AN EMPIRICAL ANALYSIS OF MARYLAND’S DEATH
SENTENCING SYSTEM WITH RESPECT TO THE
INFLUENCE OF RACE AND LEGAL JURISDICTION
FINAL REPORT
Professor Raymond Paternoster
University of Maryland
Professor Robert Brame
University of South Carolina
Sarah Bacon
Andrew Ditchfield
David Biere
Karen Beckman
Deanna Perez
Michael Strauch
Nadine Frederique
And
Kristin Gawkoski
Daniel Zeigler
Katheryn Murphy
The authors would like to acknowledge the assistance of the following people.
Without their help this study never would have been completed.
Research Staff
Sarah Bacon, Andrew Ditchfield, Karen Beckman, David Biere, Deanna Perez,
Michael Strauch, Nadine Frederique, Kristin Powkowski, Daniel Stein, and
Katheryn Murphy.
University of Maryland
Professor Charles F. Wellford, Chair of the Department of Criminology and
Criminal Justice.
Dr. Irwin Goldstein, Dean of the College of Behavioral and Social Sciences.
Members of the Steering Committee
The Honorable Bishop Robinson ( Chair)
Court of Appeals of the State of Maryland
Maryland Division of Probation and Parole
Maryland Division of Correction
State’s Attorneys in Maryland
Others
Sue Schenning, Baltimore County State’s Attorneys Office
Katy O’Donnell, State of Maryland Office of the Public Defender, Capital
Defense Division
Professor David Baldus, Joseph B. Tye Professor of Law, University of Iowa
College of Law
Professor Michael Maltz, University of Illinois- Chicago Circle
Professor David Weisburd, Hebrew University and University of Maryland
Professor Jerome Deitz, School of Law, University of Maryland
THE ADMINISTRATION OF THE DEATH PENALTY
IN MARYLAND: 1978- 1999
INTRODUCTION
There have been suggestions in the past that the imposition of the death penalty in
Maryland has been influenced by factors such as race and the particular legal jurisdiction
where the homicide occurred. Critics have pointed to the racial composition of
Maryland’s death row as evidence that there is racial disparity in the operation of the
death penalty in the state. For example, in December of 2002 all thirteen men on
Maryland’s death row were sentenced to death for killing whites and in eight of these
thirteen ( 62%) the offender was black. 1 In fact, since 1978, when the state’s new death
penalty statute took effect, there have been no fewer than four investigations into the
administration of the death penalty in Maryland, with at least some emphasis on the
issues of racial disparity and arbitrariness ( geographic disparity).
In 1987, for example, at the request of the Maryland Court of Appeals, the
Maryland Office of the Public Defender collected some preliminary information on death
sentencing patterns in the state from 1978- 1987.2 The authors of this study identified 415
homicides that were deemed to be “ death eligible” ( they presented facts which legally
qualified them for the death penalty), 90 of which resulted in a penalty phase hearing.
They found substantial county- to- county variation in the rate at which state’s attorneys
file death notices. They also reported that state’s attorneys were approximately twice as
1 Death Row U. S. A. Fall 2002.
2 Capital Punishment in Maryland 1978- 1987: A Report by the Maryland Public Defender on the
Administration of Capital Punishment.
2
likely to file a notification to seek a death sentence and not withdraw that notification
when the homicide victim was white rather than black. This report did not, however,
consider all possible death eligible cases such as those that resulted in second degree
murder convictions. More importantly, the study did not consider the numerous
characteristics about a homicide ( the number of aggravating factors, the criminal history
of the defendant, etc.) which may possibly explain any apparent racial or geographic
disparity.
In 1993 the Governor’s Commission on the Death Penalty in Maryland published
its report on the administration of capital punishment in the state from 1978 to 1993.3
This study examined death sentences that were actually imposed, penalty phase hearings
that resulted in a life or a life without parole sentence, and death notifications that were
filed but subsequently withdrawn. There was no data on such case characteristics as non-statutory
aggravators and mitigators or the criminal history of the defendant, nor did the
Commission examine all possible death eligible cases and how they are filtered through
the sentencing system. Given the data inadequacies, it is not surprising that the
conclusions were a bit ambiguous. One of the Commission’s findings ( Chapter VII,
finding # 10) was that “ the data does not establish discrimination against African
American defendants or in favor of white victims; neither does the data disprove racial
discrimination”. The Commission report ultimately concluded that “ there is no evidence
of intentional discrimination in the implementation of the death penalty in Maryland, but
racial disparities in its implementation remain a matter of legitimate concern”.
3 The Report of the Governor’s Commission on the Death Penalty, An Examination of Capital Punishment
in Maryland: 1878- 1993.
3
In 1996 the Task Force on the Fair Imposition of Capital Punishment was created
to specifically examine the issue of racial discrimination in the administration of the
death penalty in Maryland. 4 The Task Force did no original data collection or analysis,
and it limited its examination to the racial composition of Maryland’s current death row.
It observed that “[ t] he high percentage of African- American prisoners under sentence of
death and the low percentage of prisoners under sentence of death whose victims were
African- American remains a cause for concern”. 5 The basis of this conclusion was the
finding that of the seventeen condemned persons then on Maryland’s death row, fourteen
( 82%) were African American and the victims of the homicides included sixteen whites
and 6 African- Americans. The Task Force recommended a more comprehensive
empirical study of Maryland’s capital sentencing system.
Finally, in February of 2001, Professors David Baldus and George Woodworth of
the University of Iowa conducted an analysis of race disparities among 346 Maryland
first degree homicide cases where the state served notice of its intention to seek the death
penalty. 6 They found that even when considering the number of statutory aggravating
factors charged, defendants who killed white victims were more likely to advance to a
penalty trial and are more likely to be sentenced to death than those who killed a black.
This was particularly true for black offenders who killed white victims. Baldus and
Woodworth acknowledged the two central limitations of their study: ( 1) their sample did
not include all possible death eligible cases, and ( 2) they had limited information on the
4 Task Force on the Fair Imposition of the Death Penalty.
5 Task Force Report at p. 39.
6 Baldus, David C. and George Woodworth ( 2001) Race of Victim and Race of Defendant Disparities in
the Administration of Maryland’s Capital Charging and Sentencing System ( 1979- 1996): Preliminary
Finding.
4
non- statutory aggravating and mitigating factors in the case and other case characteristics.
With these limitations in mind, Baldus and Woodworth concluded that “ although our
preliminary finding may be construed as supportive of the disparate treatment hypothesis,
a definitive judgment on the issue must await the results of a study that has better controls
for case severity and defendant culpability than the preliminary results reported in this
report.” 7
There have, then, been four previous examinations into the possibility that there
are disparities ( either by race or geography) in the administration of the death penalty in
Maryland. Each of these previous attempts have been hampered by the fact that they did
not examine all homicides where death could have been requested, and they all have
failed to collect detailed case information about possible aggravating and mitigating
factors and other relevant offense and offender characteristics. As a result, there is too
little empirical information upon which to base a conclusion about the fair and even
handed imposition of the death penalty in the state.
In September of 2000, Maryland Governor Parris N. Glendening commissioned an
empirical study of the death penalty in the state of Maryland, and subsequently imposed a
moratorium on all executions in the state until the completion of the study. That study
was to examine whether or not the imposition of the death penalty in the state was
affected by race ( either of the offender, victim or both) or geography ( the jurisdiction
where the crime occurred). The influence of race and geography was to be examined at
four critical decision making points in the administration of Maryland’s capital
punishment system:
7 Baldus and Woodworth report at page 12.
5
1. the decision of the state’s attorney to file a formal notification to seek a death
sentence.
2. the decision of the state’s attorney to not withdraw a death notification once filed,
in other words, the decision to make the death notification “ stick”.
3. the decision of the state’s attorney to advance a death- eligible offense to a penalty
trial upon a conviction for first degree murder.
4. the decision of the jury or judge to sentence a defendant to death. 8
The key task of this research is to estimate the effect that race and geography has
on these four decision points while taking into account numerous case characteristics that
may explain these decisions. That is, there are numerous factors that affect these decision
making points, such as the criminal history of the offender, the number of victims, the
brutality of the murder, which must be considered when examining the effect of race and
geography.
In this report, we will first provide a brief description of the recent history of capital
punishment in the state of Maryland, including concerns about the fairness with which it
has been imposed in the past. We will then briefly describe the legal structure or
mechanics of the death penalty under Maryland law. We will then briefly discuss the
methodology we followed in this empirical study of the death penalty in Maryland, with
particular attention devoted to describing how we characterized an offense as “ death
eligible”, and the statistical strategy we followed in determining the influence of race and
geography on the four decision making points. This will be followed by a detailed
8 We do not look at the pre- prosecutorial aspects of the case, which may include interrogation,
arrest, pretrial hearings, etc. , or a decision to charge for a lesser offense such as manslaughter.
These decisions do have a filtering effect on the cases that make it to the state’s attorney’s office.
6
presentation of our results – what we found with respect to the administration of the death
penalty. In the final section of the report we summarize these findings.
A RECENT HISTORY OF THE DEATH PENALTY IN MARYLAND
In 1972, the United States Supreme Court determined that the then- existing
procedures for imposing capital punishment in two states ( Georgia and Texas) that gave
juries unlimited and standardless discretion were unconstitutional [ Furman v. Georgia
403 U. S. 238 ( 1972)]. Although the Court did not speak with one voice in its decision, it
was clear that the then existing manner in which death sentences were imposed created a
risk that defendants would be sentenced to death on the basis of constitutionally suspect
factors ( that the capital sentencing system was discriminatory) or that because the capital
sentencing system was arbitrary and capricious, there was no rational and meaningful
basis to distinguish offenders sentenced to death and those whose life was spared.
The practical impact of the Furman decision was to call into question the
constitutionality of death penalty statutes in other states, including Maryland. According
to the Maryland statute at the time, in deciding which rape and murder defendants to
sentence to death, juries were provided no guidance or standards and upon conviction a
death sentence was mandatory unless the jury specifically stated in its guilty verdict
“ without capital punishment”. On December 4, 1972 the Maryland Court of Appeals
invalidated the state’s death penalty statute on the basis of the Furman decision in
Bartholmey v. State 267 Md. 175, 297 A. 2d 696 ( 1972), and twenty- three defendants on
Maryland’s death row had their sentences vacated to life imprisonment.
7
The Maryland legislature constructed a new death penalty statute that was
intended to remedy the defect of unguided or standardless discretion identified by the
Furman Court. This new statute remedied the problem of discretion by creating eight
narrowly defined categories of first degree murder. A sentence of death was mandatory
upon the conviction of one of these categories of capital murder. This statute essentially
made the death penalty mandatory for every first degree murder where a statutory
aggravating circumstance was found. This new statute took effect on July 1, 1975. In July
of the next year, the United States Supreme Court upheld the constitutionality of capital
statutes that structured and guided the discretion of capital juries [ Gregg v. Georgia 428
U. S. 153 ( 1976); Profitt v. Florida 428 U. S. 242 ( 1976); Jurek v. Texas 428 U. S. 262
( 1976)], but struck down mandatory statutes [ Woodson v. North Carolina 428 U. S. 280
( 1976); Roberts v. Louisiana 428 U. S. 325 ( 1976)]. The problem with mandatory statutes
identified by the Court was that they failed to allow the capital decision maker to
consider the unique culpability of individual defendants and so treated them “ as members
of a faceless, undifferentiated mass to be subjected to the blind infliction of the penalty of
death” ( Woodson v. North Carolina 428 U. S. 280 1976).
Maryland once again was confronted with the problem that a key feature of its
capital punishment statute had been declared unconstitutional. Recognizing this, in
Blackwell v. State 278 Md. 466, 365 A. 2d 545 ( 1976) the Maryland Court of Appeals
vacated the two death sentences that had been handed down under the 1975 statute. The
Maryland legislature responded to the invalidation of its mandatory statute by
constructing a guided discretion statute in its 1977 and 1978 sessions. This new guided
discretion statute ( to be described in the following section) became effective on July 1,
8
1978, and provides the foundation for Maryland’s current capital punishment law. In
August of 1979 the first death sentence under the state’s new law was imposed on
Richard Danny Tichnell in Wicomico County.
MARYLAND’S CAPITAL STATUTE
Under Maryland’s death penalty statute, in order for a defendant to be sentenced
to death, the following facts must exist:
1. After 1987, the defendant must be 18 years or older at the time of the offense.
2. After 1989, the defendant cannot be mentally retarded, with mental retardation
established at the penalty phase of the capital trial by a preponderance of the
evidence. A jury finding of mental retardation must be unanimous.
3. The defendant must have been convicted of first degree murder and have been
found to be a principal in the first degree ( the defendant must have committed the
murder with his/ her own hands or employed another who committed the murder).
The “ principalship” requirement means that one who is eligible for the death
penalty must, therefore, be the actual killer or the one who pays the killer. A jury
finding of principalship must be unanimous.
4. The state’s attorney prosecuting the case must have notified defense counsel at
least 30 days in advance of the trial that the state intends to seek a death sentence
( or a sentence of life without parole) and the specific aggravating factors that the
state intends to rely on.
9
5. At least one statutory aggravating factor must have been found by the fact finder
beyond a reasonable doubt. 9
There are ten statutory aggravating circumstances under Maryland law: 10
A1: The victim of the murder was a law enforcement officer in the performance
of his/ her duties.
A2: The defendant committed the murder when confined in a correctional
institution.
A3: The defendant committed the murder while trying to escape from custody.
A4: The victim was taken in the course of a kidnapping or abduction.
A5: The victim was a child abductee.
A6: The defendant murdered pursuant to an agreement for remuneration.
A7: The defendant employed another who killed for remuneration.
A8: The defendant committed murder when under sentence of death or life
imprisonment.
A9: The same incident produced multiple murder victims.
A10: The defendant committed the murder while committing, or attempting to
commit, a carjacking, armed carjacking, robbery, arson in the first degree, rape or
sexual offense in the first degree.
The presence of at least one of these statutory aggravators is necessary to make a
defendant eligible for the death penalty, but their existence does not require the state’s
attorney to seek a death sentence.
9 Art. 27 , § . 413, Annotated Code of Maryland.
10 See Art. 27, § . 413( d), Annotated Code of Maryland.
10
State’s attorneys have the discretion not to seek a death sentence even if the facts
warrant it. Moreover, state’s attorneys have the discretion to withdraw a notification to
seek a death sentence once filed either unilaterally or in exchange for a plea from a
defendant. Futher, they have the discretion as to whether or not to advance a case to a
penalty hearing upon conviction of a charge of capital murder. Even if a death sentence is
sought by the state, and the case is advanced to a penalty hearing, the sentencer has the
discretion not to impose a death sentence if it feels that capital punishment is not
warranted in a particular case. According to Maryland law, the sentencing body in a
capital case must find at least one statutory aggravating circumstance before it may
consider a death sentence. If it does find at least one aggravating circumstance beyond a
reasonable doubt, and determines that the defendant is eligible for the death penalty, it
must then determine if there are mitigating circumstances in the case. There are eight
mitigating factors enumerated in the Maryland statute that the jury must consider: 11
M1: The defendant has not previously been convicted of a crime of violence.
M2: The victim participated in the defendant’s conduct or consented to the act
which caused the victim’s death.
M3: The defendant acted under substantial duress, domination, or provocation,
but not so substantial as to constitute a complete defense to the prosecution.
M4: The defendant’s capacity to appreciate the criminality of his conduct or to
conform his conduct to the requirements of the law was substantially impaired as
a result of mental incapacity, mental disorder, or emotional disturbance.
M5: The youthful age of the defendant at the time of the crime.
11 See Art. 27 § 413( g).
11
M6: The act of the defendant was not the sole proximate cause of the victim’s
death.
M7: It is unlikely that the defendant will engage in further criminal activity that
would constitute a continuing threat to society.
M8: Any other facts which the jury or the court specifically sets forth in writing
that it finds as mitigating circumstances in the case.
With respect to any mitigating circumstances offered by the defense, each individual
juror must determine for him/ herself if it has been proven with a preponderance of the
evidence. If there are no mitigating factors found, the presumptive sentence is death. 12 If
at least one mitigating circumstance is found, then the sentencer must weigh the
aggravating circumstances against the mitigating. To impose a sentence of death, the
aggravating circumstances must be found to “ outweigh” the mitigating circumstances by
“ a preponderance of the evidence”. If the aggravating circumstances do not outweigh the
mitigating circumstances by a preponderance of the evidence, a death sentence may not
be imposed. If the State had advised the defendant prior to trial that it would seek a life
without parole sentence, then the court or jury must decide if the convicted defendant is
to be sentenced to a regular life term or life without parole. Any death sentence imposed
in the state of Maryland is then subject to automatic appellate review by the Maryland
Court of Appeals. This initial review may not be waived by a defendant.
METHODOLOGY OF THE CURRENT STUDY
12 The Maryland statute is silent with respect to the mandatory nature of the death penalty when there are
aggravating circumstances found and no factors in mitigation. The Maryland Court of Appeals has,
however, interpreted the statute as requiring the judge or jury to impose a sentence of death in the presence
of aggravators and no mitigators, see, Scott v. State 529 A. 2d 340 ( Md. 1987).
12
This report is based upon an examination of approximately 6,000 first and
second degree murders that were committed in the state of Maryland from August of
1978 ( when the state’s new death penalty law took effect) until September of 1999. 13 An
initial list of all first and second degree murders was obtained from the Maryland
Division of Corrections Research Office. They produced a computer- generated list of all
convicted first and second degree murderers sentenced to any Maryland correctional
institution during the 1978- 1999 time period. The Maryland Division of Corrections
made available the inmate’s institutional record or file, which had much of the detailed
kind of information needed to characterize the murder, the defendant, and frequently had
information on the victim as well. 14 This Division of Corrections file had a great deal of
useful information such as the defendant’s criminal history; frequently it included the
presentence report which contained information about the defendant’s educational, social,
employment, and mental health history, information about the victim, and generally a
richly detailed description of the offense, crime scene, and some information about the
type of evidence ( if there was physical evidence available or an eye- witness). This
information was transcribed onto our initial data collection instrument, the Maryland
Screening Instrument ( MDSI), a copy of which is provided in Appendix 1. From the
information in this document we were able to determine for most of the cases whether or
not it was “ death eligible”. For those cases that were deemed death eligible, additional,
far more extensive information was collected on each case and transcribed onto the
13 The offense dates cover the period from August 1, 1978 until September 25, 1999.
14 The senior investigator entered into a research agreement with the Maryland Division of Corrections
insuring confidentiality with respect to the information extracted from the inmate files. A similar research
agreement was signed with the Maryland Division of Probation and Parole with specific reference to access
to presentence reports and a guarantee of confidentiality of information.
13
primary data collection instrument, the Maryland Data Collection Instrument ( MDCI), a
copy of which is provided in Appendix 2.
The list produced by the Division of Corrections, and access to inmate files were
important sources of information, but not the only sources relied on. For cases that both
were death eligible and had advanced to a penalty phase hearing, we examined the court
transcript and the trial judge’s report on file with the Clerk of the Maryland Court of
Appeals. An additional source of information was the file on each case in the office of the
state’s attorney for the twenty- three Maryland counties and Baltimore City. Since the
Division of Corrections list contained only those cases that resulted in a conviction, we
sought the assistance of each of the state’s attorneys to both correct and supplement our
original list of cases. We sent a list of homicide cases that the Division of Corrections had
identified as coming from that county to each of the twenty- four state’s attorneys and
asked them to verify that the list was correct, and to provide names of murder defendants
from their county who where not on the list provided them but who had been convicted
during the study’s time period of first or second degree murder, or who were charged
with first or second degree murder and were subsequently acquitted or disposed of in
ways other than a conviction. We also asked and received access to their files on each
case. These state’s attorneys files provided a substantial amount of very rich information,
which included police reports and the state’s version of the case. We were also able to
verify information initially gotten from other sources. Information on the homicide victim
was also obtained from the victim’s death certificate obtained from the Maryland Office
of Public Health.
14
Since one of the decision points examined in this research is the decision of the
state’s attorney to seek a death sentence given that a murderer was eligible for the death
penalty, the first task was to determine which among the universe of murders was eligible
for the death penalty. Determining whether or not a murderer is death eligible is a
controversial issue. In one sense the only true way to categorize a murder as eligible for
the death penalty is if the state’s attorney determines that the case meets the all of
eligibility requirements as listed in the state statute:
1. The defendant was a principal in the first degree and the state could prove this
beyond a reasonable doubt,
2. The defendant was not mentally retarded at the time of the offense ( after May of
1989) and the state could prove this with a preponderance of the evidence,
3. The defendant was not less than 18 years old at the time of the offense ( after June
of 1987),
4. The murder also included at least one statutory aggravating circumstance and the
state could prove this beyond a reasonable doubt,
5. The state’s attorney files a notice 30 days prior to trial of the state’s intention to
seek a death sentence
and then a notification to seek a sentence of death is filed.
Clearly, a murder that meets these statutory eligibility requirements and is
followed by the state’s attorney also formally filing a notice to seek a death sentence, is
death eligible and should be treated as such. However, a homicide may also meet the first
four of these requirements but the state’s attorney decides for other reasons, not to seek a
death sentence. The reasons may include the potential cost of the case to the county, the
15
reluctance of the victim’s family to support a death sentence, a low probability that a jury
would return a death sentence. In this instance, a homicide that was technically “ death
eligible” would not be followed by a decision to seek a death sentence, and the case
would be handled as a non- capital homicide. In addition, different state’s attorneys in
different offices ( or even different state’s attorneys within the same office) may evaluate
a case as to its death eligibility and come to a different conclusion. The issue of first-degree
principalship, for instance, is not always easy to determine and different state’s
attorneys may disagree as to whether or not principalship exists, or if it does exist,
whether it could be proven beyond a reasonable doubt. Similar ambiguity may exist with
respect to the presence in a murder of a statutory aggravating circumstance, and if could
be proven beyond a reasonable doubt. There are no automatic or indisputable answers to
these issues; they are inherently ambiguous.
Some murder cases, therefore, may be “ death eligible” in the sense that they meet
all of the statutory criteria for death eligibility, and yet the formal decision maker does
not treat the case as such. If one is interested, as we are here, in examining the factors that
explain the state’s attorneys’ decision to seek death in some death eligible cases but not
others, there is a need to define a death eligible case in ways other than the filing of a
formal notice to seek death. We proceeded with caution, however, because the issue as to
whether or not a murder is death eligible involves a great deal of ambiguity and
inevitable controversy.
For the purpose of this research, a case was deemed to be death eligible if:
1. the state’s attorney filed a notice of an intention to seek a death sentence, even if
that notice was later withdrawn unilaterally or in exchange for a plea.
16
2. the facts of the case clearly established that a first degree murder was committed,
the defendant was the principal in the first degree, the defendant was eligible by
age at the time of the offense, the defendant was not mentally retarded at the time
of the offense, and the murder included at least one statutory aggravating
circumstance.
From an initial pool of approximately 6,000 homicides, these two criteria produced a
universe of 1,311 death eligible cases. Initially included in the universe of 6,000
homicides was a pool of approximately 300 homicide cases where the state’s attorney did
not file a notification to seek a death sentence and the issue of death eligibility based
upon the available facts of the case was ambiguous. In these cases it was not clear if the
defendant was the principal in the first degree in the killing, or if there was a statutory
aggravating circumstance present to make the homicide death eligible. In order to
determine death eligibility in these cases we submitted them to a panel of attorneys who
had some experience in death penalty cases. The panel was put together by the senior
researcher after consultation with Ms. Sue Schenning , the Deputy State’s Attorney for
Baltimore County, and Ms. Katy O’Donnell, of the State Office of the Public Defender.
The Panel consisted of a roughly equal number of state’s attorneys, public defenders, and
private lawyers who have handled death penalty cases as former prosecutors, public
defenders or as private defense counsel. Panel members read narratives of a homicide in
question that presented the facts of the case, and were asked to make two determinations:
1. do you think this case is “ death eligible” under Maryland law? With response
options, “ yes” or “ no”.
17
2. on a scale from 1 (“ not very confident at all”) to 10 (“ very confident”) how
confident do you feel in making this determination?
Each case was read and rated by a group of panel attorneys. 15 We included as death
eligible those homicides where a majority of panel attorneys rated the case as death
eligible and where the confidence of the rating averaged 5.0 or higher. In other words, a
majority of the reviewing panel had to rate the case as death eligible and they had to state
that they were at least moderately confident in making that assessment. Out of the
approximately 300 cases reviewed by the panel of attorneys, fewer than fifty were
determined to be death eligible and were added to the pool of death eligible cases.
Our universe of cases includes the 1,311 death eligible cases in Maryland from
July 1, 1978 to December 31, 1999, and we illustrate the number of cases at each
decision point in Figure 1. Out of these 1,311 death eligible cases, state’s attorneys filed
a formal notification to seek the death penalty in 353 ( 27%). Although 353 notifications
to seek death were filed, 140 ( 40%) were withdrawn by the state’s attorneys either
unilaterally or ( most often) with a plea by the defendant. A death notification was
retained or “ stuck” in 213 ( 60%) cases. Out of these 213 cases where a death notification
was filed and retained, 180 ( 84%) were advanced to a penalty trial. A case would not be
advanced to a penalty trial for a number of reasons: the state determines that it is unlikely
to obtain a death sentence and unilaterally decides not to advance a case, there were no
aggravators found during the guilt phase, or the defendant was found not to have been a
principal in the first degree. From 180 penalty trials, a death sentence was obtained in 76
cases ( 42%). The conditional probability of a death sentence given that a case was death
15 For each case reviewed, the number of panel attorneys ranged from 5 to 10.
18
eligible was .058 ( 76/ 1,311), the conditional probability of a death sentence given that a
death notification has been filed and “ sticks” was .357.
Although the substantive issue in this report is the effect of race and geography, each
of the four decision making points in the Maryland capital sentencing system is
influenced by many homicide case characteristics ( the number of statutory and non-statutory
aggravating circumstances, the presence of factors in mitigation, etc.); case
characteristics that we have captured in our data collection instrument ( the MDCI). The
key task of this report is to examine whether race or geography has any material affect on
each of these decision points after carefully considering or “ controlling for” these case
characteristics. In other words, we will examine the role that race and geography may
play at four critical points in the Maryland capital sentencing system while
simultaneously considering important features of a case that make it more or less
deserving of a capital charge, a penalty phase hearing, or a death sentence. In trying to
determine the impact of factors such as race and geography, it is critical that these
numerous case characteristics be considered. The reason is that such characteristics are
inevitably confounded both with the variables of substantive interest ( race and
geography) and with the outcome variable we are interested in ( the decision making
points we focus on).
Suppose, for example, there is a finding that black offenders are treated more
severely than white offenders at some decision point in the capital sentencing process.
We may observe this race of offender disparity for one of two reasons: ( 1) black
offenders really are at a disadvantage and receive disparate treatment, or ( 2) black
offenders or the offenses committed by black offenders are more likely to have the
19
characteristics that incline prosecutors to seek the death penalty ( for example, they have
more extensive criminal histories, or commit their crimes in a more brutal/ egregious
manner). It is important, therefore, that as many characteristics or factors of a case be
considered or “ controlled” in order to distinguish the effect of race from the effect of
legally relevant factors that may be associated or correlated with race.
There is a direct analogy between estimating a race or geography effect and
determining the effect of smoking on lung cancer. People do not get randomly assigned
to smoke or not smoke. People who smoke, therefore, are systematically different than
those who do not – they may live more sedentary lives, they may have less healthy diets,
etc. When a researcher is looking at the effect of smoking on lung cancer, therefore, they
must also consider these other characteristics that distinguish smokers and non- smokers
besides the fact that they smoke. Our problem is the same, in order to isolate the effect of
race or geography, we need to identify and consider differences among offenders who
slay whites vs. non- whites, or who live in Montgomery vs. Prince George’s County. We
will do this by directly incorporating into our statistical model factors that we find to be
empirically related to the county ( and race) variables. A list of the factors used in this
procedure is provided in Table 9. There are 112 possible explanatory factors or case
characteristics that we use for the decision to file a notification to seek the death penalty.
For decisions beyond this ( decision to withdraw the notification, decision to advance a
case to the penalty phase, and the sentencing decision), we employ 11 additional case
characteristics based on the presence of statutory aggravating circumstances that are
shown at the bottom of Table 9.
20
Our procedure for incorporating these case characteristics was as follows. At each
decision point we first examined each case characteristic separately to see if it was related
to the county or race variable of interest. 16 Those factors that were significantly related
at the .05 level were retained for further analysis, those not meeting that criterion were
dropped. The variables that were retained at this first screening were then entered into a
full logistic regression model with the particular decision point as the outcome variable,
and a test for county and race differences conducted. The case characteristic variables
from the full model that were not significant at p < .10 were then dropped and a reduced
model estimated. The parameter estimates of the reduced model are reported in each table
in our results section discussed below. In every case the reduced model was not
significantly different from the full model. The parameter estimates reported in each table
are those from the reduced model.
RESULTS
UNADJUSTED ANALYSIS
We began our empirical examination by conducting basic descriptive analyses of
county and race patterns at each of the four key decision points in the Maryland death
penalty sentencing system: ( 1) the prosecutor’s decision to issue a notice of intention to
seek the death penalty to a death eligible defendant ( in 353 cases the prosecutor issued a
notice and in the remaining 958 cases no notice was filed); ( 2) the prosecutor’s decision
to retract or “ stick” with a death notice among the 353 noticed cases ( in 213 of these
cases the prosecutor stuck with the notice and in the remaining 140 the prosecutor
16 We pursued this multiple stage approach to the analysis rather than entering the more than 100 case
characteristics into our model in order to avoid the problem of “ overfittng” the data – having too many
variables in the statistical model relative to the number of observations.
21
retracted the death notice); ( 3) whether the case proceeds to a penalty trial among the 213
cases where the death notice sticks ( in 180 instances, the case advanced to a penalty trial
while in the remaining 33 it did not); ( 4) whether the court imposes a death sentence ( in
76 cases out of the 180 the court did impose a death sentence; in the remaining 104 it did
not). Figures 2- 5 and Tables 1- 4 present a number of key descriptive quantities
associated with the Maryland death penalty system.
Race
Basic descriptive information for the race of the offender is shown in Figure 2.
This figure shows that white offenders comprise about .24 of the pool of death eligible
cases, black offenders .74 and offenders of other races .02. The contribution of white
offenders increases slightly at the next stage, the decision to file a death notification,
where .34 of all offenders are white and the proportion of black offenders declines
slightly to .65. After the decision to file a notification to seek death, the proportion of
white and black offenders remains fairly constant to the end of the process ( death
sentencing).
Table 1 reports the unadjusted rate at which offender race groups are processed
through the Maryland death penalty system. The probability that a death notification will
be filed given a death eligible case is .24 for black offenders, and .37 for non- black
offenders ( over 90% of whom are white). At this first decision point, then, non- white
offenders are significantly more likely to have a death notice filed against them than
black offenders. At each subsequent stage of the process there are no significant
differences in the handling of black offender and non- black offender cases.
22
Figure 3 reports the proportion of white victim and black victim cases at each
stage of the death penalty system. White offenders comprise approximately 45% of all
death eligible cases. At each subsequent stage of the process, however, the proportion of
white victim cases increases. White victim homicides make up 65% of those where a
death notification is filed, 74% of the cases where a death notification “ sticks”, 77% of
the cases that are advanced to a penalty trial, and in 80% of the death sentences imposed
in Maryland during this period.
Table 2 reports the unadjusted rate at which white and non- white victim cases are
processed through the system. The probability that a state’s attorney will file a
notification to seek the death penalty in a death eligible case is .43 when there is at least
one white victim and .19 when there are no white victims. This difference is statistically
significant. State’s attorneys are more likely to retain a death notification once filed in
white victim cases compared with non- white victim cases (. 70 vs. .46), and this
difference is also statistically significant. The probability that a case will be advanced to a
penalty trial is also significantly higher in white victim (. 88) than non- white victim (. 75)
cases. There is no race of victim disparity when the decision is whether or not to sentence
someone to death given a penalty trial. Overall, however, the probability of a death
sentence given the fact that a case is death eligible is .093 higher in white victim
compared with non- white victim cases, a statistically significant difference. These
unadjusted figures suggest that the race of the victim appears to matter at least in the
early stages of the capital punishment system.
Figure 4 provides the distribution of combinations of offender’s and victim’s race
at various stages of the Maryland capital punishment system. There are two clear patterns
23
from this figure. The first is that the proportion of cases involving a black offender and a
white victim increases dramatically as you move further into the process. The other is
that the proportion of cases involving a black offender and a black victim consistently
declines. For example, black- on- white homicides comprise about .23 of all death eligible
cases, but .35 of those that produce a death notification, .40 of those where the death
notification “ sticks”, .43 of the penalty trials, and one- half of the death sentences
imposed. Black- on- black killings make up .48 of the death eligible cases, but only .28 of
the death notifications, .22 of the notifications that “ stick”, .19 of the penalty trials, and
only .18 of the 76 death sentences. The proportion of homicides involving white
offenders and white victims also increases at each successive stage of the process, but not
as dramatically as we observed for black- on- white killings.
Tables 3A to 3D report the rate at which various race of offender/ victim groups
are processed at each decision point. Table 3A shows that homicides involving white
offenders and white victims are significantly more likely than all other racial
combinations to result in a formal notification to seek the death penalty. White- on- white
killings are not treated differently from other cases at any other stage of the process.
Table 3B shows the rate at which black- on- black killings are processed. Compared with
the other racial groups, black offenders who kill blacks are significantly less likely to
have a death notification filed, and less likely to have the notification “ stick”. There is no
difference at the stage of advancing a case to a penalty trial or the rate of death
sentencing given a penalty trial. The probability of a death sentence in a death eligible
case is significantly lower for black- on- black killings, and this is because of the
differential treatment of these cases in the hands of prosecutors.
24
Table 3C reports the rate of processing of cases involving black offenders and
white victims. State’s attorneys are significantly more likely to file a formal notification
to seek the death penalty in black- on- white killings compared with other racial
combinations, and they are significantly more likely to make this notification “ stick”.
Black offenders who kill whites are not treated differently at the decision to advance a
case to a penalty trial, and the sentencing decision after a penalty trial. Black- on- white
killings that are death eligible are more likely to result in a death sentence, primarily
because of decisions made by state’s attorneys earlier in the process ( charging decision).
Table 3D reports the processing of white- on- black homicides. While whites who
kill blacks are significantly more likely to be death notified, than other racial
combinations, not much weight should be on these results since there were only 22 cases
involving white offenders and black victims. This diminishes to 3 at the penalty phase
and only 1 at the sentencing stage.
Geography
Figure 5 presents the distribution of cases for several Maryland jurisdictions at
each stage of the capital punishment process. Two things are striking. First, the
proportion of cases from Baltimore City declines substantially the further into the process
you go. Second, the proportion of cases from Baltimore County increases substantially.
Baltimore City homicides comprise .43 of all of the death eligible homicides, but only .10
of the death notifications, only .11 of the death notifications that “ stick”, .10 of the
penalty trials, and only .13 of the death sentences. Baltimore County homicides comprise
25
only .12 of all death eligible homicides but .28 of all death notifications, .39 of all
notifications that “ stick”, .42 of all penalty trials, and .45 of all death sentences.
Table 4 reports the processing of cases at each decision making point for these
same jurisdictions. There is statistically significant variation across these jurisdictions in
the probability that a death eligible case will result in a notification to seek a death
sentence. This probability is .65 for Baltimore County, .54 for Harford County, .38 for
Prince George’s County, .23 for Anne Arundel County, .19 for Montgomery County, .06
for Baltimore City, and .46 for other counties in Maryland. There is also statistically
significant jurisdictional variation in the rate at which death notifications once filed are
retained or withdrawn, from a high of .84 in Baltimore County to a low of .40 in Prince
George’s County. There is no significant variation by jurisdiction at the decision to
advance a case to a penalty trial or at the decision to sentence to death given a penalty
trial. There is statistically significant variation across the different jurisdictions in the
probability of a death sentence for all death eligible cases, due primarily to the way the
charging decisions are handled.
In sum, our unadjusted analysis would suggest the following:
1. white offenders are more likely to be death notified than non- white
offenders.
2. offenders who kill at least one white victim are more likely to be death
notified, more likely to have that notification “ stick”, and more likely to be
advanced to a penalty trial than cases without a white victim.
3. white offenders who kill whites are more likely to be death notified than
otherers.
26
4. black offenders who kill blacks are less likely to be death notified and
have that notification “ stick” than others/
5. black offenders who kill whites are more likely to be death notified and
have that notification “ stick”
6. there is substantial and significant variation in the way different state’s
attorneys in Maryland make the decision to file a notification to seek the
death penalty and whether or not that notification is withdrawn.
While we have found disparate treatment by race and geography in the processing of
cases in the Maryland death penalty system, this unadjusted analysis does not take into
account numerous facts/ circumstances about these homicides which may legitimately
explain this disparate treatment. We now proceed to examine what happens to this
evidence of disparate treatment by race and geography once case characteristics are taken
into account.
Adjusted Analysis
Professors David Baldus and George Woodworth of the University of Iowa have
previously examined offender and victim race data for the subset of death noticed cases
in Maryland. A key recommendation emerging from their report ( February 2001) was
that a study controlling for other variables in addition to the statutory aggravating factors
they examined would provide more definitive answers to questions about the Maryland
system. The mandate for the current study was to examine geographic as well as victim
and offender race disparities in Maryland after controlling for a wide variety of relevant
individual case characteristics ( those listed in Table 9). Unfortunately, there was no
27
information on some of the covariates for some of the cases. Consequently, it is
necessary to attain a balance between including as many cases as possible with as many
covariates as possible. It is not possible to optimize both of these quantities
simultaneously – an increase in the number of cases necessitates a loss of some of the
covariates and an increase in the number of covariates necessitates a loss of some of the
cases. After examining a variety of different possibilities, a list of variables with
complete data on 1,202 of the original 1,311 cases ( 91.7% of the original number) was
devised. A comparison of the notice rates and death sentence rates for the dropped cases
compared to the included cases reveals that they are not significantly different ( p( drop |
notice) = .074 compared to p( drop | not noticed) = 0.087; c_( 1) = 0.571; p > .05 and
p( drop | death sentence) = 0.066 compared to p( drop | no death sentence) = 0.084; c_( 1) =
0.319; p > .05).
To address the impact of losing these cases on the results, a series of analyses
presented in Tables 5- 8 and Figures 6- 10 was conducted. On balance, these data reveal
very little change in any of the basic descriptive quantities presented earlier. This
evidence suggests ( but does not prove) that the missing cases are a relatively
representative sample of the universe of death eligible cases.
The race of victim and race of defendant variables had additional missing data
problems. There were 18 additional cases with missing offender race information
yielding a sample of 1,202 – 18 = 1,184 cases for race of offender analyses. There were
124 cases with missing victim race information yielding a sample of 1,202 – 124 = 1,078
cases for race of victim analyses. Finally for analyses of the intersection of victim and
offender race, an additional 141 cases were lost yielding a sample of 1,202 – 141 = 1,061
28
cases. The differences in the decision outcomes between the cases with missing and
observed data on these variables are statistically and substantively significant. For
example, among the 124 cases with missing race of victim information, 121 came from
the group of individuals who were not noticed while only 3 came from the noticed cases.
Similar disparities were noted for the race of offender and the intersection of offender and
victim race variables. The lack of information associated with these cases is an important
but unavoidable weakness of this study. Readers of this report must bear in mind that
analyses involving victim race and the intersection of victim and offender race have
disproportionately eliminated death eligible cases that were not death noticed.
Table 9 presents the list of covariates used in the study. The first set of covariates
in this table were observed for the full 1,202 cases while the second set were measures of
statutory aggravating factors which were only observed for the 327 cases that were death
noticed ( 26 cases out of the original 353 death noticed cases were lost to missing data as
described above). Most of the entries in this table are proportions which means that they
can be interpreted as the number of cases having the characteristic divided by the total
number of cases. The total number of cases is 1,202 except for the statutory aggravating
factors where the total number of cases is 327.
Adjusted Analysis: Between- JurisdictionVariation
Tables 10A- 10F present the details of a multiple- variable logistic regression
analysis of county processing patterns at different stages of the death penalty system in
Maryland. Table 10A reports the results for the decision of the state’s attorney to file a
notification to seek the death penalty. The parameter estimates for the case characteristics
29
are given in Table 10A, along with the estimated effects for each county. This table
shows that there is significant jurisdiction- to- jurisdiction variation in the way the decision
to file a death notification is handled. Compared to the reference category, state’s
attorneys in Anne Arundel County, Baltimore City, and Montgomery County are
significantly less likely to file a notification to seek the death penalty while cases in
Baltimore County are significantly more likely to be death notified. It is important to note
that this substantial variation by legal jurisdiction in the decision to seek a death
sentence exists even after controlling for numerous case characteristics.
Table 10B reports the results of the decision to not withdraw a death notification
once filed. There again is substantial variation across legal jurisdictions in Maryland in
the decision to withdraw a death notification. State’s attorneys in Baltimore County are
significantly more likely to have a death notification once filed “ stick”. In examining
Table 10C there is no longer a significant jurisdiction effect. This means that in terms of
the decision by the state’s attorney to advance a case to a penalty trial, there is no
significant variation across the different legal jurisdictions in Maryland. Table 10D
reports the results for the decision to impose a death sentence given that a penalty trial
occurs. Again there is a significant jurisdictional effect with cases from Baltimore County
more likely to be sentenced to death, even after case characteristics are considered. Table
10E reports the logistic regression analysis of whether a defendant receives a death
sentence given the fact that it is a death eligible case. The results again show a significant
effect for the charging jurisdiction. Death eligible defendants in Baltimore City and
Prince George’s County are significantly less likely to be sentenced to death while those
in Baltimore County are significantly more likely to be sentenced to death. We know that
30
the reason for this is the significantly different rate at which prosecutors in the different
locations in the state make capital charges and make those capital charges “ stick” early in
the capital punishment process.
To provide an easy way to interpret the magnitude of the county effect at each
decision making point, the predicted probability of each outcome ( both before and after
the statistical controls for case characteristics) is reported for each jurisdiction in Table
10F. Looking at the decision to file a death notification, we can see that the predicted
probability that a death notice will be filed in a death eligible case ranges from a high of
.620 in Baltimore County to a low of .046 in Baltimore City. This means that given the
fact that a death eligible homicide has occurred, the probability that a notification to seek
death will be filed in Baltimore County is over 13 times higher than in Baltimore City,
even after taking into account important case characteristics. The probability of being
death notified if a case is in Baltimore County is over five times greater than if it
occurred in Montgomery County and three times greater than if it occurred in Anne
Arundel County. This jurisdiction- to- jurisdiction variation in the probability of a death
notice given a death eligible offense is statistically significant.
There is also substantial variation across the different Maryland jurisdictions in
the probability that a death notice once filed will “ stick”. This probability is highest in
Baltimore County, Anne Arundel County, and Baltimore City and lowest in Prince
George’s County. These jurisdictional variations are also statistically significant. In
looking at the probability that a case will be advanced to a penalty trial or will be
sentenced to death after a penalty trial, the variation by jurisdiction becomes much
smaller and is not statistically significant. It is very easy to see from Table 10F that the
31
probability of these latter two decisions is fairly consistent across the different
jurisdictions in the states. The effect of early prosecutorial decisions on later stages of the
capital sentencing process can be seen in the death penalty decisions. There is substantial
jurisdictional variation in the decision to impose a death sentence for a death eligible
homicide that is unexplained by case characteristics.
What these results indicate is that clearly the jurisdiction where the homicide
occurs matters and matters a great deal. There are large differences in how different legal
jurisdictions process their death penalty cases in Maryland. These differences are
manifested in how state’s attorneys charge death eligible cases and whether they retain a
capital charge or decide to withdraw it. Although the jurisdictional differences occur
early in the process they are propagated to later points and go uncorrected. It is also
important to note here that the variation in how death cases are handled in the different
legal jurisdictions in Maryland that we found in the previously reported unadjusted
analysis holds up in the multivariate analysis when numerous case characteristics are
considered. In other words, differences in how different jurisdictions handle death
eligible cases cannot be attributed to the kinds of homicides committed in those
jurisdictions.
Adjusted Analysis: Offender’s Race
Tables 11A to 11F report the results of a multivariate logistic regression analysis
looking at the race of the offender at various stages of Maryland’s death sentencing
process. Looking across the different decision points, there is no evidence that the race of
the defendant matters at any stage once case characteristics are controlled for. This is
32
best seen in Table 11F which shows the predicted probability of each stage occurring for
black and non- black defendants both before and after considering case characteristics. As
we found earlier in the unadjusted analysis, there is a slight tendency for black offenders
to be less likely to be death notified. This disappears, however, when case characteristics
are taken into account, and in fact switches direction with black offenders slightly more
at risk. The difference is quite small, however, and the differences for each stage between
black and non- black offenders are small. The largest difference occurs at the decision to
impose a death sentence given a penalty trial. There, the probability that a black offender
will be sentenced to death is .444 and the probability for non- black offenders is .376, a
statistically non- significant difference of .068. In sum, we have found no evidence that
the race of the defendant matters in the processing of capital cases in the state.
C. Victim Race
In Tables 12A to 12F we report race of victim patterns at various stages of the
state’s death penalty system in a multivariate logistic regression model that considers
both relevant case characteristics and the jurisdiction where the crime occurred. Recall
that in the unadjusted analysis killers of white victims were significantly more likely to
be death noticed, to have that death notification “ stick”, and to be advanced to a penalty
trial than those who killed non- black victims. In the adjusted analysis we find that this
difference in the handling of black victim and non- black victim cases cannot be explained
by the case characteristics of the homicide. Table 12A shows that even after case factors
and jurisdictional differences are taken into account, those who kill whites are still
significantly more likely to have the state’s attorney file a notification to seek the death
penalty. Table 12B reveals that the decision not to withdraw a death notification is also
33
related to the race of the victim. After considering jurisdiction and case characteristics,
state’s attorneys are significantly less likely to withdraw a death notification if a white
victim is killed compared with a non- white victim. This race of victim effect does not
hold up, however, at the decision of the state’s attorney to advance a case to a penalty
trial ( Table 12C) 17, and at the decision of the judge or jury to impose a death sentence
given that a penalty trial has occurred ( Table 12D). 18 In Table 12E we report the results
of a logistic regression model for defendants who are sentenced to death within the pool
of all death eligible cases. This table shows that even taking into account jurisdiction and
relevant case characteristics offenders who slay white victims are significantly more
likely to be sentenced to death than those who slay all non- white victims.
For all of our analyses we estimated a stepwise logistic regression model to see if
our results would hold up under a different model specification and with only two
exceptions they did. The stepwise logistic regression model for the race of the victim on
whether the defendant receives a death sentence for all death eligible cases was one of
these exceptions. In our first model the effect of victim’s race was significant at p < .05,
in the stepwise model reported in Table 12F, the effect of victim’s race was still present
but now it was statistically significant at only a .07 level. The estimated logistic
regression parameter was reduced from 1.216 ( with an odds multiplier of 3.37) to .721
with an odds multiplier of ( 2.06). The race of the victim still matters, but we would note
that in a different model specification it’s level of statistical significance declines.
17 The race of victim effect in this model is marginally significant with a p < .10.
18 We provide two versions of these tables. In the first there is a case characteristic that was not
significantly related to the outcome variable but was included to make a more conservative test of the racial
disparity hypothesis. In the second version this factor is dropped and there is only one significant case
characteristic included.
34
In order to better capture the magnitude of the race of victim effect, in Table 12G
we have calculated the predicted probability of each outcome in the death sentencing
process for white and non- white victim cases both before and after adjusting for case
characteristics. The adjusted probability that a state’s attorney will seek a death
notification when a white is killed is .266 and .169 when a black is killed. This means
that the probability of a death notification in a white victim cases is 1.6 times higher than
that for a black victim homicide, even after considering relevant case characteristics and
the jurisdiction where the homicide occurred. The probability of a death notification
“ sticking” is 1.5 times higher in white victim than black victim cases again after taking
into account case factors and jurisdiction. At both these early decision making points,
then, the race of the victim killed in a homicide is an important factor in determining
which death eligible defendants are notified that the state will seek the death penalty
against them, and for whom that notification will “ stick”. The last entry in Table 12F
shows that for all death eligible homicides the probability of a death sentence in a white
victim case is three times higher than in a non- white victim homicide. The estimated
probability for a death sentence among death eligible homicides in the stepwise model is
.022 for white victim cases and .011 for non- white victim cases. The probability that a
white victim death eligible homicide will result in a death sentence is now only two times
higher than in a non- white victim homicide. In the stepwise model, the effect of victim’s
race does diminish from our earlier model specification, but it still substantively matters.
In sum, we find a significant effect for the race of the victim in the way the
prosecutor initially handles death eligible homicides. State’s attorneys in Maryland are
more likely to file a notification to seek a death sentence and more likely to retain that
35
notification when the race of the victim is white rather than black. Furthermore, this race
of victim effect is not explained by case characteristics of white and non- white victims or
by the jurisdiction where the homicide occurred. This initial disparity is not corrected at
later stages of the capital sentencing process. The race of the victim does not appear to
matter when the decision is to advance a case to the penalty phase or to sentence a
defendant to death after a penalty phase hearing.
D. Offender- Victim Race Combinations
In Tables 13A to 13G we report logistic regression models for combinations of
offender’s and victim’s race. Table 13A shows that net of relevant case characteristics
and the jurisdiction where the homicide occurred, all combinations of offender’s and
victim’s race are less likely to be charged with a capital crime given that it is death
eligible than black offenders who kill white victims. 19 Both blacks who kill blacks and
homicides involving “ other” combinations of offender’s and victim’s race are
significantly less likely to have a death notification “ stick” than homicides involving
black offenders and white victims ( Table 13B). 20 There is no race of offender/ victim
effect at either the decision to advance a case to a penalty hearing ( Table 13D) or the
decision to sentence a defendant to death given a penalty hearing ( Table 13E). Table 13F
does show, however, that black offenders who slay white victims are more likely to be
19 Other combinations include es white offenders who kill blacks but also a handful of cases involving
“ other” races ( Hispanic, Asian, Native American) of either the victim or offender.
20 The only other occasion where the stepwise logistic regression model produced a different result than our
earlier model specification is in this instance. Table 13C shows that in the stepwise specification there are
no significant differences in the decision to withdraw a death notification across race of offender/ victim
groups.
36
sentenced to death than other racial combinations given the fact that a homicide is death
eligible.
Table 13G provides the predicted probabilities of each outcome for the four race
of offender/ victim combinations. The probability that the state’s attorney will file a
notification to seek the death penalty is highest in cases where a black offender kills a
white victim (. 355), and is twice as high as when a black slays another black (. 174) or
other racial combinations (. 166), and 1.7 times higher than when a white kills a white.
Even when case characteristics and jurisdiction are controlled, blacks who cross racial
lines and kill whites are more likely to be death notified. The probability that a death
notification will “ stick” is also higher ( about 1.5 times) for both whites who kill blacks
and blacks who kill blacks compared with blacks who kill blacks and “ other” racial
combinations. It is also interesting to note that given that a homicide is death eligible,
blacks who kill whites are two and one- half times more likely to be sentenced to death
than are whites who kill whites (. 043 vs. .017), three and one- half times more likely than
are blacks who kill blacks (. 043 vs. .012), and almost eleven times more likely to be
sentenced to death than “ other” racial combinations (. 043 vs. .004). Consistently, black
offenders who kill white victims are at greater risk of in Maryland’s capital sentencing
system even after controlling or numerous case characteristics and the jurisdiction where
the crime occurred.
In sum, in our analysis we have found evidence for a race of victim effect and an
effect for the combination of offender’s and victim’s race. Offenders who kill white
victims, especially if the offender is black, are significantly and substantially more likely
to be charged with a capital crime ( state’s attorney decides to file a notification to seek
37
the death penalty). Those who kill white victims are also significantly more likely to have
their death notification “ stick” than those who kill non- whites. These effects persist even
in the presence of what we think are very rigorous controls for relevant case
characteristics. Moreover, while these effects do not appear at other, later decision
making points in the capital sentencing process they are generally not corrected.
THE IMPORTANCE OF JURISDICTION
One of the most impressive findings from this research is the power that state’s
attorneys have and exercise in determining whether or not to process a death eligible
homicide as a capital crime. The variation in the treatment of cases across the different
legal jurisdictions was substantial and robust. In the Maryland death penalty system, the
jurisdiction where the crime occurs and legal prosecution begins is clearly one of the
most important factors, and cannot be ignored. We provide some supplemental analyses
to demonstrate the role of legal jurisdiction in the handling of death penalty cases.
In Table 14 we report the results of a series of logistic regression models. The
cases are all those where a notification was filed that the state intends to seek the death
penalty, and the decision is whether or not a death sentence is imposed. Model 1 shows
that considered alone the race of the victim matters, those who kill white victims are an a
substantially increased risk of being sentenced to death compared with those who kill
non- whites. In Model 2 we enter the number of statutory aggravating factors that the
prosecutor charges in the death notification. The number of statutory aggravating factors
clearly elevates the risk of a death sentence and it diminishes, but does not eliminate the
race of victim effect. The results for Model 2 are identical to those reported by Professors
38
Baldus and Woodworth in their 2001 analysis of Maryland death- noticed cases. In Model
3 we drop the number of statutory aggravating factors and add variables for the charging
jurisdiction. When the prosecuting jurisdiction is added to the model, the effect for the
victim’s race diminishes substantially, and is no longer statistically significant. This
would suggest that jurisdiction and race of victim are confounded. There are state’s
attorneys in Maryland who more frequently pursue the death penalty than others. It also
happens that there are more white victim homicides committed in these jurisdictions
where there is a more frequent pursuit of the death penalty. When both jurisdiction and
the number of statutory aggravating factors are included in the model ( Model 4), the
effect of victim’s race declines again, but only slightly.
We report a similar analysis in Table 15 but here we focus on black offenders
who kill white victims among the subset of death- notified cases. Model 1 shows that
black offenders who kill white victims are significantly more likely than other racial
combinations to be sentenced to death. This is true even when there are controls for the
number of statutory aggravating factors ( Model 2). When the jurisdiction variables are
entered in Model 3, the effect for black kills white is reduced by about twenty- two
percent, but is still significant. It continues to be significant even with controls for both
county and the number of statutory aggravating factors ( Model 4).
The effect of jurisdiction in reducing the race effect ( revealing the confound
between jurisdiction and race) is shown in Tables 16 and 17. Table 16 reports the results
of a logistic regression analysis for death eligible cases for the race of the victim on both
the death notice decision and the death sentence decision. Without jurisdiction controls,
there is a very strong relationship between killing a white victim and being death noticed
39
and being sentenced to death. With just the addition of the jurisdiction controls, however,
this relationship, though still statistically significant, is reduced dramatically. For the
death notice decision the difference in the estimated probability of a death notice between
white and non- white victims declines from .256 to .110, a 132% decrease. For the death
sentence decision, the difference in the predicted probability of a death sentence between
white and non- white victims declines from .090 to .037, a 143% decrease.
The different treatment that is given to death penalty cases for different race
groups is summarized in Table 18. From this table it is very clear to see that the two
counties with the highest death notice and death sentencing rates ( Baltimore and Harford)
are also the two counties with the highest rates of white victim and black defendant white
victim death eligible homicides. What this implies is that any attempt to deal with any
racial disparity in the imposition of the death penalty in Maryland cannot ignore the
substantial variability that exists in different state’s attorneys’ offices in the processing of
death cases.
CONCLUSION
The analysis presented in this report has explored a number of issues related to the
death penalty sentencing system in Maryland. The primary focus has been on the
possible effects of geography and race of victim and race of defendant variables. The
analysis suggests that both classes of variables play an important role in the Maryland
system. The evidence indicates that these factors exert their greatest effects at the death
notice and death notice retraction decisions. Later stages of the system do not appear to
exacerbate or increase the magnitude of these effects. But the effects of the earliest
40
decisions in the Maryland system are apparent in the actual imposition of death sentences
and capital punishment when the group of individuals who receive these sentences are
compared to the entire state’s universe of death eligible cases.
The research presented in this report is limited in several respects. First, as
discussed earlier there were significant quantities of missing data on the race of some
victims. These cases were disproportionately lost at the notice decision and some of the
most important effects estimated in this study revolved around the death notice decision.
Nevertheless, the effects of geography remained strong at this decision point even with
all of the cases with missing victim race included in the analysis. Efforts to rigorously
address this missing data problem related to victim race will be developed in the months
ahead.
A second weakness which will be more difficult to address involves the inability
to hold statutory aggravating factors constant at the notice decision. This is a logical
problem since prosecutors are not required to identify statutory aggravating factors unless
a death notification is issued. Nevertheless, these variables appear to play an important
role in decisions related to the imposition of capital punishment in Maryland and future
research will benefit from addressing this issue.
A third issue that this report has not addressed is whether the statewide results
estimated here hold equally for all counties. A challenge for research on this topic is that
the sample sizes for statistical analysis become very small as the sample is subdivided.
Some specialized statistical methods for addressing small sample problems are becoming
more feasible to implement with current statistical computing technology and this is
another issue that will be explored with the Maryland death penalty data.
41
These weaknesses notwithstanding, the report does identify several clear
statewide patterns among the cases that are fully observed. These patterns include
statistically significant effects for geographic, race of victim, and joint offender- victim
race groups on the imposition of death sentences in Maryland. The data suggest that
most of these patterns become apparent at the earliest stages of processing within the
state’s death penalty system.
Figure 1
Number of Cases Progressing Through Each Stage of the Maryland Death Penalty System
Notes:
Stage 1 = Universe of death eligible cases in Maryland from July 1, 1978 to December 31, 1999
( N = 1,311).
Stage 2 = Subset of death eligible cases where the prosecutor files notice of intent to seek the death
penalty ( N = 353). The conditional probability of filing notice given a death eligible case is 353/ 1311 =
0.269.
Stage 3 = Subset of death- noticed cases where the prosecutor does not retract notice ( i. e., the death notice
“ sticks.”) ( N = 213). The conditional probability of notice sticking given a death eligible case is 213/ 1311
= 0.162 and the conditional probability of notice sticking given that the prosecutor files notice is 213/ 353
= 0.603.
Stage 4 = Subset of “ stuck” death- noticed cases advancing to a penalty trial ( N = 180). The conditional
probability of a case advancing to the penalty phase given that the prosecutor sticks with the death notice
is 180/ 213 = 0.845.
Stage 5 = Subset of penalty trial cases resulting in the imposition of a death sentence ( N = 76). The
conditional probability of a death sentence being imposed given a death eligible case is 76/ 1311 is 0.058.
The conditional probability of a death sentence being imposed given a death notice that sticks is 76/ 213 =
0.357. Finally, the conditional probability of a death sentence being imposed given a penalty trial is 76/
180 = 0.422.
Number of Cases
Stage of System
0
200
400
600
800
1000
1200
1400
N = 1311 353 213 180 76
Stage 1 Stage 2 Stage 3 Stage 4 Stage 5
Figure 2
Offender Race Distribution
Notes:
Stage 1 = Universe of death eligible cases in Maryland from July 1, 1978 to December 31, 1999
( N = 1,311). There is no information about the race of the defendant in 20 ( 1.5%) of the cases.
Stage 2 = Subset of death eligible cases where the prosecutor files notice of intent to seek the death
penalty ( N = 353). The defendant’s race is unknown in 1 case.
Stage 3 = Subset of death- noticed cases where the prosecutor does not retract notice ( i. e., the death notice
“ sticks.”) ( N = 213). The defendant’s race is unknown in 1 case.
Stage 4 = Subset of cases that advance to a penalty trial ( N = 180). The defendant’s race is unknown in 1
case.
Stage 5 = Subset of cases reaching the penalty phase resulting in the imposition of a death sentence ( N =
76). The defendant’s race is observed for all 76 cases.
Proportion of Cases
Stage of System
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Other Offender 0.022 0.011 0.009 0.011 0.000
Black Offender 0.743 0.653 0.637 0.639 0.684
White Offender 0.236 0.335 0.354 0.350 0.316
Stage 1 Stage 2 Stage 3 Stage 4 Stage 5
Figure 3
Victim Race Distribution
Notes:
Stage 1 = Universe of death eligible cases in Maryland from July 1, 1978 to December 31, 1999
( N = 1,311). There is no information about the race of victim in 139 ( 10.6%) of the cases. The case is
considered a “ white victim case” if at least one white person is killed ( in 11 cases at least one white and at
least one black were killed; these cases are considered “ white” because at least one white person was
killed). The nonwhite victim group ( i. e., cases with no white victims) is comprised mainly of cases with
at least one black victim ( N = 593; 92.1%).
Stage 2 = Subset of death eligible cases where the prosecutor files notice of intent to seek the death
penalty ( N = 353). The victim’s race is unknown in 6 ( 1.7%) of the 353 cases. The nonwhite victim
group is comprised mainly of cases with at least one black victim ( N = 109; 90.8%).
Stage 3 = Subset of death- noticed cases where the prosecutor does not retract notice ( i. e., the death notice
“ sticks.”) ( N = 213). The nonwhite victim group is comprised mainly of cases with at least one black
victim ( N = 49; 89.1%).
Stage 4 = Subset of cases that advance to a penalty trial ( N = 180). The nonwhite victim group is com-prised
mainly of cases with at least one black victim ( N = 36; 87.8%).
Stage 5 = Subset of cases reaching the penalty phase resulting in the imposition of a death sentence ( N =
76). The victim’s race is observed for all 76 cases. The nonwhite victim group is comprised entirely of
cases with at least one black victim ( N = 15).
Proportion of Cases
Stage of System
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Nonwhite 0.547 0.346 0.258 0.228 0.197
W hite 0.453 0.654 0.742 0.772 0.803
1 2 3 4 5
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Other Combination 0.054 0.038 0.033 0.033 0.000
White Defendant/ Black Victim 0.019 0.032 0.014 0.006 0.013
Black Defendant/ White Victim 0.230 0.347 0.401 0.428 0.500
Black Defendant/ Black Victim 0.479 0.280 0.217 0.194 0.184
White Defendant/ White Victim 0.219 0.304 0.335 0.339 0.303
1 2 3 4 5
Figure 4
Joint Offender- Victim Race Distribution
Notes:
Stage 1 = Universe of death eligible cases in Maryland from July 1, 1978 to December 31, 1999
( N = 1,311). Either the race of the victim or the defendant is unknown in 158 ( 12.1%) of the cases.
Stage 2 = Subset of death eligible cases where the prosecutor files notice of intent to seek the death
penalty ( N = 353). Either the race of the victim or the defendant is unknown in 7 of these 353 ( 2.0%)
cases.
Stage 3 = Subset of death- noticed cases where the prosecutor does not retract notice ( i. e., the death notice
“ sticks.”) ( N = 213). Either the race of the victim or the defendant is unknown in 1 of these cases.
Stage 4 = Subset of cases advancing to a penalty trial ( N = 180). Both the race of the victim and the
defendant are observed for all 180 cases.
Stage 5 = Subset of cases reaching the penalty phase resulting in the imposition of a death sentence ( N =
76). Both the race of the victim and the defendant are observed for all 76 cases.
Proportion of Cases
Stage of System
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Other Counties 0.140 0.238 0.202 0.200 0.184
Prince George's 0.176 0.247 0.164 0.144 0.105
Montgomery 0.045 0.031 0.033 0.039 0.026
Harford 0.027 0.054 0.038 0.033 0.040
Baltimore County 0.117 0.281 0.390 0.417 0.447
Baltimore City 0.435 0.099 0.113 0.100 0.132
Anne Arundel 0.061 0.051 0.061 0.067 0.066
1 2 3 4 5
Figure 5
County Contributions to Each Stage of the Maryland Death Penalty System
Proportion of Cases
Stage of System
Notes:
Stage 1 = Universe of death eligible cases in Maryland from July 1, 1978 to December 31, 1999
( N = 1,311). The county in which the charge is brought is unknown in six of these cases. For all subsequent
stages there is no missing county information.
Stage 2 = Subset of death eligible cases where the prosecutor files notice of intent to seek the death penalty ( N
= 353).
Stage 3 = Subset of death- noticed cases where the prosecutor does not retract notice ( i. e., the death notice
“ sticks.”) ( N = 213).
Stage 4 = Subset of cases advancing to a penalty trial ( N = 180).
Stage 5 = Subset of cases reaching the penalty phase resulting in the imposition of a death sentence ( N = 76).
Figure 6
Number of Cases Progressing Through Each Stage of the
Maryland Death Penalty System After Listwise Deletion of Missing Cases
Notes:
Stage 1 = Universe of death eligible cases in Maryland from July 1, 1978 to December 31, 1999
( N = 1,202).
Stage 2 = Subset of death eligible cases where the prosecutor files notice of intent to seek the death
penalty ( N = 327). The conditional probability of filing notice given a death eligible case is 327/ 1202 =
0.272.
Stage 3 = Subset of death- noticed cases where the prosecutor does not retract notice ( i. e., the death notice
“ sticks.”) ( N = 198). The conditional probability of notice sticking given a death eligible case is 198/
1202 = 0.165 and the conditional probability of notice sticking given that the prosecutor files notice is
198/ 327 = 0.606.
Stage 4 = Subset of “ stuck” death- noticed cases advancing to a penalty trial ( N = 169). The conditional
probability of a case advancing to the penalty phase given that the prosecutor sticks with the death notice
is 169/ 198 = 0.854.
Stage 5 = Subset of penalty trial cases resulting in the imposition of a death sentence ( N = 71). The
conditional probability of a death sentence being imposed given a death eligible case is 71/ 1202 is 0.059.
The conditional probability of a death sentence being imposed given a penalty trial is 71/ 169 = 0.420.
Number of Cases
Stage of System
0
200
400
600
800
1000
1200
1400
N = 1202 327 198 169 71
Stage 1 Stage 2 Stage 3 Stage 4 Stage 5
Figure 7
Offender Race Distribution after Listwise Deletion of Missing Cases
Notes:
Stage 1 = Universe of death eligible cases in Maryland from July 1, 1978 to December 31, 1999
( N = 1,202). There is no information about the race of the defendant in 18 ( 1.5%) of the cases.
Stage 2 = Subset of death eligible cases where the prosecutor files notice of intent to seek the death
penalty ( N = 327). The defendant’s race is unknown in 1 case.
Stage 3 = Subset of death- noticed cases where the prosecutor does not retract notice ( i. e., the death notice
“ sticks.”) ( N = 198). The defendant’s race is unknown in 1 case.
Stage 4 = Subset of cases that advance to a penalty trial ( N = 169). The defendant’s race is observed for
all 169 cases.
Stage 5 = Subset of cases reaching the penalty phase resulting in the imposition of a death sentence ( N =
71). The defendant’s race is observed for all 71 cases.
Proportion of Cases
Stage of System
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Other Offender 0.026 0.011 0.010 0.012 0.000
Black Offender 0.734 0.645 0.619 0.621 0.676
White Offender 0.242 0.344 0.371 0.367 0.324
Stage 1 Stage 2 Stage 3 Stage 4 Stage 5
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
Nonwhite 0.547 0.336 0.242 0.213 0.211
W hite 0.453 0.664 0.758 0.787 0.789
1 2 3 4 5
Figure 8
Victim Race Distribution After Listwise Deletion of Missing Cases
Notes:
Stage 1 = Universe of death eligible cases in Maryland from July 1, 1978 to December 31, 1999
( N = 1,202). There is no information about the race of victim in 124 ( 10.3%) of the cases. The case is
considered a “ white victim case” if at least one white person is killed. The nonwhite victim group ( i. e.,
cases with no white victims) is comprised mainly of cases with at least one black victim ( N = 543;
92.0%).
Stage 2 = Subset of death eligible cases where the prosecutor files notice of intent to seek the death
penalty ( N = 327). The victim’s race is unknown in 3 ( 0.9%) of the 327 cases. The nonwhite victim
group is comprised mainly of cases with at least one black victim ( N = 98; 90.0%).
Stage 3 = Subset of death- noticed cases where the prosecutor does not retract notice ( i. e., the death notice
“ sticks.”) ( N = 198). The nonwhite victim group is comprised mainly of cases with at least one black
victim ( N = 42; 87.5%).
Stage 4 = Subset of cases that advance to a penalty trial ( N = 169). The nonwhite victim group is com-prised
mainly of cases with at least one black victim ( N = 31; 86.1%).
Stage 5 = Subset of cases reaching the penalty phase resulting in the imposition of a death sentence ( N =
71). The victim’s race is observed for all 71 cases. The nonwhite victim group is comprised entirely of
cases with at least one black victim ( N = 15).
Proportion of Cases
Stage of System
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Other Combination 0.055 0.040 0.036 0.035 0.000
White Defendant/ Black Victim 0.018 0.028 0.015 0.006 0.014
Black Defendant/ White Victim 0.221 0.344 0.401 0.426 0.479
Black Defendant/ Black Victim 0.478 0.272 0.198 0.178 0.197
White Defendant/ White Victim 0.228 0.316 0.350 0.355 0.310
1 2 3 4 5
Figure 9
Joint Offender- Victim Race Distribution after Listwise Deletion of Missing Cases
Notes:
Stage 1 = Universe of death eligible cases in Maryland from July 1, 1978 to December 31, 1999
( N = 1,202). Either the race of the victim or the defendant is unknown in 141 ( 11.7%) of the cases.
Stage 2 = Subset of death eligible cases where the prosecutor files notice of intent to seek the death
penalty ( N = 327). Either the race of the victim or the defendant is unknown in 4 of these 327 ( 1.2%)
cases.
Stage 3 = Subset of death- noticed cases where the prosecutor does not retract notice ( i. e., the death notice
“ sticks.”) ( N = 198). Either the race of the victim or the defendant is unknown in 1 of these cases.
Stage 4 = Subset of cases advancing to a penalty trial ( N = 169). Both the race of the victim and the
defendant are observed for all 169 cases.
Stage 5 = Subset of cases reaching the penalty phase resulting in the imposition of a death sentence ( N =
71). Both the race of the victim and the defendant are observed for all 71 cases.
Proportion of Cases
Stage of System
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Other Counties 0.142 0.245 0.212 0.207 0.183
Prince George's 0.179 0.242 0.157 0.136 0.113
Montgomery 0.046 0.034 0.035 0.041 0.028
Harford 0.027 0.052 0.040 0.036 0.042
Baltimore County 0.118 0.287 0.399 0.426 0.465
Baltimore City 0.424 0.085 0.091 0.083 0.099
Anne Arundel 0.064 0.055 0.066 0.071 0.070
1 2 3 4 5
Figure 10
County Contributions After Listwise Deletion of Missing Cases
Proportion of Cases
Stage of System
Notes:
Stage 1 = Universe of death eligible cases in Maryland from July 1, 1978 to December 31, 1999
( N = 1,202).
Stage 2 = Subset of death eligible cases where the prosecutor files notice of intent to seek the death penalty
( N = 327).
Stage 3 = Subset of death- noticed cases where the prosecutor does not retract notice ( i. e., the death notice
“ sticks.”) ( N = 198).
Stage 4 = Subset of cases advancing to a penalty trial ( N = 169).
Stage 5 = Subset of cases reaching the penalty phase resulting in the imposition of a death sentence ( N = 71).
Table 1
Processing of Offender Race Groups at Various Stages of Maryland Death Penalty System
Test # 1
p( death notice | death eligible offense, offender is black) = 0.240
p( death notice | death eligible offense, offender is not black) = 0.368
N = 1,291 ( 20 Missing Cases); χ ² w/ 1 df = 20.259; p < .05
Test # 2
p( death notice sticks | death notice, offender is black) = 0.581
p( death notice sticks | death notice, offender is not black) = 0.631
N = 352 ( 1 Missing Case); χ ² w/ 1 df = 0.650; p > .05
Test # 3
p( penalty trial | death notice sticks, offender is black) = 0.852
p( penalty trial | death notice sticks, offender is not black) = 0.844
N = 212 ( 1 Missing Case); χ ² w/ 1 df = 0.023; p > .05
Test # 4
p( death sentence | penalty trial, offender is black) = 0.452
p( death sentence | penalty trial, offender is not black) = 0.369
N = 180; χ ² w/ 1 df = 1.171; p > .05
Test # 5
p( death sentence | death eligible offense, offender is black) = 0.054
p( death sentence | death eligible offense, offender is not black) = 0.072
N = 1291 ( 20 Missing Cases); χ ² w/ 1 df = 1.453; p > .05
} 0.240 - 0.368 = - 0.128
} 0.581- 0.631 = - 0.050
} 0.852- 0.844 = 0.012
} 0.452- 0.369 = 0.083
} 0.054- 0.072 = - 0.018
Table 2
Processing of Victim Race Groups at Various Stages of Maryland Death Penalty System
Test # 1
p( death notice | death eligible offense, at least one victim is white) = 0.430
p( death notice | death eligible offense, no white victim) = 0.186
N = 1,172 ( 139 Missing Cases); χ ² w/ 1 df = 82.600; p < .05
Test # 2
p( death notice sticks | death notice, at least one victim is white) = 0.696
p( death notice sticks | death notice, no white victim) = 0.458
N = 347 ( 6 Missing Cases); χ ² w/ 1 df = 18.712; p < .05
Test # 3
p( penalty trial | death notice sticks, at least one victim is white) = 0.880
p( penalty trial | death notice sticks, no white victim) = 0.746
N = 213; χ ² w/ 1 df = 5.620; p < .05
Test # 4
p( death sentence | penalty trial, at least one victim is white) = 0.439
p( death sentence | penalty trial, no white victim) = 0.366
N = 180; χ ² w/ 1 df = 0.692; p > .05
Test # 5
p( death sentence | death eligible offense, at least one victim is white) = 0.116
p( death sentence | death eligible offense, no white victim) = 0.023
N = 1172 ( 139 Missing Cases); χ ² w/ 1 df = 40.705; p < .05
} 0.430- 0.186 = 0.244
} 0.696- 0.458 = 0.238
} 0.880- 0.746 = 0.134
} 0.439- 0.366 = 0.073
} 0.116- 0.023 = .093
Table 3A
Processing of White Defendant- White Victim Cases at Various Stages of the Maryland Death Penalty System
Test # 1
p( death notice | death eligible offense, white defendant & at least one white victim ) = 0.417
p( death notice | death eligible offense, all other victim- offender race combinations) = 0.268
N = 1,153 ( 158 Missing Cases); χ ² w/ 1 df = 20.867; p < .05
Test # 2
p( death notice sticks | death notice, white defendant & at least one white victim ) = 0.676
p( death notice sticks | death notice, all other victim- offender race combinations) = 0.585
N = 346 ( 7 Missing Cases); χ ² w/ 1 df = 2.560; p > .05
Test # 3
p( penalty trial | death notice sticks, white defendant & at least one white victim ) = 0.859
p( penalty trial | death notice sticks, all other victim- offender race combinations) = 0.844
N = 212 ( 1 Missing Case); χ ² w/ 1 df = 0.085; p > .05
Test # 4
p( death sentence | penalty trial, white defendant & at least one white victim ) = 0.377
p( death sentence | penalty trial, all other victim- offender race combinations) = 0.445
N = 180; χ ² w/ 1 df = 0.772; p > .05
Test # 5
p( death sentence | death eligible case, white defendant & at least one white victim ) = 0.091
p( death sentence | death eligible case, all other victim- offender race combinations) = 0.059
N = 1,153 ( 158 Cases Missing); χ ² w/ 1 df = 3.367; p > .05
Note: any case with at least one white victim is defined as a “ white victim” case.
} 0.417- 0.268 = 0.149
} 0.676- 0.585 = 0.091
} 0.859- 0.844 = 0.015
} 0.377- 0.445 = - 0.068
} 0.091- 0.059 = 0.032
Table 3B
Processing of Black Defendant- Black Victim Cases at Various Stages of the Maryland Death Penalty System
Test # 1
p( death notice | death eligible offense, black defendant & black victim) = 0.176
p( death notice | death eligible offense, all other victim- offender race combinations) = 0.414
N = 1,153 ( 158 Missing Cases); χ ² w/ 1 df = 77.979; p < .05
Test # 2
p( death notice sticks | death eligible offense, black defendant & black victim) = 0.474
p( death notice sticks | death eligible offense, all other victim- offender race combinations) = 0.667
N = 346 ( 7 Missing Cases); χ ² w/ 1 df = 10.894; p < .05
Test # 3
p( penalty trial | death notice sticks, black defendant & black victim) = 0.761
p( penalty trial | death notice sticks, all other victim- offender race combinations) = 0.874
N = 212 ( 1 Missing Case); χ ² w/ 1 df = 3.565; p > .05
Test # 4
p( death sentence | penalty trial, black defendant & black victim) = 0.400
p( death sentence | penalty trial, all other victim- offender race combinations) = 0.428
N = 180; χ ² w/ 1 df = 0.089; p > .05
Test # 5
p( death sentence | death eligible case, black defendant & black victim) = 0.025
p( death sentence | death eligible case, all other victim- offender race combinations) = 0.103
N = 1,153 ( 158 Missing Cases); χ ² w/ 1 df = 28.285; p < .05
Note: any case with no white victims and at least one black victim is defined as a “ black victim” case.
} 0.176- 0.414 = - 0.238
} 0.474- 0.667 = - 0.193
} 0.761- 0.874 = - 0.113
} 0.400- 0.428 = - 0.028
} 0.025- 0.103 = -. 078
Table 3C
Processing of Black Defendant- White Victim Cases at Various Stages of the Maryland Death Penalty System
Test # 1
p( death notice | death eligible offense, black defendant & white victim) = 0.453
p( death notice | death eligible offense, all other victim- offender race combinations) = 0.255
N = 1,153 ( 158 Missing Cases); χ ² w/ 1 df = 38.221; p < .05
Test # 2
p( death notice sticks | death eligible offense, black defendant & white victim) = 0.708
p( death notice sticks | death eligible offense, all other victim- offender race combinations) = 0.562
N = 346 ( 7 Missing Cases); χ ² w/ 1 df = 7.078; p < .05
Test # 3
p( penalty trial | death notice sticks, black defendant & white victim) = 0.906
p( penalty trial | death notice sticks, all other victim- offender race combinations) = 0.811
N = 212 ( 1 Missing Case); χ ² w/ 1 df = 3.565; p > .05
Test # 4
p( death sentence | penalty trial, black defendant & white victim) = 0.494
p( death sentence | penalty trial, all other victim- offender race combinations) = 0.369
N = 180; χ ² w/ 1 df = 2.803; p > .05
Test # 5
p( death sentence | death eligible case, black defendant & white victim) = 0.143
p( death sentence | death eligible case, all other victim- offender race combinations) = 0.043
N = 1,153 ( 158 Missing Cases); χ ² w/ 1 df = 33.549; p < .05
Note: any case with at least one white victim is defined as a “ white victim” case.
} 0.453- 0.255 = 0.198
} 0.708- 0.562 = 0.146
} 0.906- 0.811 = 0.095
} 0.494- 0.369 = 0.125
} 0.143- 0.043 = 0.100
Table 3D
Processing of White Defendant- Black Victim Cases at Various Stages of the Maryland Death Penalty System
Test # 1
p( death notice | death eligible offense, white defendant & black victim) = 0.500
p( death notice | death eligible offense, all other victim- offender race combinations) = 0.296
N = 1,153 ( 158 Missing Cases); χ ² w/ 1 df = 4.268; p < .05
Note: there are only 22 cases with a white defendant and a black victim in this test.
Test # 2
p( death notice sticks | death eligible offense, white defendant & black victim) = 0.273
p( death notice sticks | death eligible offense, all other victim- offender race combinations) = 0.624
N = 346 ( 7 Missing Cases); χ ² w/ 1 df = 5.534; p < .05
Note: there are only 11 cases with a white defendant and a black victim in this test.
Test # 3
p( penalty trial | death notice sticks, white defendant & black victim) = 0.333
p( penalty trial | death notice sticks, all other victim- offender race combinations) = 0.857
N = 212 ( 1 Missing Case); χ ² w/ 1 df = 6.315; p < .05
Note: there are only 3 cases with a white defendant and a black victim in this test.
Test # 4
p( death sentence | penalty trial, white defendant & black victim) = 1.000
p( death sentence | penalty trial, all other victim- offender race combinations) = 0.419
N = 180; χ ² w/ 1 df = 1.376; p > .05
Note: there is only one case with a white defendant and a black victim in this test.
Test # 5
p( death sentence | death eligible case, white defendant & black victim) = 0.046
p( death sentence | death eligible case, all other victim- offender race combinations) = 0.066
N = 1,153 ( 158 Missing Cases); χ ² w/ 1 df = 0.153; p > .05
Note: only one case with a white defendant and a black victim receives the death penalty.
Note: any case with no white victims and at least one black victim is defined as a “ black victim” case.
} 0.500- 0.296 = 0.204
} 0.273- 0.624 = - 0.351
} 0.333- 0.857 = - 0.524
} 1.0- 0.419 = 0.581
} 0.046- 0.066 = - 0.020
Table 4
County Processing Patterns at Various Stages of the Maryland Death Penalty System ( N = 1,305; 6 Cases Missing)
County
p( Death Notice |
Death Eligible Case)
p( Notice Sticks |
Death Notice)
p( Penalty Trial |
Notice Sticks)
p( Death Sentence |
Penalty Trial)
p( Death Sentence |
Death Eligible Case)
Unconditional
Prevalence
Anne Arundel
Baltimore City
Baltimore County
Harford County
Montgomery County
Prince George’s County
All Other Counties
Total Number of Cases
( Not Including Cases with
Missing County Information)
χ ² with 6 degrees of freedom
* p < .05
Note: Missing cases affect the unconditional prevalence, p( Death Notice | Death Eligible Case) and p( Death Sentence | Death Eligible Case)
calculations. The other calculations have complete county information. The unconditional prevalence statistic is calculated by dividing the number
of death eligible cases in each county by the total number of cases ( N = 1,305).
0.061
0.435
0.117
0.027
0.045
0.176
0.140
1,305
0.228
0.062
0.651
0.543
0.186
0.378
0.459
1,305
299.436
0.722
0.686
0.838
0.421
0.636
0.402
0.512
353
45.227
0.923
0.750
0.904
0.750
1.000
0.743
0.837
213
9.083
0.417
0.556
0.453
0.500
0.286
0.308
0.389
180
3.856
0.063
0.018
0.224
0.086
0.034
0.035
0.077
1,305
* * 97.478 *
Table 5
Processing of Offender Race Groups at Various Stages of Maryland
Death Penalty System After Listwise Deletion of Missing Cases
Test # 1
p( death notice | death eligible offense, offender is black) = 0.242
p( death notice | death eligible offense, offender is not black) = 0.368
N = 1,184 ( 18 Missing Cases); χ ² w/ 1 df = 18.571; p < .05
Test # 2
p( death notice sticks | death notice, offender is black) = 0.581
p( death notice sticks | death notice, offender is not black) = 0.647
N = 326 ( 1 Missing Case); χ ² w/ 1 df = 1.345; p > .05
Test # 3
p( penalty trial | death notice sticks, offender is black) = 0.861
p( penalty trial | death notice sticks, offender is not black) = 0.853
N = 197 ( 1 Missing Case); χ ² w/ 1 df = 0.020; p > .05
Test # 4
p( death sentence | penalty trial, offender is black) = 0.457
p( death sentence | penalty trial, offender is not black) = 0.359
N = 169; χ ² w/ 1 df = 1.560; p > .05
Test # 5
p( death sentence | death eligible offense, offender is black) = 0.055
p( death sentence | death eligible offense, offender is not black) = 0.073
N = 1,184 ( 18 Missing Cases); χ ² w/ 1 df = 1.297; p > .05
} 0.242 - 0.368 = - 0.126
} 0.581- 0.647 = - 0.066
} 0.861- 0.853 = 0.008
} 0.457- 0.359 = 0.098
} 0.055- 0.073 = - 0.018
Table 6
Processing of Victim Race Groups at Various Stages of Maryland
Death Penalty System After Listwise Deletion of Missing Cases
Test # 1
p( death notice | death eligible offense, at least one victim is white) = 0.441
p( death notice | death eligible offense, no white victim) = 0.185
N = 1,078 ( 124 Missing Cases); χ ² w/ 1 df = 83.152; p < .05
Test # 2
p( death notice sticks | death notice, at least one victim is white) = 0.698
p( death notice sticks | death notice, no white victim) = 0.440
N = 324 ( 3 Missing Cases); χ ² w/ 1 df = 20.150; p < .05
Test # 3
p( penalty trial | death notice sticks, at least one victim is white) = 0.887
p( penalty trial | death notice sticks, no white victim) = 0.750
N = 198; χ ² w/ 1 df = 5.433; p < .05
Test # 4
p( death sentence | penalty trial, at least one victim is white) = 0.421
p( death sentence | penalty trial, no white victim) = 0.417
N = 169; χ ² w/ 1 df = 0.002; p > .05
Test # 5
p( death sentence | death eligible offense, at least one victim is white) = 0.115
p( death sentence | death eligible offense, no white victim) = 0.025
N = 1,078 ( 124 Missing Cases); χ ² w/ 1 df = 34.642; p < .05
} 0.441- 0.185 = 0.256
} 0.698- 0.440 = 0.258
} 0.887- 0.750 = 0.137
} 0.421- 0.417 = 0.004
} 0.115- 0.025 = .090
Table 7A
Processing of White Defendant- White Victim Cases at Various Stages of the Maryland
Death Penalty System After Listwise Deletion of Missing Cases
Test # 1
p( death notice | death eligible offense, white defendant & at least one white victim ) = 0.422
p( death notice | death eligible offense, all other victim- offender race combinations) = 0.270
N = 1,061 ( 141 Missing Cases); χ ² w/ 1 df = 20.287; p < .05
Test # 2
p( death notice sticks | death notice, white defendant & at least one white victim ) = 0.677
p( death notice sticks | death notice, all other victim- offender race combinations) = 0.579
N = 323 ( 4 Missing Cases); χ ² w/ 1 df = 2.776; p > .05
Test # 3
p( penalty trial | death notice sticks, white defendant & at least one white victim ) = 0.870
p( penalty trial | death notice sticks, all other victim- offender race combinations) = 0.852
N = 197 ( 1 Missing Case); χ ² w/ 1 df = 0.119; p > .05
Test # 4
p( death sentence | penalty trial, white defendant & at least one white victim ) = 0.367
p( death sentence | penalty trial, all other victim- offender race combinations) = 0.450
N = 169; χ ² w/ 1 df = 1.091; p > .05
Test # 5
p( death sentence | death eligible case, white defendant & at least one white victim ) = 0.091
p( death sentence | death eligible case, all other victim- offender race combinations) = 0.060
N = 1,061 ( 141 Cases Missing); χ ² w/ 1 df = 2.890; p > .05
Note: any case with at least one white victim is defined as a “ white victim” case.
} 0.422- 0.270 = 0.152
} 0.677- 0.579 = 0.098
} 0.870- 0.852 = 0.018
} 0.367- 0.450 = - 0.089
} 0.091- 0.060 = 0.031
Table 7B
Processing of Black Defendant- Black Victim Cases at Various Stages of the Maryland
Death Penalty System After Listwise Deletion of Missing Cases
Test # 1
p( death notice | death eligible offense, black defendant & black victim) = 0.174
p( death notice | death eligible offense, all other victim- offender race combinations) = 0.424
N = 1,061 ( 141 Missing Cases); χ ² w/ 1 df = 78.523; p < .05
Test # 2
p( death notice sticks | death eligible offense, black defendant & black victim) = 0.443
p( death notice sticks | death eligible offense, all other victim- offender race combinations) = 0.672
N = 323 ( 4 Missing Cases); χ ² w/ 1 df = 14.132; p < .05
Test # 3
p( penalty trial | death notice sticks, black defendant & black victim) = 0.769
p( penalty trial | death notice sticks, all other victim- offender race combinations) = 0.880
N = 197 ( 1 Missing Case); χ ² w/ 1 df = 3.133; p > .05
Test # 4
p( death sentence | penalty trial, black defendant & black victim) = 0.467
p( death sentence | penalty trial, all other victim- offender race combinations) = 0.410
N = 169; χ ² w/ 1 df = 0.324; p > .05
Test # 5
p( death sentence | death eligible case, black defendant & black victim) = 0.028
p( death sentence | death eligible case, all other victim- offender race combinations) = 0.103
N = 1,061 ( 141 Missing Cases); χ ² w/ 1 df = 24.024; p < .05
Note: any case with no white victims and at least one black victim is defined as a “ black victim” case.
} 0.174- 0.424 = - 0.250
} 0.443- 0.672 = - 0.229
} 0.769- 0.880 = - 0.111
} 0.467- 0.410 = 0.057
} 0.028- 0.103 = -. 075
Table 7C
Processing of Black Defendant- White Victim Cases at Various Stages of the Maryland
Death Penalty System After Listwise Deletion of Missing Cases
Test # 1
p( death notice | death eligible offense, black defendant & white victim) = 0.472
p( death notice | death eligible offense, all other victim- offender race combinations) = 0.257
N = 1,061 ( 141 Missing Cases); χ ² w/ 1 df = 40.191; p < .05
Test # 2
p( death notice sticks | death eligible offense, black defendant & white victim) = 0.712
p( death notice sticks | death eligible offense, all other victim- offender race combinations) = 0.556
N = 323 ( 4 Missing Cases); χ ² w/ 1 df = 7.367; p < .05
Test # 3
p( penalty trial | death notice sticks, black defendant & white victim) = 0.911
p( penalty trial | death notice sticks, all other victim- offender race combinations) = 0.822
N = 197 ( 1 Missing Case); χ ² w/ 1 df = 3.099; p > .05
Test # 4
p( death sentence | penalty trial, black defendant & white victim) = 0.472
p( death sentence | penalty trial, all other victim- offender race combinations) = 0.381
N = 169; χ ² w/ 1 df = 1.398; p > .05
Test # 5
p( death sentence | death eligible case, black defendant & white victim) = 0.145
p( death sentence | death eligible case, all other victim- offender race combinations) = 0.045
N = 1,061 ( 141 Missing Cases); χ ² w/ 1 df = 29.234; p < .05
Note: any case with at least one white victim is defined as a “ white victim” case.
} 0.472- 0.257 = 0.215
} 0.712- 0.556 = 0.156
} 0.911- 0.822 = 0.089
} 0.472- 0.381 = 0.091
} 0.145- 0.045 = 0.100
Table 7D
Processing of White Defendant- Black Victim Cases at Various Stages of the Maryland
Death Penalty System After Listwise Deletion of Missing Cases
Test # 1
p( death notice | death eligible offense, white defendant & black victim) = 0.474
p( death notice | death eligible offense, all other victim- offender race combinations) = 0.301
N = 1,061 ( 141 Missing Cases); χ ² w/ 1 df = 2.617; p > .05
Note: there are only 19 cases with a white defendant and a black victim in this test.
Test # 2
p( death notice sticks | death eligible offense, white defendant & black victim) = 0.333
p( death notice sticks | death eligible offense, all other victim- offender race combinations) = 0.618
N = 323 ( 4 Missing Cases); χ ² w/ 1 df = 2.977; p > .05
Note: there are only 9 cases with a white defendant and a black victim in this test.
Test # 3
p( penalty trial | death notice sticks, white defendant & black victim) = 0.333
p( penalty trial | death notice sticks, all other victim- offender race combinations) = 0.866
N = 197 ( 1 Missing Case); χ ² w/ 1 df = 6.874; p < .05
Note: there are only 3 cases with a white defendant and a black victim in this test.
Test # 4
p( death sentence | penalty trial, white defendant & black victim) = 1.000
p( death sentence | penalty trial, all other victim- offender race combinations) = 0.417
N = 169; χ ² w/ 1 df = 1.389; p > .05
Note: there is only one case with a white defendant and a black victim in this test.
Test # 5
p( death sentence | death eligible case, white defendant & black victim) = 0.053
p( death sentence | death eligible case, all other victim- offender race combinations) = 0.067
N = 1,061 ( 141 Missing Cases); χ ² w/ 1 df = 0.063; p > .05
Note: only one case with a white defendant and a black victim receives the death penalty.
Note: any case with no white victims and at least one black victim is defined as a “ black victim” case.
} 0.474- 0.301 = 0.173
} 0.333- 0.618 = - 0.285
} 0.333- 0.866 = - 0.533
} 1.0- 0.417 = 0.583
} 0.053- 0.067 = - 0.014
Table 8
County Processing Patterns at Various Stages of the Maryland
Death Penalty System After Listwise Deletion of Missing Cases ( N = 1,202)
County
p( Death Notice |
Death Eligible Case)
p( Notice Sticks |
Death Notice)
p( Penalty Trial |
Notice Sticks)
p( Death Sentence |
Penalty Trial)
p( Death Sentence |
Death Eligible Case)
Unconditional
Prevalence
Anne Arundel
Baltimore City
Baltimore County
Harford County
Montgomery County
Prince George’s County
All Other Counties
Total Number of Cases
χ ² with 6 degrees of freedom
* p < .05
Note: The unconditional prevalence statistic is calculated by dividing the number of death eligible cases in each county by the total number of cases ( N = 1,202).
0.064
0.424
0.118
0.027
0.046
0.179
0.142
1,202
0.234
0.055
0.662
0.531
0.200
0.367
0.468
1,202
286.296
0.722
0.643
0.840
0.471
0.636
0.392
0.525
327
41.436
0.923
0.778
0.911
0.750
1.000
0.742
0.833
198
8.557
0.417
0.500
0.458
0.500
0.286
0.348
0.371
169
2.309
0.065
0.014
0.232
0.094
0.036
0.037
0.076
1,202
* * 99.604 *
Table 9
List of Covariates Used For Statistical Analysis
Covariates Studied For Universe of Death Eligible Cases ( N = 1,202) Mean/ Proportion
1. Number of prior violent felony convictions ( 0,1,2,3+) 0.558
2. Multiple victim case 0.183
3. Any of the victims a stranger to defendant 0.414
4. Any of the victims has a criminal history 0.083
5. Defendant has a history of alcohol abuse 0.339
6. Defendant has a history of drug abuse 0.506
7. Defendant has history of mental illness/ emotional problems 0.217
8. Defendant under age 21 at time of offense 0.265
9. Defendant over age 60 0.005
10. Defendant unable to control conduct due to alcohol/ drugs 0.289
11. Defendant unable to control conduct due to mental/ emotional problems 0.127
12. Defendant under control/ influence of another person 0.080
13. Defendant’s participation in crime was minor 0.022
14. Defendant claims killing was accidental 0.067
15. Defendant was physically abused as a child 0.113
16. Defendant was sexually abused as a child 0.043
17. Defendant had generally good character 0.075
18. Defendant had trouble in school 0.504
19. Defendant had trouble holding a job 0.385
20. Defendant has history of physical abuse as a child 0.116
21. Defendant has history of sexual abuse as a child 0.043
22. Defendant has spouse and/ or family 0.285
23. Defendant admitted crime 0.334
24. Defendant expressed remorse for crime 0.126
25. Defendant has history of mental illness/ emotional problems 0.205
26. Defendant has history of drug or alcohol use/ abuse 0.512
27. Defendant has an organic brain disorder 0.027
28. Defendant maintains innocence 0.427
29. Defendant has no major criminal history 0.240
30. Defendant aided or assisted the victim 0.004
31. Defendant surrendered within 24 hours 0.032
32. Defendant was not the actual killer 0.037
33. Defendant lay in wait for/ ambushed the victim 0.363
34. Defendant showed no remorse for the killing 0.116
35. Defendant expressed pleasure at the killing 0.031
36. Defendant alleged to have committed additional crimes contemporaneously 0.297
37. Defendant actively evaded arrest 0.095
38. Defendant was a fugitive for a prior violent crime 0.012
39. Defendant escaped from custody 0.017
40. Defendant implicated in other killing( s) 0.052
41. Defendant interfered with judicial process 0.017
42. Defendant has previously threatened/ attempted to kill victim 0.017
43. Defendant threatened victim in front of family 0.036
44. Defendant threatened other family members 0.028
45. Defendant threatened to kill victim in advance 0.021
46. Defendant abandoned victim who might otherwise have lived 0.075
47. Defendant persisted in attack even after death was certain 0.140
48. Defendant forced his/ her way into place of murder of any of the victims 0.176
49. Weapon brought to the murder scene of any of the victims 0.659
50. Any of the victims killed with a bizarre or unusual weapon 0.097
Table 9 ( continued)
List of Covariates Used For Statistical Analysis
Covariates Studied For Universe of Death Eligible Cases ( N = 1,202) Mean/ Proportion
51. Any of the victims forced to beg/ plead for their lives 0.069
52. Any of the victims’ murder planned for more than five minutes 0.282
53. Any of the victims offered no resistance to killer 0.240
54. Any of the victims not clothed or in bedclothes at time of killing 0.170
55. Any of the victims suffered multiple trauma 0.212
56. Any of the victims bound/ gagged or otherwise restrained 0.121
57. Any of the victims forced to do something against their will 0.166
58. Any of the victims held hostage prior to killing 0.037
59. Any of the victims tortured or mutilated before killing 0.056
60. Any of the victims mutilated after killing 0.034
61. Any of the victims brutally clubbed, beaten, stomped on 0.146
62. Any of the victims shot more than one time 0.265
63. Any of the victims shot in the face 0.089
64. Any of the victims killed execution style 0.129
65. Defendant tried to hide or dispose of bodies of any of the victims 0.130
66. Defendant lay in wait for any of the victims 0.098
67. Any of the victims stabbed many times or had throat slashed 0.183
68. There was another victim that was injured but not killed by defendant 0.131
69. Any of the victims killed in front of another person ( not co- defendant) 0.339
70. Crime scene was described as a bloody mess or particularly gruesome 0.121
71. Any of the victims’ murder took a long time to complete 0.087
72. Physical details of crime are unusually repulsive/ horrific 0.044
73. Any of the victims bedridden or physically handicapped 0.021
74. Any of the victims mentally/ emotionally impaired 0.004
75. Any of the victims defenseless due to youth 0.057
76. Any of the victims defenseless due to advanced age 0.115
77. Any of the victims pregnant 0.009
78. Any of the victims asleep, just awakened or in bedroom 0.116
79. Any of the victims in own house when defendant intruded 0.247
80. Any of the victims defenseless due to gross size/ strength disparity 0.131
81. Any of the victims defenseless due to intoxication 0.082
82. Any of the victims defenseless due to frail condition/ illness 0.033
83. Any of the victims have children or grandchildren 0.225
84. Any of the victims killed after kidnapping/ abduction 0.060
85. Any of the victims verbally/ physically mistreated prior to killing 0.336
86. Any of the victims dismembered before killing 0.004
87. Any of the victims mutilated in some way before killing 0.030
88. Any of the victims sexually abused before killing 0.094
89. Any of the victims burned before killing 0.026
90. Defendant slapped, kicked, or punched any of the victims before death 0.165
91. Any of the victims thrown in a body of water before being killed 0.004
92. Any of the victims subjected to unknown form of abuse before killing 0.022
93. Any of the victims dismembered after being killed 0.004
94. Any of the victims mutilated after being killed 0.017
95. Any of the victims sexually abused after killing 0.018
96. Any of the victims burned after killing 0.030
97. Defendant slapped, kicked, or punched any of the victims after death 0.003
98. Any of the victims put in the trash or dump after death 0.015
Table 9 ( continued)
List of Covariates Used For Statistical Analysis
Covariates Studied For Universe of Death Eligible Cases ( N = 1,202) Mean/ Proportion
99. Any of the victims thrown in a body of water after being killed 0.020
100. Any of the victims subjected to unknown form of abuse after killing 0.008
101. Defendant made full confession to first- degree murder to police 0.182
102. Defendant made full confession to second- degree murder 0.063
103. Defendant made full confession to aggravating circumstances 0.151
104. Defendant made partial/ qualified confession to first- degree murder 0.025
105. Defendant made partial/ qualified confession to second- degree murder 0.045
106. Defendant made partial/ qualified confession to aggravating circumstances 0.047
107. One eyewitness to the event testified 0.259
108. More than one eyewitness to the event testified 0.209
109. Physical evidence linking defendant to the crime was present 0.256
110. An informant or jail- house snitch testified against defendant 0.092
111. Defense claims case is based on circumstantial evidence 0.022
112. Defense claims state’s burden of proof not met 0.020
Statutory Aggravating Factors ( Death- Noticed Cases Only; N = 327) Mean/ Proportion
113. Victim was a law enforcement officer 0.055
114. Murder committed while defendant was in an institution 0.043
115. Murder committed in effort to evade capture by authorities 0.037
116. Murder committed in course of kidnapping 0.156
117. Victim was a child under the age of 12 0.000
118. Defendant carried out a contract killing 0.049
119. Defendant solicited killing 0.018
120. Defendant was serving a sentence of life imprisonment or death 0.012
121. Multiple victim murder 0.205
122. Murder committed along with carjacking/ robbery/ rape/ arson 0.810
123. Number of Statutory Aggravating Factors Present ( 1, 2, 3+) 1.544
Table 10A
Logistic Regression Model Estimating Effects of County on Notice Decision ( N = 1,202)
Covariate Coefficient χ ²
Intercept
Number of prior violent felony convictions
Multiple victim case
Any of the victims a stranger to defendant
Defendant has a history of alcohol abuse
Defendant under age 21 at time of offense
Defendant unable to control conduct due to mental/ emotional problems
Defendant actively evaded arrest
Defendant persisted in attack even after death was certain
Any of the victims forced to beg/ plead for their lives
Any of the victims killed execution style
Any of the victims’ murder took a long time to complete
Any of the victims in own house when defendant intruded
Any of the victims defenseless due to frail condition/ illness
Any of the victims sexually abused before killing
Defendant made full confession to aggravating circumstances
Defense claims case is based on circumstantial evidence
County = Anne Arundel
County = Baltimore City
County = Baltimore County
County = Harford
County = Montgomery
County = Prince George’s
County = Other Counties ( Reference Category)
Type III Test For County Effect ( χ ² ( 6) = 191.92; p < .05)
Log- likelihood ( Reduced Model; 23 Parameter Estimates)
Log- likelihood ( Full Model; 76 Parameter Estimates)
Likelihood Ratio Test of Full v. Reduced Model w/ 53 df
- 0.635
0.152
0.636
0.626
- 0.436
- 0.615
0.810
0.449
0.704
0.611
0.436
- 0.691
- 0.307
1.511
0.844
0.523
- 0.151
- 1.306
- 2.948
0.565
0.076
- 1.943
- 0.409
--------
- 490.13
- 466.67
46.92
8.94
2.99
8.68
14.14
5.73
9.72
11.55
2.92
9.19
3.69
3.27
5.39
2.54
10.76
9.93
5.77
0.10
14.60
117.32
4.78
0.03
20.61
3.02
*
***** *
*
***
***
*
NS
* p < .05
Table 10B
Logistic Regression Model Estimating Effects
of County on “ Notice Sticks” Decision ( N = 327)
Covariate Coefficient χ ²
Intercept
Any prior violent felony convictions
Defendant has a history of alcohol abuse
Defendant maintains innocence
Defendant expressed pleasure at the killing
Defendant forced his/ her way into place of murder of any of the victims
Any of the victims defenseless due to advanced age
Any of the victims defenseless due to gross size/ strength disparity
Victim was a law enforcement officer
County = Anne Arundel
County = Baltimore City
County = Baltimore County
County = Harford
County = Montgomery
County = Prince George’s
County = Other Counties ( Reference Category)
Type III Test For County Effect ( χ ² ( 6) = 26.97; p < .05)
Log- likelihood ( Reduced Model; 15 Parameter Estimates)
Log- likelihood ( Full Model; 64 Parameter Estimates)
Likelihood Ratio Test of Full v. Reduced Model w/ 49 df
- 0.956
0.631
0.122
0.938
1.906
1.123
0.960
- 0.568
2.911
1.036
0.364
1.542
0.008
0.070
- 0.375
--------
- 165.67
- 146.91
37.52
8.85
4.14
0.18
10.26
5.27
9.19
4.85
2.38
7.40
2.75
15.31
0.47
0.00
0.01
1.00
**
****
*
*
NS
* p < .05
Table 10C
Logistic Regression Model Estimating Effects of
County on Whether Case Advances to a Penalty Trial ( N = 198)
Covariate Coefficient χ ²
Intercept
Defendant has history of mental illness/ emotional problems
Defendant expressed remorse for crime
Defendant maintains innocence
Defendant alleged to have committed additional crimes contemporaneously
Physical evidence linking defendant to the crime was present
County = Anne Arundel
County = Baltimore City
County = Baltimore County
County = Harford
County = Montgomery
County = Prince George’s
County = Other Counties ( Reference Category)
Type III Test For County Effect ( χ ² ( 5) = 4.00; p > .05)
Log- likelihood ( Reduced Model; 11 Parameter Estimates)
Log- likelihood ( Full Model; 39 Parameter Estimates)
Likelihood Ratio Test of Full v. Reduced Model w/ 28 df
- 0.135
1.463
0.679
1.842
1.292
1.466
1.018
0.020
0.567
- 0.579
-------
- 0.598
-------
- 61.00
- 46.53
28.94
0.06
4.49
0.88
11.74
4.88
4.54
0.69
0.00
0.75
0.25
0.77
*
***
NS
* p < .05
Table 10D
Logistic Regression Model Estimating Effects of
County on Whether Defendant Receives a Death Sentence ( N = 169)
Covariate Coefficient χ ²
Intercept
Multiple victim case
Defendant has spouse and/ or family
Defendant expressed remorse for crime
Defendant alleged to have committed additional crimes contemporaneously
Any of the victims offered no resistance to killer
Any of the victims bed