U.S. Department of Justice
Office of Justice Programs
Bureau of Justice Statistics
Special Report

Murder in Families


July 1994, NCJ 143498



----------------------------------------------------------------
This file is text only without graphics and many of the tables.
A zip archive of the tables in this report in spreadsheet format.
(.wk1) and the full report including tables and graphics in 
.pdf format are available from:
http://www.ojp.usdoj.gov/bjs/abstract/mf.htm
        
-----------------------------------------------------------------

By John M. Dawson
and Patrick A. Langan, Ph.D.
BJS Statisticians
          
A survey of murder cases disposed 
in 1988 in the courts of large urban
counties indicated that 16% of murder
victims were members of the defendant's 
family. The remainder were murdered by 
friends or acquaintances (64%) or by 
strangers (20%).  These findings are 
drawn from a representative sample 
survey of State and county prosecutors' 
records. The survey covered disposed 
charges against nearly 10,000 murder
defendants, whose murder cases
accounted for over 8,000 victims.

Other findings include the
following:

*  Among murder victims 6.5% were killed 
by their spouses, 3.5% by their parents, 
1.9% by their own children, 1.5% by their 
siblings, and 2.6% by some other family
member.

*  A third of family murders involved 
a female as the killer.  In sibling
murders, females were 15% of killers,
and in murders of parents, 18%.  But
in spouse murders, women represented 
41% of killers.  In murders of their 
offspring, women predominated, 
accounting for 55% of killers.

*   Among black marital partners, wives 
were just about as likely to kill their 
husbands as husbands were to kill their 
wives: 47% of the victims of a spouse 
were husbands and 53% were wives. Among 
white victims murdered by their spouse, 
wives were much less likely to be the 
killers:  38% of the 

victims were husbands and 62% were
wives.

*  Forty-five percent of family murder
victims were female, compared to 18% of 
nonfamily murder victims.

When a mother killed her own child, the 
offspring she killed was more likely to 
be a son than a daughter:  64% sons versus 
36% daughters.  But when a father killed 
his own child, the offspring he killed was 
about as likely to be a daughter as a 
son: 52% daughters versus 48% sons.
-------------------------------------

July 1994

The United States has over 3,000
counties, but more than half of all
murders occur in just 75 of them, the
Nation's most populous jurisdictions.
This report taps a rich source of
murder data -- prosecutors' files in a
sample of these large urban places --
for detailed information on the nature
and extent of a particular type of
murder: those that occur within families.  
In addition, the report uses these files 
justice systems respond to family murder.

This study was possible as a result of
the generous cooperation of urban
prosecutors and their staffs in
jurisdictions throughout the Nation. 
On behalf of BJS, I want to express
my sincere appreciation.

Lawrence A. Greenfeld
Acting Director
----------------------------------------- 
 
*   When a son killed a parent, his
victim was about as likely to be the
mother as the father:  47% mothers
versus 53% fathers.  But when a
daughter killed a parent, her victim
was more likely to be the father than
the mother:  81% fathers versus 19%
mothers.

*  In murders of persons under age 12,
the victims' parents accounted for 57%
of the murderers.

*  Eleven percent of all victims age 
60 or older were killed by a son or
daughter.

*  No significant difference in the
conviction rate separated family
murder defendants (76%) from
nonfamily murder defendants (72%). 

*  Convicted family murder defendants
(88%) were as likely to receive a
prison sentence as convicted nonfamily 
murder defendants (91%).

*  Firearms were used in the killing 
of 42% of all family murder victims,
compared to 63% of all nonfamily
murder victims.

*  Seventy-four percent of murder
defendants had a prior criminal record
of arrest or conviction for a crime.  A
substantial percentage of murder victims, 
44%, also had a prior criminal record.  
However, 19% of family murder victims 
had a prior record, compared to 51% of 
nonfamily murder victims.  Also, 56% of 
family murder defendants, compared to 
77% of other murder defendants, had a 
prior record.

------------------------------------------
Compared to other murder victims,
family murder victims were --

More often

*  female than male:
45% versus 18%
*  under age 12:
19% versus 2%)
*  age 60 or older:
12% versus 6%
*  killed during the daytime:
39% versus 25%
*  killed in the victim's own 
home: 82% versus 22%

Less often

*  identified as alcohol users:
33% versus 51%
*  unemployed:
7% versus 16%
*  killed by a firearm:
42% versus 63%
*  armed:  15% versus 20% (page 4)
*  killed by multiple assailants:
9% versus 19%
*  identified as having a past record
or arrest or conviction:
*  involved in murders in which both
the defendant and the victim had a
prior criminal record:
14% versus 43%).

Compared to other murder defend-
ants, defendants in family murders--

More often

*  were age 30 or older:
57% versus 32%
*  had a history of mental 
illness: 14% versus 3%
*  committed a daytime murder:
38% versus 24%
*  committed the murder in the 
defendant's own home:
64% versus 10%.

Less often

*  were under age 30:
43% versus 68%
*  were identified as alcohol 
users: 48% versus 68%
*  were unemployed:
29% versus 37%
*  had a victim of the same sex:
34% versus 80%
*  used a firearm:
43% versus 64%
*  had a record of arrest or 
conviction: 56% versus 77%
*  were involved in murders in which
both the defendant and the victim had
a prior criminal record: 15% versus 
44%.
------------------------------------------

The survey

Survey data were compiled from State
prosecutors' files and were based on 
a representative sample of all murder
cases disposed in large urban counties 
in 1988.  The murders were committed 
in 1988 or earlier.  The Nation's 75 
largest counties (as defined by the 
number of arrests and population size) 
formed the population from which 33 
counties were systematically sampled 
for the survey. 

Within each of the 33 sampled counties, 
a criminal case was eligible for sampling 
if (a) 1 or more defendants in the case 
were charged with murder, and (b) at 
least 1 murder defendant in the case was 
disposed by a court in 1988.  The sample
consisted of 2,539 murder cases against 
3,119 defendants and involving 2,655 
victims.  When statistically weighted, 
the 3,119 defendants in the sampled cases
represented 9,576 murder defendants in 
the Nation's 75 largest counties, and the 
2,655 victims represented 8,063 victims 
in the 75 largest counties. The weighted 
estimate of 8,063 victims accounted for 
39% of the nationwide total of 20,860 
murder victims in 1988.

Murder includes (1) intentionally
causing the death of another person
without extreme provocation or legal
justification, (2) causing the death of
another while committing or attempting
to commit another crime, and (3)
nonnegligent or voluntary manslaughter.  
Murder excludes negligent or involuntary 
manslaughter as well as attempted murder, 
which is classified as aggravated assault. 
Murder also includes accessory to
murder, aiding and abetting a murder,
and facilitating a murder.  When the
term murder is used in this report
with-out qualification, it includes
nonnegligent manslaughter. Defendant 
in this report refers to a person 
arrested for murder and presented by 
the police for prosecution.

Victims

Murder cases disposed in the 75 largest 
counties in 1988 involved an estimated 
8,063 victims. Sixteen percent of victims 
had a family relationship to at least 
one defendant in the case.  The most 
frequent specific relationship was that 
of spouse; the least frequent, sibling. 
Nonfamily victim-offender relationships
characterized 84% of the victims, who
were a stranger to, an acquaintance
of, or a friend of the defendant or
defendants.

Defendants

Murder cases disposed in the 75 largest 
counties in 1988 involved an estimated 
9,576 defendants.  Thirteen percent had 
a family relationship to at least one of 
the victims in the case. Nonfamily victim-
offender relationships characterized 87% 
of the defendants who were a stranger to, 
an acquaintance of, or a friend of the 
victim or victims.

Family murder defendants and their
victims comprised almost 15% of all
victims and defendants recorded in the
murder cases in 1988.  Husbands and
wives were the most likely to be
involved in family murders.  Spouses
were 4 in 10 of all defendants and
victims involved in a family murder: 

The second most frequent type of family 
murder, with offspring as victims and 
parents as assailants, were 21% of the 
total. (See the box describing murders 
by parents of their children.)  Those 
cases in which offspring were assailants 
and parents were the victims comprised 
12% of all family murder victims. 

As among victims, most of the nonfamily 
murder defendants (93%) and most of the 
family murder defendants (66%) were male. 
Offspring murder was the only murder
category in which females predominated 
as killers. In offspring murders, mothers 
accounted for 55% of the defendants. 

Sons, more often than daughters, were 
the defendants in the murders of parents:  
82% versus 18%.  Compared to defendants 
in other types of family murder, offspring 
accused of killing their parents were 
the youngest of the assailants, two-
thirds being under age 30.

Husbands killed wives more frequently
than wives killed husbands. Overall,
husbands comprised about 60% of the
assailants in spouse killings. The
predominance of husbands as the
defendant, however, varied by race. 
In black murders, wives were about as
likely as husbands to be charged with
the murder of their spouse.  Of the 283
black-on-black spouse killings, 53% of
the assailants were husbands, compared 
to 62% of the 218 white-on-white spouse 
killings.  In the 11 Asian, Native 
American, Pacific Islander, or Alaska 
Native spouse murders, the husband 
killed the wife.

Drinking, mental illness, and
unemployment

About half of the nonfamily murder victims 
and a third of the family murder victims 
consumed alcoholic drinks before the crime.  
Compared to victims in other types of 
family murder, victims in spouse murders 
were the most likely to have been drinking 
(50%). Sixty-eight percent of nonfamily 
murder defendants and 48% of family murder 
defendants were drinking at about the time 
of the murder.

Family murderers were more likely
than nonfamily murderers to have a
history of mental illness (14%). 
Those who killed their parents were
particularly likely to have such a
history (25%).

Parents who were murdered were, apart 
from offspring murder victims, the 
least likely to be unemployed. 

Sex of offender and victim

Defendants in family murder cases were 
much less likely to be accused of
murdering someone of the same sex
(34% of defendants) than were
defendants in murder cases not
involving family members (80%). 
When the sex of the victim and
offender is considered, spousal
murder, which by definition includes 
a man and a woman, provides the
primary source of the difference
between family and nonfamily murder. 
If spousal murder is excluded from
consideration and "other" family
members are included, murderers and
victims were of the same sex in 65%
of family murders.

Female defendants were more likely
than male defendants to have murdered 
a person of the opposite sex.  When a 
mother murdered her own child, the 
offspring she killed was more often a 
son (64%) than a daughter (36%). Among 
fathers who murdered, 48% of their 
victims were a son and 52% were a 
daughter.  When a daughter killed a 
parent, the victim was more likely to 
be a father (81%)than a mother (19%).  
Among sons who murdered a parent, 53% 
of the victims were a father.  When a 
sister murdered a sibling, 55% of the 
victims were a brother.  Among brothers 
who killed a sibling, 74% of the victims
were a brother.

Guns used as the murder weapon

Over 60% of nonfamily murders and over 
40% of family murders were committed with 
a gun. Compared to victims in other types 
of family murder, victims in spouse murders 
were the most likely to have died from a 
gunshot (53%).  Offspring were the least 
likely to be shot to death: 1 in 5 
offspring murdered by a parent died 
from a bullet wound.

Armed victims and 
victim-precipitated murders 

Nineteen percent of murder victims
were armed with a gun, knife, or other
deadly weapon.  A smaller percentage
of family murder victims (15%) than of
nonfamily murder victims (20%) were
armed.

Some armed victims used the deadly
weapon to provoke the defendant. 
Others provoked the defendant with a
nonlethal weapon or their fists or by
pushing the defendant.  Altogether,
19% of the victims in some way
provoked the defendant.  The 
provocation did not vary significantly
between family (17%) and nonfamily
murders (20%).

Multiple victims and assailants

Victims in murders of family members
were about half as likely as nonfamily
murder victims to have had multiple
assailants.  However, similar
percentages of defendants in both
types of murder (family murders, 6%,
and other murders, 5%) were charged
with killing more than one person. 
Compared to defendants in other
types of family murder, defendants
accused of killing their offspring or
their parents were the most likely to
have multiple victims.  These murders
were also the most likely to involve
multiple assailants.

Criminal history

A little over half of the defendants in
family murders, but over three-quarters 
of defendants in nonfamily murders, had 
been arrested in the past.  Defendants 
were more likely than their victims to 
have a criminal history.  Nevertheless, 
44% of murder victims (51% nonfamily 
and 19% family) had a prior history of
arrest or conviction.  

Whether or not the victim had a criminal 
history, the killer had such a history 
in most cases.  The likelihood of the 
killer's having a prior record was
greater when the victim also had a
record.  Victims with a prior criminal
record accounted for 44% of all
victims;  83% of these victims were
killed by someone with a prior criminal
history. (See Methodology, for calculation 
of the percentage of defendants with a 
prior record.)  For family murder victims 
the percentages were lower:  21% had a 
criminal record and 69% of these victims 
were killed by someone with a prior record. 
Among the murderers of family victims
without a criminal record, 42% had such 
a history.

Age of victim and victim-assailant
family relationship

When a person under age 12 is murdered, 
a family member is the best suspect, 
according to survey results; family 
members killed 63% of the child murder 
victims.

However, a family member is an unlikely 
suspect in murders of persons in their 
teens.  Among victims age 20 or older, 
the likelihood that a family member was 
the killer increased as the victim's age 
increased.  

A family member was involved in the
murder of 27% of the murder victims
age 60 or older.  Among all murders of
persons in the oldest group, offspring
as the killer accounted for 11%. Among 
murders of family members age 60 or older, 
the most frequent assailant category was 
not the spouse, as for younger adult 
victims, but the victim's offspring. The
assailant was an offspring in 42% of
family member victims age 60 or older
and a spouse in 24%. The most frequent 
family member category varied by the 
victim's age:

Murder weapons used against
young children and elderly parents

When parents killed their offspring
under age 12, they rarely used a
firearm or knife. Firearms or knives
were responsible for the deaths of 7%
of offspring victims under 12.  When
sons and daughters killed their parents
age 60 or older, in most cases they did
not use a firearm or knife.  Firearms or
knives accounted for 44% of family
murder victims age 60 or older.

---------------------------------------
Victims who were children
under age 12

A parent was the assailant in the 
majority (57%) of family murders
involving victims under 12.

For all murder victims under age 12,
death was often preceded by child
abuse.  In 57% of cases the assailant
had abused the murder victim under
age 12.  Among offspring murder
victims who were under age 12, before
their death  79% had suffered abused
by the assailant.

Rape or sexual assault preceded the
death of 6% of murder victims under
age 12.  These crimes occurred less
often when the assailant was a parent,
accounting for 1% of offspring murder
victims under age 12.
------------------------------------------

Strangulation, use of a blunt
instrument, and pounding by fists or
feet were among the more frequent
methods of death when firearms or
knives were not used.

------------------------------------------- 
        
Parental murder of offspring
under age 12

Prosecutors' files contained information
on reasons why a parent murdered an
offspring under age 12.  One or more 
reasons were given for 62 of the total 
84 offspring murder victims under age 
12. The following presents reasons and 
the number of victims:

-- Unspecified forms of child abuse
(18)
-- Victim's behavior, such as crying
or misbehavior (15)
-- Parent's emotional instability 
or retardation (9)
-- Unwanted newborn baby (8)
-- Unintended consequence of the 
commission of another crime (lethal
conflict between the parents) (6)
-- Neglect (5)
-- Difficulty handling the
responsibility
of child rearing (3)
-- Child held hostage (1).

Examination of the details concerning the 
method of killing covered all but 3 of 
the victims.  By far the most frequent 
method of murder was beating:  punching 
with fists, kicking, throwing, pushing, 
slapping, hitting (with belts, hammers, 
or wooden brushes), and striking body
against furniture (shower head or walls). 

With five of the victims counted under 
two or three methods of murder,
specific methods and the number of
victims were as follows:
-- Beating (35)
-- "Shaken baby syndrome" (10)
-- Arson (6)
-- Newborn disposed of in toilet
or trash can (6)
-- Drowning in bathtub (6)
-- Firearm (5)
-- Suffocation/strangulation (5)
-- Neglect (dehydration, starvation,
and failure to use infant heart 
monitor)(4)
-- Stabbing (3)
-- Starvation (2)
-- Other methods, including
poisoning with carbon monoxide, 
lethal doses of drugs, running over 
with a car, boiling, and putting in 
freezer (5).

Of the five victims who were shot to
death, three died because the
assailant accidentally fired a gun
while committing another  crime;
therefore, two offspring victims
under age 12 were intentionally
killed with a firearm.
----------------------------------------

Arrested for murder or
manslaughter

The family relationship of the
defendant to the murder victim made
little difference in whether the
defendant was charged with first-degree 
murder or a less serious type of homicide.  
When a family member was the victim, 
73% of the defendants were charged with 
first degree murder, compared to 74% of
the defendants charged with murdering a 
stranger or acquaintance.  Compared to 
defendants in other types of family murder, 
defendants in offspring murders were the 
most likely to have had voluntary or 
nonnegligent manslaughter as the most 
serious arrest charge. 

Time to arrest and disposition

In more than half of the family murder
cases, but in about a third of the cases
of other types of victims, the arrest
occurred on the day of the crime. 
Spouses and siblings were identified
as the murderer more quickly than
parents or offspring.  Overall, family
murder cases required less time to
disposition than other types of murder
cases:  Prosecution was completed
within 6 months for 34% of family
murder defendants versus 29% 
of nonfamily murder defendants. 

Justice system response 
to family murder

The criticism is sometimes made that
police, prosecutors, and judges treat
family violence less seriously than
violence between strangers and other
unrelated persons.***Delbert S. Elliott,
"Criminal Justice Procedures in Family
Violence Crime," in Lloyd Ohlin and 
Micheal Tonry (eds.), Family Violence
(Chicago: University of Chicago Press,
1989), pp. 428.***  The urban county
data provide little support for such
criticism.  In several important
respects, the criminal justice outcomes
of family murder defendants were about 
the same overall as those of other murder 
defendants.  Where differences in the 
overall case outcome existed, the more 
detailed statistical testing of data 
removed the characteristic of nonfamily-
family as a possible source of those 
differences.

Compared to other murder defendants, 
those in family murder cases were  

*  As likely to be charged with
first-degree murder

*  No more likely to have their cases
diverted, rejected for prosecution, or 
to be acquitted

*  Less likely to be dismissed by the
court 

*  As likely to have their cases result 
in a conviction for some crime and,
specifically, as likely to be convicted 
of murder . 

Compared to other defendants at
sentencing, convicted family murder
defendants, including those initially
charged with murder of a family
member but convicted of some other
offense, were not significantly less
likely to receive --

*  a prison sentence:  88% versus 91%
or

*  a sentence to life imprisonment:  
13% versus 16%.

Compared to nonfamily murder defendants, 
however, convicted family murderers were --

*  more likely to receive a less severe
sentence, probation:  7% versus 3%

*  given shorter prison sentences:  
13 years, on average, versus 15 years.

Analyzing differences between
sentences for family murder and 
other types of murder

These findings of more probation
sentences and shorter prison terms do
not necessarily reflect greater
tolerance of family violence by justice
system officials.  Numerous differences 
between family murder and other murder 
might possibly account for less severe 
sentencing for family murder. To test 
that possibility, regression analysis 
was applied to the data.

This analysis accounts for the
simultaneous effects on sentencing of
the following factors:

offense seriousness (degree of murder)
nature of conviction (trial or guilty 
plea)defendant criminal history age of 
defendant number of victims victim 
precipitation county (place-to-place 
variability in sentencing can be 
substantial) race and sex of both 
victim and defendant. 

Results showed that sentencing
differences are not statistically
significant (.05-level) when differences
in case characteristics are taken into
account.  

Regression analysis did not confirm
differences in sentencing severity
between family and nonfamily murders
because defendants in family murders
less often had characteristics that are
associated with the more severe
sentences.  

As indicated in the discussion and
tables describing family murders, for
example --

-- Voluntary or nonvoluntary
manslaughter is the least severely
punished category of murder.  A
greater percentage of the family-murder 
defendants had voluntary or nonnegligent
manslaughter as their most serious
conviction offense.

-- Having a prior arrest or conviction is 
associated with receiving a prison
sentence and with longer terms of
confinement.  A smaller percentage of
family-murder defendants had such a
criminal history.

-- Convicted murder defendants under
age 30 were less likely to be
sentenced to probation.  A smaller
percentage of family-murder
defendants were younger than 30.

-- Family murder convictions were less
likely than nonfamily murder convictions 
to have resulted from a trial.  Conviction 
based on trial rather than guilty plea 
increases the likelihood of any of the 
more severe sentences (life or long 
prison term).

Methodology

Sample selection

The 33 counties in the sample were
selected to be representative of the
Nation's 75 largest counties.  The
ranking of counties in which the 75
largest were identified was based on a
combination of crime data (1980 and
1984 Uniform Crime Report Part I
arrests) and population data (1980
population from the Census Bureau's
City County Data Book).  Rankings
reflected the size of the prosecutors'
offices.  The original sample plan
identified 34 counties, 1 of which
declined to participate.

The following are the 33 counties
whose prosecutors' offices participated
in the study reported here:  

Arizona: Pima 
California: Los Angeles, Orange, Kern, San Diego, Riverside 
Colorado: Denver, Arapahoe 
Connecticut: New Haven 
Florida: Dade, Orange, Broward 
Illinois: Cook 
Louisiana: Orleans 
Maryland:  Prince Georges 
Massachusetts:  Middlesex 
Michigan: Wayne
Missouri: St. Louis
New Mexico: Bernalillo
New York: Kings, Monroe, New York, Queens
Ohio: Franklin, Montgomery
Oklahoma: Oklahoma
Pennsylvania: Philadelphia, Allegheny
Tennessee: Shelby
Texas: Dallas, Tarrant, Travis
Washington: King

A total of 2,539 murder cases were
sampled.  These cases were a sample
of about half of all murder-charge
cases disposed in the sampled 33
counties in 1988.  Not eligible for
sample selection were nonmurder
defendants or any whose most serious
charge was attempted murder, negligent 
or involuntary manslaughter, or vehicular 
homicide.  In counties with 200 or fewer 
disposed murder cases in 1988, all were 
selected for inclusion in the sample. In 
counties with more than 200, a systematic 
sample of 200 was chosen. Only 6 of the 
33 counties had more than 200 murder 
cases. 

Virtually all cases meeting the 1988-
disposition criterion were disposed for 
all defendants in the case.  Of the more 
than 3,100 defendants on whom data were 
obtained, only 13 had not yet had their 
cases adjudicated at the time the survey 
was carried out in 1990. Another 25 
defendants had died of suicide or other 
causes.

Nonavailability of cases

The survey goal to track murder cases
across justice system stages was not
met in nine counties.  In one of the
nine, legal restrictions barred access
to cases rejected by the prosecutor. 
In the remaining eight counties, some
of the sampled cases could not be
located.  

To an unknown degree, these data 
access problems help explain why no 
case from the nine was coded as
"rejected by prosecutor."  Though
there is no reason for all of the
unavailable cases to be rejections in
all nine counties, assuming that all
such cases were rejections results in
an estimated rejection rate of 12%,
instead of 8%, as shown in table 11. 
Other outcomes would have been
similarly affected.  The percentage of
defendants tried and convicted would
have been 33% instead of 34%; the
percentage pleading guilty would have
been 37% instead of 39%; and the
percentage receiving an incarceration
sentence of more than 1 year would
have been 62% instead of 65%.  

Computation of estimates 
from sample data

Case weights were applied to statistics
on the sampled cases to expand them 
to estimates for the universe of the 
75 largest counties, the key assumption
being that cases not sampled were
similar to the cases sampled.  A case
weight was the inverse of the
probability that a case would be in the
survey.  That probability was the
product of both the probability that 
a given county would be chosen and the
probability of selection of that case in
that county.  Case weights were adjusted 
to compensate for the loss of one 
nonparticipating county.

Statistically weighted, the 3,119
defendants in the sample cases
represented 9,576 murder defendants
in the Nation's 75 largest counties. 
The 2,655 victims represented 8,063
victims in the 75 largest counties.

Coding of circumstances 
and victim-killer relationships

Information about a murder case
usually included details about the
relationship between the victim and
the defendant as well as the
circumstances that existed at the time
of the murder.  The rules for
describing relationships and
circumstances were those used by
local police in reporting murder cases
to the FBI.  These rules were
developed by the FBI for publication of
its Supplemental Homicide Reports. 
The reporting rules include a set of
codes to describe the principal
victim-assailant relationship and the
circumstances in which they were
involved at the time the murder
occurred.  

In the survey reported here, however,
provision was made for coding as
many as three kinds of relationships
and three kinds of circumstances.  For
example, if the victim was the
assailant's brother and was also the
assailant's drug supplier, both a family
relationship and a drug relationship
would be recorded.  Likewise, more
than one type of circumstance might
have existed at the time of the murder. 
Altogether, 79 separate relationship
codes and 85 circumstance codes
were available for coding cases.

Among all pairs of victims and
assailants found in the prosecutors'
murder files, a majority required only 
a single relationship or circumstance
code.  The percentages requiring
more are as follows:

Response rates

Except as noted below, this report
focused exclusively on characteristics
that were successfully obtained in a
high percentage of sample cases
("response rate").

The case records identified age, race, 
sex, and ethnicity for nearly all
defendants (approximately 98%).  The
same was true of victims, except that
the victim's age was available in only
16% of cases.  Also obtained in nearly
all cases were the relationships
between victims and defendants; the
circumstances preceding the
homicide; the arrest or indictment
charge; and whether the defendant
was convicted, and if so, the
conviction offense.  In incarceration or
probation cases, the length of the term
of sentence was usually known.

Defendant criminal history was
available in three-quarters of the
cases, but victim criminal history was
obtained in only a third of the cases. 
The juvenile portion of the criminal
history information was probably less
complete than the adult portion.

Victim and offender information
compiled on both drug use at the time
of the offense and drug use history
were not presented in the report
because of concerns about data
credibility.  Drug use, for example, 
was far below what previous surveys 
have documented.  Also, the data show
drug use to have been more common
among victims than defendants, a
finding that did not seem believable. 
Consequently, survey data on drug
consumption and type of drug
consumed were not used.

Source of percentages of defendants
who had a criminal record and whose
victim also had a criminal record

Overall --
44% of victims had a criminal record 
(from table 8:  36.9% + 7.5%).
    83% of these victims were killed 
    by someone with a prior record 
    (36.9/44.4).

Family murders --
21% of victims had a criminal record
(14.3% + 6.4%).
   69% of these victims were killed 
   by someone with a prior record 
   (14.3/20.7).

78% of victims had no criminal record
(45.7% + 33.6%).
   43% of these victims were killed 
   by someone with a prior record 
   (33.6/78.3).

Comparison with other BJS murder
data collections

Selected data reported here can be
compared with other BJS publications
that contain information on murder
cases.

Conviction rate

The 73% rate of conviction reported in
table 12 is higher than the 66%
reported for murder defendants in the
National Pretrial Reporting Program
(NPRP).  See table 13 in Felony
Defendants in Large Urban Counties,
1988, BJS, NCJ-122385, April 1990. 
The NPRP studied a sample 
of felony cases obtained from court
records in 40 of the 75 largest
counties in the Nation.  Those cases
were followed to disposition or for up
to a maximum of 1 year. 

The following two reports give data for
cases accepted by the prosecutor. 
The comparable conviction rate in this
report would be 79%, rather than 73%,
after deducting the cases rejected for
prosecution.

The Offender-Based Transaction
Statistics (OBTS) program reported a
76% conviction rate among murder
cases that were prosecuted in 14
States.  See table 4 in Tracking
Offenders, 1988, BJS Bulletin, NCJ
29861, June 1991.  The OBTS uses
arrest records, disposition information,
and data from fingerprint cards that
are submitted by local law
enforcement agencies to State
criminal information repositories.  This
76% conviction rate is not measurably
different than the 79% estimate in this
report.

Conviction rates of murder cases filed 
in court are reported for a selection of
10 counties in table 2 in The
Prosecution of Felony Arrests, 1988,
BJS, NCJ-130914, February 1992. 
The local prosecutors in those 10
counties provided the data.  The rates
in those counties among murder cases
disposed during 1988 ranged from
57% to 84%.  Four of ten had rates
higher than the 79% reported here.

Number of murder convictions

Table 12 showed 63% of murder defendants 
convicted of murder, for a total of 
approximately 6,000 convictions.  The 
comparable number in the National Judicial 
Reporting Program (NJRP) for the 75 largest
counties in the United States during
1988 is approximately 5,000, which is
not measurably different than the 6,000 
estimate reported here.  See table 2.1a in 
National Judicial Reporting Program, 1988, 
BJS, NCJ-135945, December 1992. However, the 
63% of defendants convicted of murder is 
higher than the comparable 48% reported by 
the NPRP.

Sentences to prison, jail, or probation

The NJRP and NPRP reports include
the sentences received by those
convicted of murder, comparable to
table 13 of this report.  All three
studies show that of such defendants,
more than 90% were sentenced to a
prison term, fewer than about 5% were
sentenced to jail, and about 3% were
sentenced to probation without any
incarceration.  The OBTS, however,
reported these percentages as 81%,
11%, and 5% respectively.  

Standard errors

Data collected in this murder study
were from a sample rather than a
complete enumeration.  Because
counties and cases were sampled, a
sampling error (standard error) is
associated with each number in the
report.  In general, if the difference
between two numbers is greater than
twice the standard error for that
difference, we can say that we are at
least 95% confident that the two
numbers are in fact different; that is,
the apparent difference is not simply
the result of surveying a sample rather
than the entire population.  

Similarly, if the difference between two
numbers is greater than 1.6 standard
errors, we are at least 90% confident
that the two numbers are different.  In
this report the term "statistically
significant" was used to denote a
difference in which we have at least
90% confidence.  Except where
explicitly indicated otherwise, all
differences discussed in this report
had a confidence level at or above 90
percent.  
 
Why a standard error may be large
relative to the difference it pertains 
to includes the two most typical reasons:
(1) the measurements or observations
being compared (such as a sex difference 
in average prison sentence length) are 
highly variable from one case to another 
and (2) a small sample size.

Standard errors for selected key
variables in the report are presented
below.

Related reading  

Data used in this report were previously 
used in the report by John M. Dawson and 
Barbara Boland (Murder in Large Urban 
Counties, 1988, BJS Special Report, NCJ-
140614, May 1993).

The data presented in this report may 
be obtained from the National Archive 
of Criminal Justice Data at the
University of Michigan, 1-800-999-0960.  
The name of the data set is Murder in Large 
Urban Counties, 1988 (ICPSR 9907).  The
data are available in either dBASE or
SAS format.

--------------------------------------
John M. Dawson and Patrick A.
Langan, Ph.D., wrote this report. 
Dawson monitored the Prosecution
of Felony Arrests project, through
which the report's data were
collected.  Peter Brien verified the
data.  Tom Hester edited and
produced it, assisted by Priscilla
Middletion.  Marilyn Marbrook,
assisted by Jayne Robinson, Jodi
Brown, and Priscilla Middleton,
administered the final production.

Abt Associates personnel for the
data collection phase were as
follows:  Barbara Boland (survey
design), Jan Chaiken (sampling
plan), Wayne Logan (case
classification scheme and data form
design), Lynn Warner (coordination
of field operations), Marcia Schieck
and Mark Searight (field data
collection), and Bill Martin
(database design).

July 1994, NCJ-143498
------------------------------------------

End of report
11/22/02 ih