opportunity to consider long-term prevention strategies.

When crime and violence rates are very high, the public demands and the criminal
justice system provides immediate short-term interventions to reduce violence.
However, when rates are lower, attention may be more properly focused on devel-
oping longer-term interventions, particularly primary prevention, and in providing
the chance to develop and evaluate new programs. This may involve changes in
the approach to gun ownership, such as registration and licensing, development of
better surveillance and data systems, and early childhood interventions to allow
children to grow up to be healthy, successful, and contributing members of society.
Visit the Annual Reviews home page at www.AnnualReviews.org
LITERATURE CITED
1. Addams J. 1972. The Spirit of Youth and 7. Blumstein A. 1998. US criminal justice
City Streets. Urbana: Univ. Ill. Press. 162 conundrum: rising prison populations and
pp. stable crime rates. Crime Delinq. 44:127–
2. Alcohol, Tobacco, and Firearms. 1999a. The 35
Youth Crime Gun Interdiction Initiative. Per- 8. Blumstein A, Beck A. 1999. Prisons: Pop-
form. Rep. US Dep. Treas., Washington, ulation growth in U.S. prisons, 1980–1996.
DC In Crime and Justice: An Annual Review of
3. Alcohol, Tobacco, and Firearms. 1999b. Research, 26:391–435, ed. M Tonry, J Pe-
Gun Crime in the Age Group 18–20. Rep. tersilia. In press
US Dep. Treas., Washington, DC 9. Blumstein A, Cork D. 1996. Linking gun
4. Anderson DC. 1997. The mystery of the availability to youth gun violence. Law Con-
falling crime rate. Am. Prospect. 32(7): temp. Probl. 59:5–24
49 10. Blumstein A, Rosenfeld R. 1998. Exploring
5. Baumer E, Lauritsen J, Rosenfeld R, Wright recent trends in U.S. homicide rates. J. Crim.
R. 1998. The influence of crack cocaine Law Criminol. 88:1175–216
on robbery, burglary, and homicide rates: 11. Blumstein A, Rosenfeld R. 1999. Trends in
a cross-city, longitudinal analysis. J. Res. rates of violence in the U.S.A. Stud. Crime
Crime Delinq. 35:316–40 Crime Prev. 22:1–29
6. Blumstein A. 1995. Youth violence, guns, 12. Bureau of the Census. 1996. Statistical Ab-
and the illicit-drug industry. J. Crim. Law stract of the United States 1996. Springfield,
Criminol. 86:10–36 VA: Natl. Tech. Inf. Serv.
P1: FQP
13. Bureau of Justice Statistics. 1997. Felony violence resources. Homicide Stud. 3:187–
Sentences in State Courts, 1994. US Dep. 214
Justice, Washington, DC 25. Easterlin RA. 1987. Birth and Fortune:
14. Bureau of Justice Statistics. 1999. Prison The Impact of Numbers on Personal Wel-
and jail inmates at midyear 1998. BJS Bull. fare. New York: Basic Books. 2nd ed.
No. NCJ-173414. US Dep. Justice, Wash- 26. Fagan J. 1996. The Criminalization of Do-
ington, DC mestic Violence: Promises and Limits.
15. Butterfield F. 1995. Many cities in the U.S. Washington, DC: Natl. Inst. Justice
?
show sharp drop in homicide rates. NY 27. Federal Bureau of Investigation. 1972–
Times, August 13, p. 11 1997. Crime in the United States. Unif.
16. Callahan CM, Rivara FP. 1992. Urban Crime Rep. US Govt. Print. Off., Wash-
high school youth and handguns: a school- ington, DC

based survey. J. Am. Med. Assoc. 267: 28. Fingerhut LA, Ingram DD, Feldman JJ.
3038–42 1998. Homicide rates among US teenagers
17. Centers for Disease Control and Preven- and young adults: differences by mecha-
tion. 1994. Deaths resulting from firearm- nism, level of urbanization, race and sex,
and motor vehicle-related injuries—Uni- 1987 through 1995. J. Am. Med. Assoc.
ted States, 1968–1991. Morbid. Mortal. 280:423–27
Wkly. Rep. 43:37–42 29. Fox JA, Zawitz MW. 1999. Homicide
18. Centers for Disease Control and Preven- Trends in the United States. Crime Data
tion. 1995. Youth Risk Behavior Sur- Brief. US Dep. Justice, Bur. Justice Stat.,
veillance—United States, 1993. Morbid. Washington, DC
Mortal. Wkly. Rep. 23:1–56 30. Freeman RB. 1994. Crime and the job mar-
19. Committee on Trauma Research, Comm. ket. Natl. Bur. Econ. Res., Working Pap. No.
Life Sci., Natl. Res. Counc. Inst. Med. 4910. Cambridge, MA: Natl. Bur. Econ.
1985. Injury in America: A Continuing Res. 44 pp.
Public Health Problem. Washington, DC: 31. Freeman RB. 1996. Why do so many young
Natl. Acad. Press. 164 pp. American men commit crimes and what
20. Cook P. 1998. The epidemic of youth gun might we do about it? Work. Pap. Ser.,
violence. In Perspectives on Crime and Natl. Bur. Econ. Res., Work. Pap. No. 5451.
Justice. Washington, DC: Natl. Inst. Jus- Cambridge, MA: Natl. Bur. Econ. Res. 35
tice pp.
21. Cook PJ, Molliconi S, Cole TE. 1995. Reg- 32. Golub AL, Johnson BD. 1997. Crack’s de-
ulating gun markets. J. Crim. Law Crimi- cline: some surprises across U.S. cities.
nol. 86:59–91 Res. Brief. US Dep. Justice, Washington,
22. Cork D. 2000. Examining space-time DC
interaction in city-level homicide data: 33. Greenfeld LA, Rand MR, Craven O, Klaus
crack markets and the diffusion of guns PA, Perkins CA, et al. 1998. Violence by
among youth. J. Quant. Criminol. 15:In Intimates. Washington, DC: US Dep. Jus-
press tice
23. DiIulio JJ. 1995. Rule of law: why violent 34. Greenfeld LA, Zawitz MW. 1995. Weapon
crime rates have dropped. Wall St. J. Sept. offenses and offenders. Bur. Justice Stat.
6, p. A19 Sel. Find. Nov.
24. Dugan L, Nagin D, Rosenfeld R. 1999. Ex- 35. Hemenway D, Prothrow-Stith D, Bergstein
plaining the decline in intimate partner JM, Ander R, Kennedy BP. 1996. Gun car-
homicide: the effects of changing do- rying among adolescents. Law Contemp.
mesticity, women’s status, and domestic Probl. 59:39–54
P1: FQP
36. Hershey RD. 1997a. US jobless rate de- 50. Rosenfeld R, Decker S. 1996. Consent to
clines to 4.7%, lowest since 1973. NY search and seize: evaluating an innovative
Times, Nov. 8, p. A1 youth firearm suppression program. Law
37. Hershey RD. 1997b. Confidence of con- Contemp. Probl. 59:197–220
sumers is surging. NY Times, Dec. 31, 51. Sheley JF, Wright JD. 1993. Gun Acquisi-
p. C1 tion and Possession in Selected Samples.
38. Jacobs B. 1999. Dealing Crack: The So- Washington, DC: Natl. Inst. Justice
cial World of Streetcorner Selling. Boston, 52. Sherman L, Shaw JW, Rogan DP. 1995.
?
MA: Northeastern Univ. Press. 163 pp. The Kansas City gun experiment. NCJ
39. Janofsky M. 1998. Trend, some cities see 150855. Res. Brief. US Dep. Justice, Wash-
murders rise. NY Times, Jan. 15, p. A16 ington, DC

40. Kennedy DM, Piehl AM, Braga AA. 1996. 53. Smith MD. 1986. The era of increased vi-
Youth violence in Boston: gun markets, se- olence in the United States: age, period, or
rious youth offenders, and a use-reduction cohort effect. Sociol. Q. 27:239–51
strategy. Law Contemp. Probl. 59:147–96 54. Snyder HN, Finnegan TA. 1997. Easy Ac-
41. Kocieniewski D. 1997. New York City cess to the FBI’s Supplementary Homicide
murder rate may hit 30-year low. NY Times, Reports: 1980–95. Washington, DC: Off.
Dec. 25, p. B1 Juv. Justice Delinq. Prevent.
42. Lattimore PK, Riley KJ, Trudeau T, Leiter 55. Steffensmeier D, Streifel C, Shihadeh ES.
J, Edwards S. 1997. A study of homicide 1992. Cohort size and arrest rates over the
in eight US cities. Res. Brief, pp. 1–9. Natl. life course: the Easterlin hypothesis recon-
Inst. Justice, Washington, DC sidered. Am. Sociol. Rev. 57:306–14
43. Loftin C. 1986. Assaultive violence as a 56. Uchitelle L. 1997. 6 years in the plus col-
contagious social process. Bull. NY Acad. umn for the US economy. NY Times, Mar.
Med. 62:550–55 12, p. C1
44. Maxim PS. 1985. Cohort size and juvenile 57. Webster DW, Gainer PS, Champoin HR.
delinquency: a test of the Easterlin hypoth- 1993. Weapon carrying among inner-city
esis. Soc. Forces 63:661 junior high school students: defensive be-
45. O’Brien RM. 1989. Relative cohort size havior vs. aggressive delinquency. Am. J.
and age-specific crime rates: an age- Public Health 83:1604
period-relative-cohort-size model. Crimi- 58. Williams KR, Flewelling RL. 1988. Fam-
nology 29:57 ily, acquaintance, and stranger homicide:
46. Nasar S, Mitchell KB. 1998. Booming job alternative procedures for rate calculations.
market draws young black men into fold. Criminology 25:543–60
NY Times, May 23, p. A1 59. Wintemute GJ, Drake CM, Beaumont JJ,
47. Riedel, M. 1998. Counting stranger homi- Wright MA, Parham CA. 1998. Prior mis-
cides: a case study of statistical prestidig- demeanor convictions as a risk factor for
itation. Homicide Stud. 2:206–19 later violent and firearm-related criminal
48. Rosenfeld R. 1997. Changing relationships activity among authorized purchasers of
between men and women: a note on the de- handguns. J. Am. Med. Assoc. 280:2083–
cline in intimate partner homicide. Homi- 87
cide Stud. 1:72–83 60. Wright MA, Wintemute GJ, Rivara FP.
49. Rosenfeld R, Decker S. 1993. Where pub- 1999. Effectiveness of denial of handgun
lic health and law enforcement meet: mon- purchase to persons believed to be at high
itoring and preventing youth violence. Am. risk for firearm violence. Am. J. Public
J. Police 12:11–57 Health 89:88–90
Annual Review of Public Health
Volume 21, 2000
CONTENTS
PUBLIC HEALTH GENETICS: An Emerging Interdisciplinary Field for
the Post-Genomic Era, Gilbert S. Omenn 1
HOST-PATHOGEN INTERACTIONS IN EMERGING AND RE-
EMERGING INFECTIOUS DISEASES: A Genomic Perspective of
Tuberculosis, Malaria, Human Immunodeficiency Virus Infection,
Hepatitis B, and Cholera, Janet M. McNicholl, Marie V. Downer,
Venkatachalam Udhayakumar, Chester A. Alper, David L. Swerdlow
15
NUTRITION, GENETICS, AND RISKS OF CANCER, Cheryl L. Rock,
Johanna W. Lampe, Ruth E. Patterson 47
POPULATION SCREENING IN HEREDITARY

HEMOCHROMATOSIS, Arno G. Motulsky, Ernest Beutler 65

THE INTERFACE OF GENETICS AND PUBLIC HEALTH: Research
and Educational Challenges, Melissa A. Austin, Patricia A. Peyser, Muin
J. Khoury 81
LOOKING BACK ON ""CAUSAL THINKING IN THE HEALTH
SCIENCES, J. S. Kaufman, C. Poole 101
CAUSAL EFFECTS IN CLINICAL AND EPIDEMIOLOGICAL
STUDIES VIA POTENTIAL OUTCOMES: Concepts and Analytical
Approaches, Roderick J. Little, Donald B. Rubin 121
BUILDING BRIDGES BETWEEN POPULATIONS AND SAMPLES
IN EPIDEMIOLOGICAL STUDIES, W. Kalsbeek, G. Heiss 147
MULTILEVEL ANALYSIS IN PUBLIC HEALTH RESEARCH, Ana V.
Diez-Roux 171
SHOULD WE USE A CASE-CROSSOVER DESIGN, M. Maclure, and
M. A. Mittleman 193
WATER RECLAMATION AND UNRESTRICTED NONPOTABLE
REUSE: A New Tool in Urban Water Management, Daniel A. Okun 223
EPIDEMIOLOGY AND PREVENTION OF INJURIES AMONG
ADOLESCENT WORKERS IN THE UNITED STATES, Carol W.
Runyan, Ronda C. Zakocs 247
THE EFFECTS OF CHANGING WEATHER ON PUBLIC HEALTH,
Jonathan A. Patz, David Engelberg, John Last 271
TOXICOLOGICAL BASES FOR THE SETTING OF HEALTH-
RELATED AIR POLLUTION STANDARDS, M. Lippmann, R. B.
Schlesinger 309
RELIGION AND HEALTH: Public Health Research and Practice, Linda
M. Chatters 335
A REVIEW OF COLLABORATIVE PARTNERSHIPS AS A
STRATEGY FOR IMPROVING COMMUNITY HEALTH, Stergios
Tsai Roussos, Stephen B. Fawcett 369
ORAL HEALTH IN THE UNITED STATES: The Post-Fluoride
Generation, P. Milgrom, S. Reisine 403
THE NEW PUBLIC HEALTH LITIGATION, W. E. Parmet, R. A.
Daynard 437
BABY AND THE BRAIN: Advances in Child Development, Janet A.
DiPietro 455
HEALTH PROMOTION IN THE CITY: A Review of Current Practice
and Future Prospects in the United States, N. Freudenberg 473
THE RISE AND DECLINE OF HOMICIDE- AND WHY, Alfred
Blumstein, Frederick P. Rivara, Richard Rosenfeld 505
NCOME INEQUALITY AND HEALTH: What Does the Literature Tell
Us?, Adam Wagstaff, Eddy van Doorslaer 543
EVALUATING THE STATE CHILDREN''S HEALTH INSURANCE
PROGRAM: Critical Considerations, Barbara Starfield 569
PREFERENCE-BASED MEASURES IN ECONOMIC EVALUATION
IN HEALTH CARE, Peter J. Neumann, Sue J. Goldie, Milton C.
Weinstein 587
TELEMEDICINE: A New Health Care Delivery System, Rashid L.
Bashshur, Timothy G. Reardon, Gary W. Shannon 613
THE CHANGING NATURE OF RURAL HEALTH CARE, Thomas C.
Ricketts 639
ASSESSMENT IN LONG-TERM CARE, R. L. Kane, R. A. Kane 659

The Rise and Decline of Homicide and Why

Enviado por

Dados do documento

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

The Rise and Decline of Homicide and Why

Enviado por

Direitos autorais:

Formatos disponíveis

P1: FQP

March 31, 2000 10:55 Annual Reviews CHAP-21

Annu. Rev. Public Health. 2000. 21:505–41

THE RISE AND DECLINE OF HOMICIDE—

1Carnegie Mellon University, Pittsburgh, Pennsylvania

Key Words violence, firearms, drugs, cocaine, youth, crime

506 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

in the United States by the year 2000 (17).

THE RISE AND DECLINE OF HOMICIDE 507

508 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

THE RISE AND DECLINE OF HOMICIDE 509

510 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

in the aggregate rate of victimization occurred in the category of homicides by

THE RISE AND DECLINE OF HOMICIDE 511

Relationship Factors A final component of homicide with important implica-

512 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

Figure 6 Trends in the percent distribution of homicide victim-offender relationships. Closed

THE RISE AND DECLINE OF HOMICIDE 513

514 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

THE RISE AND DECLINE OF HOMICIDE 515

516 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

THE RISE AND DECLINE OF HOMICIDE 517

518 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

FACTORS CONTRIBUTING TO THE INCREASE

Changing Demographic Composition

THE RISE AND DECLINE OF HOMICIDE 519

The Changing Role of Handguns, Especially Among Juveniles

520 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

whites, certainly suggests an important role of the increased availability of guns.

522 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

How Did the Young People Get the Guns:

THE RISE AND DECLINE OF HOMICIDE 523

524 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

THE RISE AND DECLINE OF HOMICIDE 525

526 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

THE RISE AND DECLINE OF HOMICIDE 527

528 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

THE RISE AND DECLINE OF HOMICIDE 529

37% of male high school students owned guns (51).

ROLE OF THE BIG CITIES

The largest cities contribute disproportionately to patterns of serious violence for

530 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

THE RISE AND DECLINE OF HOMICIDE 531

532 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

SOME OBSERVATIONS ABOUT CAUSES

THE RISE AND DECLINE OF HOMICIDE 533

Reactive Response to Factors Contributing to the Rise

mixture of aggressive stop-and-frisk methods, especially in high-violence neigh-

534 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

THE RISE AND DECLINE OF HOMICIDE 535

Independent Effects Especially on Juveniles and Young Adults

536 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

Economic Expansion Some of the recent decline in homicide rates is almost

THE RISE AND DECLINE OF HOMICIDE 537

crime in a permanent way.

538 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD

the state itself. Thus, efforts to intervene on illegal-firearm traffickers, especially

Factors Affecting Older Populations

29% of this growth (8).

Declining Domesticity and Intimate-Partner Violence In addition to the inca-

THE RISE AND DECLINE OF HOMICIDE 539

early 1980s, provides an opportunity to consider long-term prevention strategies.

Visit the Annual Reviews home page at www.AnnualReviews.org

540 BLUMSTEIN ¥ RIVARA ¥ ROSENFELD