1;..v v

.4—

A);
...J a!
at);

3‘s};

: ....7.:..
:1. : ...:
9,. unfit... . .
. é
. .2.. ....
...»... ..........?..$tfi.. 139%.
.Q.z§i.:. u

.u.
. .
a . 7 . .
A... . ......u. f...
e i.
, $2.3. ...: mm:
.31.... .2. m .
4.32:...» ..tznfi .3... my.
.. . . .
. .3... “.297...er
.-.... .

. um .... .-.?

a” ..lu
i

 

 

Lilli”...
...: i .. ,
. . . I1l\.o<vll‘
1.3... ..
$3.24 .. I... .. .. .4.
grits? .....zr 19.“;
.....t; (13...: ......d... ......
+1 a}
. ...
..-. 5.6...
..2 : .. x i.
...;
4...?.-. i;-
v‘.. a!) I '9‘?
.14... q .
t S‘Klvlo
‘1 t3!0?f..lJ.-.$n

. {p .r.r.i\. .
..v .01 .. : [til 3|
(2‘13“. .. 1’13...)

u.V;....... ~$.v.

 

.|.v\“l
a :......:.i
(I! it \....v...

\

. . ,r ... :
_&§fimzfi

 

. 7v 3WHI....!L|HAH‘H.IE riders—.4.
..Huafl...” fiwwmbafiibgfifflmmfiainfihi

 

2 ' LIBRARY
21% Michigan State
University

This is to certify that the
thesis entitled

THE LEGAL PROSECUTION OF ADULT RAPE CASES
PROCESSED BY A
SEXUAL ASSAULT NURSE EXAMINER PROGRAM

presented by

Debra Ann Patterson

has been accepted towards fulfillment
of the requirements for the

MA. degree in Psychology

 

 

 

Ma‘l‘d’? Professor’s Signature
IZ '- /2 - 05'

 

Date

MSU is an Affirmative Action/Equal Opportunity Institution

-:— -------o-c-o-’-—------.--

- _.—-.-.-.-.-.-.-o-o--

-'-.-n-v-I-C-‘-I-O-O-l-0-I-O-l-o-I-o-n-u—.—

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 c:/CIRC/DateDue.indd-p.15

THE LEGAL PROSECUTION OF ADULT RAPE CASES
PROCESSED BY A
SEXUAL ASSAULT NURSE EXAMINER PROGRAM

By

Debra Ann Patterson

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTERS OF ARTS

Department of Psychology

2005

ABSTRACT

THE LEGAL PROSECUTION OF ADULT RAPE CASES
PROCESSED BY A SEXUAL ASSAULT NURSE EXAMINER PROGRAM

By
Debra Ann Patterson

Prior research has documented that many reported rapes are not prosecuted by the
legal system (termed case attrition). Most of the research on rape case attrition was
conducted in the 19803. Since then, the advent of sexual assault nurse examiner (SAN E)
programs created significant changes in the way many communities throughout the
country respond to rape including the treatment of victims, and collection of medical
forensic evidence. The purpose of this study was to retrospectively examine case attrition
in cases treated in a SANE program during a three-year period. Furthermore, this study
examines whether the presence and type of forensic medical evidence collected by the
SANEs predicts case attrition.

Results indicate that referral, prosecution, conviction, and prison sentencing rates
were higher in the focal SANE program compared to published rates in non-SANE cases.
Using logistic regression, the findings show forensic medical evidence primarily

predicted police referring and prosecutors warranting cases.

For my parents, Lori Groppi and Richard Patterson for their love and support

This study is in memory of Dave Woodford, a dedicated champion of improving the

criminal justice system

ACKNOWLEDGEMENTS

I learned many years ago that meaningful work only occurs in the company of
great women. My work, as well as my life, has been blessed with many brilliant women
who must be thanked. Prior to my leap into academia, my life was graced with two
wonderful mentors who changed my life profoundly: Debbie McPeek and Sue Coats who
taught me to think critically, to trust my instincts, and to stand up for my beliefs while
being prepared for the price of my stance. I am grateful everyday for having them in my
life. I also would like to thank two wonderful friends, Maria Penaranda Sheehy and Becki
Verkest, for seeing my potential before I could and gently pushing me on a new path.

I would like to thank my committee: Cris Sullivan, for encouraging me to step
back into academia and for her humor and gentle reminders to stop taking life too
seriously; and Deb Bybee, a statistical goddess, for knowing how to explain stats in
understandable language and for her kindness. I cannot express enough my appreciation
and thanks to Rebecca Campbell, the chair of my committee, for generously sharing her
time and expertise, her remarkable ability to know when to challenge me and when to
point out my strengths, and providing me with invaluable opportunities to guide my
development as a researcher. Furthermore, she eased my transition into academia by
valuing my expertise as a practitioner, and teaching me how to utilize my professional
experience in the research development process.

In addition, many other individuals significantly contributed to the development
of this research: thank you to Dominica Rossman Tokarski for helping with data

collection, always asking “what else can I do to help”, and for her enthusiasm about this

project; to Emily Dworkin for being the queen of coding at the speed of lightening; to
Kim Greenfelder, Therese Tobin, and Rebecca Oster for their help with the data
collection process and for their insight into the legal system; and to Renae Diegel and
Jamie Ferrell for their continual support, encouragement, and sage advice throughout this
project.

I would like to thank the ladies of suite 139 (aka. the Evalu-ettes), Adrienne
Adams, Megan Greeson, and Maria Valenti, for their encouragement and well-paced
reminders to put away the thesis and enjoy life. A special thank you must go to Lauren
Lichty who has generously provided me with friendship, belly-aching laughter when the
work hours became too long, acceptance of my eccentric behavior, challenging questions
to push my thinking, and honest critique.

Finally, I would like to thank my family for their love, encouragement, and
support: my mother, Lori Groppi, for continually standing by me regardless of the
challenge; my father, Richard Patterson, for believing in my dreams; my aunt, Shirley
Grindling, for always believing that I can; my brother, Scott Patterson, for being my
greatest friend in life; and my nephew, Marcus Patterson, who at the precious age of two
reminds me what is most important in life. There are numerous other people who
supported me during this process including Jenny Patterson, Ed Groppi, Steve and Sue

Patterson, Arlene Leich, and Erika Carr.

TABLE OF CONTENTS

LIST OF TABLES ............................................................................. viii
LIST OF FIGURES ........................................................................... ix
OVERVIEW ..................................................................................... 1
Literature Review ..................................................................... 4
Steps of rape case processing .................................................... 5
Attrition in rape case processing ................................................. 7
Factors affecting rape case processing ........................................... 9
Victim characteristics .......................................................... 10
Case Characteristics ........................................................... 13
Forensic Medical evidence ................................................... 19

An intervention to address the problems of rape case attrition: SAN E
Programs ..................................................................................................... 21
CURRENT STUDY ........................................................................... 27
Project Aims and Research Questions ................................................. 27
Research Setting and Context ......................................................... 3O
METHODS ...................................................................................... 33
Sample .................................................................................. 33
Procedures .............................................................................. 41
Measures ............................................................................... 44
Data Analyses Plan .................................................................... 48
RESULTS ....................................................................................... 50
Univariate and Multivariate Data Screening ....................................... 50
Research Question #1: Rates of Case Attrition .................................... 51
Research Question #2: Factors Predicting Case Attrition ....................... 54
Referring model .................................................................. 54
Warranting model ................................................................ 65
DISCUSSION .................................................................................. 76
Summary of Major Findings ........................................................ 76
Case attrition ...................................................................... 76
Police referring ..................................................................... 78
Warranting ........................................................................ 81
Implications of Findings ............................................................. 82
Limitations of Study ................................................................. 88
Implications for Research and Practice ............................................ 90

APPENDIX A

A Comparison of SANE and Hospital Evidence Collection Methods. . . . . . 92

vi

APPENDIX B
Adult Sexual Assault Case Coding Sheet .......................................... 98

APPENDD( C
Operational and Coding Definitions ................................................ 103

APPENDIX D
Comparison of Original Coding of Variables vs. Coding of Variables for
analyses ........................................................................................................... 108

APPENDIX E
Comparison of Published Attrition Rates of Rape Cases and Current Study. 113

REFERENCES ................................................................................. 1 14

vii

LIST OF TABLES

Table l - Demographics of Victims ........................................................
Table 2 - Demographics of Case Characteristics .........................................
Table 3 - Demographics of Forensic Medical Evidence ...................................

Table 4 - Correlations of the Predictor Variables and Outcome for the Referring
Model .................................................................................

Table 5 - Correlations of the Predictor Variables and Outcome for the Warranting
Model ................................................................................

Table 6 - Logistic Regression Analyses Predicting Police Referring Cases to
Prosecutor ............................................................................

Table 7 - Chi-Square Difference Test for Types of Predictors for Police Decisions.

Table 8 - Relationship between Law Enforcement Agencies and Case
Characteristics of Referred Cases ................................................

Table 9 - Logistic Regression Analyses Predicting Prosecutor Warranting Cases
for Prosecution ......................................................................

Table 10 - Chi-Square Difference Test for Types of Predictors for Prosecutor
Decisions ....................................................................................................

viii

36

38

40

46

47

60

62

64

70

72

LIST OF FIGURES

Figure 1 — Case Excluded from Sample Based on the Eligibility Criteria .............. 34
Figure 2 - Reasons Cases were Excluded from Eligibility Sample ..................... 35
Figure 3 - Case Attrition at Each Stage of the Legal System ............................. 53

OVERVIEW

Rape has been recognized as a national problem in need of governmental response
(Bachar & Koss, 2001 ). Epidemiological data indicate that at least 17% of women will
be raped in their adult lifetimes (Koss, Gidycz, & Wisniewski, 1987; Russell, 1984;
Tjaden & Thoennes, 1998). Both the prevalence of rape and its debilitating psychological
(e. g. posttraumatic stress) and physical health consequences (Herman, 1992; Koss, 1993;
Burgess & Holmstrom, 1974; Koss & Heslet, 1992) emphasize the need to respond to this
social problem. Although the psychological and physical health problem victims
experience are largely attributable to the rape itself, some of this distress is also due to
how the legal and medical systems respond to rape victims (Campbell, 1998; Campbell &
Raja, 1999; Campbell et al., 1999; Campbell, Wasco, Ahrens, Sefl & Barnes, 2001;
Madigan & Gamble, 1991; Martin & Powell, 1995). Rape victims often do not receive
needed services and are treated by social system personnel in ways that they experience
as upsetting and victim blaming (Campbell et al., 2001).

Studies estimate that only 18% to 44% of all reported rape cases are referred to
prosecutors by police officials (Chandler & Tomey, 1981; Crandall & Helitzer, 2003;
Galvin & Polk, 1982; Frazier & Haney, 1996; LaFree, 1980). Of those cases referred by
police, prosecutors warrant an estimated 46% to 72%. Overall, only 14% to 18% of all
reported cases are prosecuted. However, these referral and prosecution rates were
obtained in studies conducted in the 19803, and since then, there have been significant
changes in rape evidence collection and documentation procedures. In the past, rape
victims were sent to hospital emergency rooms to have forensic evidence collected. The

emergency department personnel ofien lacked training in forensic evidence collection,

which decreased their ability to collect and document all available forensic evidence
(Littel, 2001). To address these problems, some communities have established Sexual
Assault Nurse Examiner (SAN E) programs (Lang, 1999). A SANE program is staffed by
registered nurses who conduct forensic exams of rape victims (Lang, 1999). Nurses
receive specialized training in forensic evidence collection, sexual assault trauma
response, forensic techniques using specialized equipment, expert witness testimony,
assessment and documentation of injuries, identifying patterned injury, and maintenance
of chain-of-evidence (Cohen, Donahue, & Kovener, 1996; Ledray, 1998). The
emergence of SAN E programs may be changing how the criminal justice system
processes and prosecutes reported rapes.

The first purpose of this study was to examine the referral and warranting rates,
and case outcomes for rape cases examined in a SANE program. Previous studies on rape
case attrition in the criminal justice system have typically examined later stages (e. g.
warranting) or late stages (e.g. case outcomes) of case processing. This study offers a
more refined analysis by focusing not only on these later stages, but also on earlier stages
of processing, such as whether law enforcement refers cases for warranting. This step in
the process of rape prosecution is essential to understand because it is one of the first
opportunities for cases to be filtered out of the system. Because SANE programs have
created significant changes in the way many communities respond to rape, new research
is needed that re-examines the issue of case attrition within the context of these
alternative reform programs. Preliminary evidence indicates that SANE programs may
prevent case attrition and may increase the likelihood that cases will move further in the

legal system (Crandall & Helitzer, 2003; Littel, 2001).

The second purpose of this study was to examine what factors predict whether a
rape case will move forward in each step of legal prosecution (i.e., referring, warranting).
Previous studies on the factors that predict case attrition have not focused on the nuances
of forensic evidence (e. g., type of injury). Because SANE programs are especially
detailed in the collection and documentation of evidence, this study was able to examine
the role of specific types of physical and anogenital injury and other forensic evidence
(e. g. trace evidence) on cases moving forward. In addition, this study explored whether
the presence and type of forensic evidence collected by the SAN E nurses affected case
attrition. Drawing on previous literature, this study also examined whether victim
characteristics (e.g. age, race) and case characteristics (e. g. victims’ use of alcohol at the
time of the assault) predicted whether rape cases moved forward in each step of the legal
system. This study compared the predictive value of forensic medical evidence in cases
moving forward relative to victim characteristics and case characteristics. This is
important to examine because it is improbable that cases move forward based on only one

particular victim, case, or evidentiary characteristic.

LITERATURE REVIEW

Rape or sexual assault is often legally defined as: “(1) any vaginal, anal, or oral
penetration by a penis, object or other body part; (2) lack of consent, communicated with
verbal or physical signs of resistance, or if the victim is unable to consent by means of
incapacitation because of age, disability, or alcohol or drug intoxication; and (3) threat of
or actual use of force” (Giardino, Datner & Asher, 2003, p.211). The process of
reporting and prosecuting a rape in the United States criminal justice system is quite
complicated and has been the focus of a substantial body of social science research since
the 1980s. In this literature, the phrase “rape case processing” is used to refer to the
actions of law enforcement, prosecutors and judicial systems to move rape cases through
the multiple steps of the legal system (LaFree, 1980). This literature review begins with a
description of the stages involved in rape case processing. Following this contextual
information, the empirical literature on case attrition and factors that influence rape case
processing in the legal system is critically examined. This research has consistently
indicated that attrition is very high in cases of sexual assault, which has prompted many
communities throughout the United Stated to explore options for improving the
community response to rape. One such approach has been the formation of Sexual
Assault Nurse Examiner (SAN E) Programs whereby medical and legal professionals
work together to document the evidence of sexual assaults. This literature review
concludes with an exploration of how SANE programs may address the problems of rape

case attrition.

Steps of Rape Case Processing

Rape victims come into contact with the legal or medical system through many
different means. Some victims call or go directly to the police department and are
subsequently referred to a medical facility. Other victims go directly to a hospital
emergency room whereby medical personnel contact the police to make a report. Some
victims contact a crisis hotline and then are referred to a medical facility or the police.
While the first point of contact with these systems varies by jurisdiction and by victim,
the following example depicts the typical steps involved in rape case processing. The
victim’s first contact is often with a dispatched road patrol officer who takes a report and
secures any evidence of the crime. The victim is sent to the hospital for a rape evidence
exam (Martin & Powell, 1995) or to a SANE program if available. The purpose of this
exam is to identify and treat any injuries that the victim may have sustained in the assault
and to collect forensic evidence from the victim’s body. The evidence collection process
includes a visual assessment of genital and body injury for evidence of the rape and
specimen collection from the body surfaces for evidence of DNA (Lang, 1999).

At this stage in the process, a law enforcement investigator is typically assigned to
the case and responsible for gathering evidence about the case by interviewing the victim,
as well as the suspect (if known) (Galton, 1975-76). Drawing from the reports by the
patrol officer and the forensic evidence, the investigator assesses the validity of the
victim’s story (Martin & Powell, 1995) and decides whether to treat the incident as a
crime and, if so, how to classify the crime (Kerstetter, 1990). If the officer decides that

there is probable cause that a rape occurred, the case is called “founded” (Martin &

Powell, 1995). If the officer decides probable cause is not warranted the case is
reclassified as unfounded (Bouffard, 2000; Martin & Powell, 1995) and the case ceases to
move forward in the system. Throughout this review and in the current study, founding
will be called “referring” to clarify that it is only in the later stages of processing (e.g.,
trial or plea bargain) that a case is ultimately “founded.” At this stage in rape case
processing, the critical issue is whether a case is referred (i.e., sent on) to the prosecutors
for further consideration.

If the investigator makes the recommendation to refer the case, it must be
approved by his/her supervisor and then sent to the prosecutor who considers the
investigator’s advice in making a tentative filing decision before interviewing the victim
(Frohmann, 1998). The prosecutor can either decide to have a warrant issued for the
assailant (termed “warranting the case”), return the case for further investigation, or drop
the case (Galton, 1975-76). If the prosecutor determines there is sufficient evidence
documenting that a crime occurred, the case is warranted, charges are determined, and the
offender is arrested and arraigned. Some jurisdictions convene a grand jury to determine
if there is sufficient evidence for the case to go to trial. In other jurisdictions, the next
court proceeding is the preliminary hearing at which a judge determines if the legal
requirements of the crime have been met. The victim may provide brief testimony about
the rape at the preliminary hearing. Based on the information presented at the hearing,
the judge decides whether to “bound over” the case for trial or dismiss the case entirely.
If the case is bound over for trial, the offender can plea-bargain with the prosecutor for

lesser charges to avoid a trial. If the case does go to trial, the assailant will be either

found guilty or acquitted. Offenders who accept plea-bargains or are found guilty, receive
a sentence of jail/prison time or other punishment (Giardino, Datner & Asher, 2003).
Attrition in Rape Case Processing

Given the complexity of the legal system, every rape case entering the system is
not expected to be fully processed, which has generated a long-standing body of research
on case attrition. The majority of studies focusing on attrition of rape cases in the legal
system were conducted in the 19808 after many states had significantly reformed their
rape laws. The intent behind these legal reforms was to streamline rape case processing
and prevent case attrition. However, most studies ascertained that despite these reforms,
few reported rapes are referred by investigators, and even fewer cases are warranted and
prosecuted. For example, LaFree (1980) conducted one of the first comprehensive
studies examining attrition of rape cases by examining police, prosecution and court
records from 1970 to 1975. In addition, he observed and interviewed law enforcement
and prosecutors to provide further context for his quantitative findings. In a sample of
881 forcible sex offense cases, he discovered that police referred 37% of the rape cases to
the prosecutor, who warranted 47% of the referred cases. Of all the reported cases in his
sample, only 14% were prosecuted.

Subsequent studies also conducted in the 19803 post-reform era also utilized
secondary data to examine attrition rates in rape case processing and found similar results
to LaFree’s original analysis. These studies revealed that referring rates among
investigators varied from 41% (Chandler & Tomey, 1981) to 44% (Galvin & Polk, 1982).
Of those cases referred by police, the rate that prosecutors warrant cases varied between

46% (Chandler & Tomey, 1981) to 57 % (Galvin & Polk, 1982). Of cases prosecuted, the

rate at which they end with a guilty verdict or guilty plea bargain varied from 83%
(LaFree, 1980), to 92% (Galvin & Polk, 1982), to 96% (Chandler & Tomey, 1981). Of
all reported cases, only 17% (Galvin & Polk, 1982) to 18% (Chandler & Tomey, 1981)
were prosecuted.

More recent studies have continued the methodological tradition of analyzing
secondary data to examine attrition rates in rape case processing. Frazier and Haney
(1996) conducted a study examining attrition of rape cases utilizing police and
prosecution court records from 1991. With a sample of 569 criminal sexual conduct
cases, they determined that investigators referred 22% of the rape cases to the prosecutor,
who warranted 72% of the referred cases. Of all the reported cases in their sample, only
16% were prosecuted. Of the cases prosecuted, 76% ended with a guilty verdict or plea
bargain. Although the warranting rate of 72% was higher than in previous studies (which
were typically 46%-57%), the overall percentage of cases that were prosecuted was
remarkably similar (near 16%). Subsequent studies have suggested the rates
investigators refer rape cases vary from 18% (Bouffard, 2000) to 38% (Crandall &
Helitzer, 2003). Among cases referred by police, Spears, Beichner, and Davis-Frenzel
(2001) ascertained that prosecutors warranted 56% of the cases. Of the cases prosecuted
in the Spears’s et a1. study, 97% ended with a guilty verdict or guilty plea bargain.

It appears from these studies that rape cases have a high probability of resulting in
a conviction if they are warranted. That is, 83% (LaFree, 1980) to 97% (Spears et al.,
2001) of prosecuted cases end in conviction. It also appears that more rape cases are
being warranted. That is, studies from the 19808 reveal that 46% to 57% of cases were

warranted compared to 56% to 72% of cases being warranted in more recent studies.

However, there appears to have been a decline in the percentage of cases being referred.
Studies from the 19805 indicated that 37% to 44% of cases were being referred compared
to 18% to 38% of cases being referred in more recent studies. Yet, when examining all
cases reported, there has not been much change since the 1980s. For example, the 19805
studies show only 17% (Galvin & Polk, 1982) to 18% (Chandler & Tomey, 1981) of all
reported cases were prosecuted compared to 16% of reported cases being prosecuted in
more recent studies (Frazier & Haney, 1996). Unfortunately, studies examining the
attrition of rape cases through the entire legal system have not occurred since the mid-
nineties.

To date, there has been only one study that has explicitly examined how rape case
attrition impacts victims. Campbell, Wasco, Ahrens, Sefl, and Barnes (2001) interviewed
a representative sample of rape victims in a major metropolitan area and found that
victims who did not have their cases prosecuted were more likely to rate their contact
with the legal system as hurtful. They also discovered that victims who rated their contact
with the legal system as hurtful exhibited higher psychological and physical health
distress. Given that the legal system is one of the few means of justice and protection for
rape victims, and that there are distressful consequences to victims whose cases are not
referred, it is important to examine this pattern of attrition in rape processing.

Factors Affecting Rape Case Processing

As many reported rape cases are not fully processed, researchers have begun

examining what factors predict rape case attrition. To date, the literature has examined

three main factors that could influence rape case attrition: 1) victim characteristics (e. g.

race, age); 2) case characteristics (e. g. relationship between the victim and offender); and
3) forensic medical evidence (e. g. injury, trace evidence).

Victim characteristics. Even if law enforcement personnel and prosecutors believe
that a rape occurred, a case might still be rejected if officials do not think that the victim
will make a credible witness. The perception that a victim will make a credible witness
may be influenced by characteristics of the victim such as age, race and social status. For
example, previous research suggests that victim age influences case outcomes. In studies
that included victims of all ages, prosecutors were significantly less likely to file charges
if the victim was a child (Rose & Randall, 1982; Spear & Spohn, 1996, 1997). For
example, Spears and Spohn (1997) discovered that cases with adult victims were three
times more likely to be prosecuted than cases with child victims. They also indicated that
cases involving children were more likely to be dismissed due to lack of evidence while
adult cases were more likely to be dismissed when victims delayed reporting, were
engaged in behavior deemed risky, or their moral character was questioned by the
prosecutor. Rose and Randall (1982) found that investigators view adolescents and
younger women as more suspicious compared to children and elderly victims. The law
enforcement investigators believed that adolescents and younger women are more likely
to fabricate the rape in order to hide consensual sexuality activity from their parents. In
addition, the investigators also viewed this age group as having instigated the assault.
While younger women’ motivations were questioned, children and the elderly were
viewed as legitimate victims, but poor witnesses.

While some studies have included child victims in their samples, others have

focused specifically on teen and adult victims and demonstrated mixed results. For

10

example, LaFree (1981) noted that cases involving victims under the age of 18 were less
likely to be warranted. The detectives interviewed believed that date rape caSes in which
both the victim and offender were young should not result in felony prosecution. Other
studies have shown that cases involving adolescent victims are more likely to be
warranted. For example, Chandler and Tomey (1981) indicated 78% of cases with
adolescent victims were warranted compared to 53% of cases with victims in their
twenties and 42% of cases with victims in their thirties. In a more recent study, Spears et
al. (2001) determined that prosecutors were nearly four times as likely to warrant a case if
the victim was between the age of thirteen and sixteen compared to victims over the age
of sixteen.

Race of the victim is another element of rape cases that may have an impact on
legal outcomes. Anecdotal literature suggests that the criminal justice system treats the
rape of ethnic minority women less seriously than the rape of Caucasian women.
Specifically, the literature suggests that one reason these cases are treated less seriously is
that there is a lack of confidence by law enforcement in the veracity of African-American
victims (Wriggins, 1983). The majority of studies examining the role of race on rape case
outcomes have focused on the race of the offender or the racial composition of both the
victim and offender. Half of the studies illustrate that race played an influential role on
case outcomes (LaFree, 1980, 1981; Chandler & Tomey, 1981; Rose & Randall, 1982;
Bradmiller & Walters, 1985; Frohmann, 1997; Spears et al, 2001). For example, LaFree
(1980) noted that cases involving African-American offenders assaulting Caucasian
victims were more likely to be warranted. This racial composition accounted for 3% of

the variance in warranting cases. Chandler and Tomey (1981) discovered that cases were

11

more likely to be warranted if the victim was Caucasian or if the defendant was non-
Caucasian. Of 260 cases, 61% of cases involving Caucasian victims were warranted
compared to 48% of cases involving non-Caucasian victims and 31% of cases involving
Hawaiian victims. More recent studies also suggest that race plays a role in prosecutorial
decisions in warranting cases. Spears et al. (2001) found that prosecutors rejected cases
more often when the victim was a racial minority. Similarly, Campbell et al. (2001)
discovered that Caucasian women were more likely to have their cases prosecuted than
minority women.

Nonetheless, the other half of the studies show that race did not have a significant
effect on case outcomes (Kerstetter, 1990; Frazier & Haney, 1996; Spears & Spohn,
1996, 1997; Bouffard, 2000). For example, Frazier and Haney (1996) did not find a
relationship between race and cases being referred. Bouffard (2000) examined if the dyad
African-American offender and Caucasian victim played a role in case outcomes. His
findings indicated that cases involving African-American offenders and Caucasian
victims were no more, or less, likely to be referred. He also examined race for victims
and offenders separately and did not find a significant relationship using both Logit and
Probit models.

Another victim characteristic that could affect case processing is social class
status. Although few studies have examined this variable, they consistently find that
social class influences cases moving forward in the system. Rose and Randall’s (1982)
study revealed that investigators exhibited class prejudice toward those who hold lower
as well as upper class status and held stereotypes about those who live in certain

economically homogenous areas. The investigators tended to typify those living in low

12

socio-economic areas as having low morality. Despite this prejudice, the investigators
tended to look for other indices of credibility. Therefore, a victim with lower socio-
economic status who appears credible may be treated the same as a victim with middle
class status. However, indices of credibility (e.g. seeking medical attention) are heavily
class biased. In a more recent study, Frohmann (1997) found that prosecutors also hold
class prejudices. Prosecutors were less likely to view cases involving victims with lower
socio-economic status as prosecutable because they believe jurors would not view these
victims as credible.

Finally, another under-studied issue in this literature is whether the victim’s
disability status affects case processing. Despite the fact that disabled women are ten
times more likely to experience sexual victimization (Sobsey & Doe, 1991), the role of
the victim’s disability on legal outcomes has been absent in the research literature.
Currently, it is unknown if the disabled population is treated differently than the non-
disabled population by investigators when referring cases and prosecutors when
warranting cases.

Case characteristics. Alcohol and drug use by the victim is one element of rape
cases that may have an impact on legal outcomes. The majority of studies exploring the
role of victim alcohol and drug use before or during the rape have shown that alcohol and
drug use decreases the probability that a case moves forward in the legal system
(Chandler & Tomey, 1981; Spears & Spohn, 1996; Frohmann, 1997). For example,
Chandler and Tomey (1981) discovered that cases were more likely to be warranted if the
victim had not used alcohol or drugs. In a sample of 260 cases, 60% of the dismissed

cases involved victims who had been drinking whereas only 8% of the warranted cases

13

involved victims who had been drinking. Additional analyses showed that absence of
alcohol or drug use by the victim was the second most important factor in prosecutorial
decisions in warranting cases. Likewise, Spears and Spohn (1996) indicated that
prosecutors were less willing to warrant cases in which the victim was drinking or using
drugs at the time of the assault. Alcohol or drug use diminishes the victim’s credibility in
the eyes of many criminal justice system personnel. In fact, Kerstetter (1990) determined
that victims who use alcohol or drugs are less likely to have their cases move forward for
both stranger and acquaintance cases. Furthermore, Campbell (1998) discovered that
victims who were drinking at the time of the rape were four times more likely to have
their cases dropped early in the stages of the legal system. However, Frazier and Haney
(1996) did not find a significant relationship between the victims’ use of drugs and
alcohol and cases being referred in a sample of 105 cases from police and court records.
Another element of rape cases that may impact legal outcomes is whether
penetration occurred. In the 19803 most states’ definition of rape were expanded to
include anal and oral penetration in addition to vaginal penetration (Giardino et al.,
2003). Studies examining the role of penetration have focused on the presence, but not
the type of penetration in cases moving forward in the legal system, and results have been
mixed. For instance, Bradmiller and Walters (1985) found that cases involving
penetration were significantly correlated with warranting the case and accounted for 17%
of the variance in predicting cases being warranted. Similarly, Frazier and Haney (1996)
noted that cases involving penetration were somewhat more likely to be referred, but the
effect was not significant. However, Bouffard (2000) employed logistic regression and

determined that cases involving penetration decreased their probability of being referred.

14

Most of the debate in the literature regarding case characteristics has focused on
victim-offender relationship. Prior research suggests that rape cases involving
acquaintances may be assessed and treated differently than incidents involving strangers.
The major investigative concern in stranger rape cases is identifying the assailant.
Prosecutors’ decisions to warrant stranger rape cases are determined primarily by
evidentiary strength of the case. Conversely, decisions to warrant acquaintance rape cases
are determined largely by assessing the lack of consent by and the credibility of the
victim (Kerstetter & Van Winkle, 1990).

Studies exploring the role of the victim-offender relationship in rape processing
have shown mixed results. The majority of studies and anecdotal literature suggest that if
the offender is a stranger to the victim, there is a greater chance that the allegation will be
taken seriously and referred than if the offender is an acquaintance (LeGrand, 1975;
Chappel et a1, 1977; Chandler & Tomey, 1981; Rose and Randall, 1982; Bradmiller &
Walters, 1985; Kerstetter, 1990). For example, Bradmiller and Walters (1985) conducted
a secondary analysis on 89 cases from a court program and found that warranting cases
was significantly related to the relationship between the victim-offender. Specifically,
prosecutors were less likely to warrant cases in which the offender was a relative of the
victim. Hierarchical multiple regression showed that the victim-offender relationship
accounted for 4.2% of the variance. Similarly, Kerstetter (1990) indicated that stranger
rape cases are more likely to be referred and warranted. In order to refer a case,
investigators noted that they have to establish the identity of the offender, that sexual
activity occurred, and that the victim did not consent. Because identifying the offender is

not a concern in acquaintance rape cases, determining whether the victim consented

15

becomes a primary focus of the investigation. Because identifying the offender is often
easier than determining lack of consent, stranger cases are more often prosecuted.

By contrast, Frazier and Haney (1996) suggest that overall, suspects in
acquaintance rape cases are more likely to be referred than stranger rape cases. However,
when comparing only stranger rapes with identified suspects to acquaintance rapes,
stranger rape cases (66%) are more likely to be referred than acquaintance rape cases
(41%). Although acquaintance rape cases have the advantage of an identified offender,
their findings suggest they are viewed as less prosecutable. This study demonstrates that
inclusion of unidentified offenders in the analyses regarding victim-offender relationship
and case processing must be re-examined because it substantially and differentially
affects the resulting findings. Because it is impossible for the investigator to refer a case
when a suspect has not been identified, it is important to only compare stranger cases
with identified suspects to acquaintances cases.

Some studies have shown that the victim-offender relationship had no effect on
legal outcomes of rape cases (Bachman, 1998; Spohn & Homey, 1993; Spears & Spohn,
1996, 1997). For example, Bachman (1998) analyzed the National Crime Victimization
Survey dataset from 1992 to 1994 with a sample size of 348 and indicated that there was
not a statistically significant relationship between victim-offender relationship and
referring cases. The study measured victim/offender relationship as a dichotomous
variable: intimate offenders (spouse, ex-spouse, boyfriend/girlfriend and ex-boyfriend/ex-
girlfriend) and all other offenders (family members, fiiends/acquaintances and strangers).
Because investigators treat stranger and acquaintance cases differently, it is possible that

including strangers in the same category as acquaintances led to the non-significant

16

finding. Spears and Spohn (1996) analyzed 321 cases from prosecutor files and also
indicated that the victim-offender relationship did not influence the prosecutor’s decision
to warrant cases. It is unclear if their sample included unidentified stranger cases or only
the cases in which the suspect was identified.

Other studies suggest that acquaintance cases are more likely to move forward in
the system. Through interviews and analysis of prosecutor files, Spears, Beichner, and
Davis-Frenzel (2001) found that prosecutors reject stranger cases more often than
acquaintance cases. They noted that most of the victims of stranger rape in this sample
engaged in behavior likely to be questioned by the prosecutors (e. g. delayed reporting,
prostitution, walking alone at night). Interestingly, interviews conducted with prosecutors
indicated that they would be less inclined to warrant acquaintance cases. Bouffard (2000)
used logistic regression to analyze 326 cases collected from police records. He
ascertained that cases were more likely to be referred to the prosecutor if a prior victim-
offender relationship existed. However, the study also suggests a higher probability that
the prosecutor would dismiss cases with a prior victim-offender relationship. Although it
is impossible for a case to move forward with an unidentified suspect, these cases were
still included in the analyses. Because police and prosecutors cannot refer or warrant
cases without a suspect, it is possible that including cases without an identified suspect
affected the results of the study.

Offender tactics is another important element of rape cases that may have an
impact on legal outcomes. Sex offenders use various types of tactics to control victims
(Cleveland, Harrington, Koss, & Lynons, 1999). Studies examining the role of offender

tactics in legal outcomes have been limited to three types of tactics. The most common

17

tactic studied is weapon use. The majority of these studies have shown that the presence
of a weapon increased the probability that a case would move forward in the legal system
(Chandler & Tomey, 1981; LaFree 1981; Rose & Randall, 1982; Kerstetter 1990;
Campbell et al., 2001). For instance, Chandler and Tomey (1981) indicated that cases
were more likely to be warranted if a weapon was used. In a sample of 260 cases, 60% of
the cases involving a weapon were warranted while only 5% of the cases involving a
weapon were dismissed. Similarly, Rose and Randall (1982) found through observations
of 610 police investigations that weapon use affected the investigators’ judgment about
victim credibility. Investigators were more likely to refer cases in which they doubted the
victims’ credibility if the case involved weapon use. Because victim credibility is ofien
doubted in rape cases involving acquaintances, weapon use may be related to moving
acquaintance rape cases through the legal system. In fact, Kerstetter (1990) revealed that
weapon use was the sole significant factor in the prosecutorial decision in warranting
acquaintance rape cases. Weapon use was particularly important to cases in which the
offender claimed that the victim consented to the sexual activity. In a more recent study,
Campbell et al. (2001) found through interviewing victims that rape cases involving
weapons were significantly more likely to be prosecuted than rape cases that did not
involve weapon use. However, Frazier and Haney’s (1996) study revealed that weapon
use had no effect on cases being referred by investigators. Similarly, Spears and Spohn
(1997) indicated that weapon use did not affect the prosecutor’s decision to charge or not.
The second type of perpetrator tactic that has been examined in the literature is
the use of force. These studies show that the use of force increases the probability that a

case will move forward in the legal system (Rose & Randall, 1982; Bradmiller &

18

Walters, 1985; Martin & Powell, 1995). For example, Rose and Randall (1982) found
that the use of force was important in demonstrating to the investigators that the victim
did not consent. Correspondingly, Bradmiller and Walters (1985) discovered that the
amount of force was significantly correlated with warranting cases. In addition, amount
of force along with the age of the victim accounted for 32.7% of the variance. They
determined that amount of force was a more powerful predictor than age of the victim.

A less commonly studied tactic is the use of threats. Frazier and Haney (1996)
noted that cases were more likely to be referred if the offender threatened the victims.
However, Spears and Spohn (1997) ascertained that both threats and force did not
significantly affect prosecutorial decisions on warranting.

Finally, another under-studied issue in this literature is whether the length of time
between the rape and exam affects case processing. Currently, it is unknown if cases in
which the victim immediately seeks a forensic medical exam are more likely to be
forwarded than cases in which the victim waits a couple of days before seeking a forensic
medical exam.

Medical evidence. The term “medical evidence” is ofien used to encompass
evidence that can be collected from or observed on the body of the victim, including
injury, and trace evidence. Medical evidence includes the identification and
documentation of injury such as physical and anogenital injury. Although the law no
longer requires proof of resistance, many judges and jurors still view injury as necessary
proof the victim did not consent (Giardino et al., 2003). Finally, medical evidence also
includes trace evidence such as foreign material found on the victim’s body. Trace

evidence may place the offender at the crime scene or corroborate the victim’s story. The

19

presence or absence of medical evidence could impact the chance of cases moving
forward in the system.

Most studies suggest that medical evidence, in particular documented injuries, is
predictive of cases moving further through the criminal justice system (Chandler &
Tomey, 1981; Feldman-Summers & Palmers, 1980; Frazier & Haney, 1996; Galton,
1976; Kerstetter, 1990; Martin & Powell, 1995; Rose & Randall, 1982; Spohn & Spears,
1997; Spohn, Beichner, & Davis-Frenzal, 2001). For example, Galton (1976) provided
investigators with a questionnaire focusing on six hypothetical rape cases that were based
on actual reports. The investigators were asked to suggest a disposition of the cases and
indicate the facts most significant to the case. The investigators noted that they expected
to see bruises or lacerations on the victims’ body, which indicated to them that she
resisted. However, mere surface scratches were not enough for the investigators to
suggest referring the cases. Similarly, Feldman-Summers and Palmer (1980) surveyed
prosecutors, judges and staff of two rape crisis centers and revealed that rape-processing
officials characterize false complaints as those that lack physical injury.

The majority of studies examining medical evidence have analyzed police,
prosecutor or court records. For example, Rose and Randall (1982) analyzed 610 police
reports received by one police department during a six-month period. They found that
investigators perceived legitimate rape cases as having corroborating physical evidence
such as injury or torn clothing. Further, if the victim’s credibility was questioned, the
case had a higher chance of being referred if corroborating physical evidence existed.
Evidence of physical force helped to demonstrate non-consent to investigators. These

findings suggest that physical evidence plays an important role in referring acquaintance

20

rape cases in which lack of consent is often questioned. In addition, they discovered that
the more severe the injury, the more serious the investigators perceived the crime. Other
researchers have shown similar evidence regarding medical evidence. For example, cases
were more likely to be warranted if the victim sustained injury (Chandler & Tomey,
1981; Spears et al., 2001) or when any physical evidence was present (Frazier & Haney,
1996). However, Spears and Spohn (1997) found that prosecutors were somewhat more
likely to warrant cases when physical evidence was present although their findings did
not reach statistical significance; the trend was in the expected direction.

Other studies in this body of literature have used observations and interviews with
social system personnel to examine the role of medical evidence in moving cases
forward. For example, Kerstetter (1990) interviewed and observed twenty detectives and
discovered that the police are more likely to refer and treat the case seriously if the victim
suffered injuries to her sex organs. Using similar methodology, Martin and Powell (1995)
determined that prosecutors prosecuted cases more aggressively if the victim was
physically injured and/or force could be demonstrated. They also discovered that law
enforcement concentrate on building their cases by collecting and verifying evidence
rather than accepting victims’ accounts. This suggests that evidence has more weight in
moving cases forward than the victim’s testimony of the rape.

Finally, an under-studied area in the literature is whether the identification of
trace evidence affects case processing. Presently, it is unknown if identifying and
documenting trace evidence from the victim’s body or clothing increases the probability

of cases being referred and warranted.

21

An Intervention to Address the Problems of Rape Case Attrition: SANE Programs

As noted previously, many of the studies on rape case processing were conducted
in the 19808 and early 1990s. Since that time, there have been significant changes in the
way many communities throughout the United States respond to rape victims. The
advent of sexual assault nurse examiner (SANE) programs created a fundamental shifi in
many aspects of rape processing including the response to victims, the collection and
documentation of forensic medical evidence, preservation of chain of custody,
availability of expert witnesses for trial and consultation, and increased collaboration,
communication and cross training among key agents in rape case processing. In seeking
new ways to treat victims and collect evidence with more sensitivity and effectiveness,
some communities have established sexual assault nurse examiner (SANE) programs
(Lang, 1999). A SANE program is staffed by registered nurses or nurse practitioners who
conduct a forensic exam of the sexual assault victim (Lang, 1999). The clinician receives
specialized training in forensic evidence collection, sexual assault trauma response,
forensic techniques using specialized equipment, expert witness testimony, assessment of
injuries, identifying patterned injury, documentation of injuries, and maintenance of
chain-of-evidence (Cohen et al., 1996; Ledray, 1998).

The goal of SANE programs is to provide supportive care to victims while
improving prosecution of suspects (Amdt, 1988). SANE programs attempt to improve
prosecution by: 1) providing better-quality forensic evidence collection with specially
trained personnel and specialized forensic equipment; 2) providing expert witness

testimony by personnel trained in providing this testimony; and 3) providing training and

22

case consultation for law enforcement officials to increase their understanding of medical
evidence findings (Littel, 2001).

Forensic evidence exams conducted by SANE programs include a medical history
of the victim that guides the evidence collection, properly packaging the victim’s
clothing, conducting a head to toe assessment and specimen collection from the body
surfaces (e.g., hair), visual assessment of genital trauma, body fluid and orifice specimen
collection (Lang 1999) (see Appendix A for a detailed comparison of exam procedures
for hospitals and SAN E programs). Unlike hospitals, most SANE programs utilize
specialized forensic equipment and techniques, such as a colposcope, which is a lighted
magnifying instrument that greatly increases the identification of anogenital trauma
(Slaughter & Brown, 1992). With a colposcope, a medical practitioner is three times
more likely to detect microlacerations, bruises and other injuries in rape victims (V oelker,
1996). A camera can be attached to the colposcope to photodocument genital injuries
(Lang, 1999). Some SANE programs also use Toluidine Blue dye in the detection of
genital trauma by enhancing the visualization of microlacerations (Cohen et al., 1996;
Ledray, 1998). Toulidine Blue dye is a nuclear stain that adheres to cellular material and
enhances the identification of injury. The dye causes the injured area to darken
significantly, making visualization of an injury much easier (Giardino, 2003). McCauley,
Guzinski, Welch, Gorman, and Osmers (1987) found tears in only 4% of cases using
gross visualization, but after Toluidine Blue was applied, the rate increased to 58%. Some
SAN E programs utilize an alternative light source which can detect bodily fluids that

may not be detected with visual inspection alone (Giardino, 2003).

23

Studies have shown that the quality of the forensic medical evidence collected by
SANE nurses is significantly more complete than that obtained by hospital personnel and
that the documentation of injury and maintenance of chain of custody were better in
SANE-collected kits compared to non-SANE kits (Ledray & Sirnmelink, 1997; Sievers,
Murphy & Miller, 2003). Sievers et al. (2003) determined that 92 % of SANE-collected
cases had preserved the chain of evidence compared to 81% by hospital personnel.
Derhammer, Lucente, Reed, and Young (2000) indicated that 92% of SAN E cases
preserved the chain of evidence compared to 68% percent. Ledray and Simmelink (1997)
discovered that SANE preserved the chain of evidence 100 percent of the time while the
hospital only properly sealed 97% of the kits, only documented the appropriate
information on 73% of the rape kit boxes, and only properly stored the victims’ clothing
in 88% of the cases.

With respect to the completeness of rape evidence kits, Derhammer et a1. (2000)
revealed that 95% of the SAN E cases had complete evidence collection compared to 71%
of hospital cases. Ledray and Simmelink (1997) noted that SANE programs were more
likely to collect swabs that matched the documented orifice penetration (96% of SAN E
compared to 85% by the hospital). Kits were more likely to contain an extra tube of blood
for alcohol and/or drug analysis (100% of SANE cases compared to 15% of hospital
cases). In addition, SANE programs included proper documentation in the rape kit (100%
compared to 79% in hospital cases). Sievers et al. (2003) found that the rape evidence
kits processed by SAN E programs contained the right amount of swabs, pubic hair and

head hair. In addition they discovered that the rape evidence kits processed by SANE

24

were more likely to have complete documentation (6. g. specimen envelopes were
labeled).

To date, research on how SAN E programs may impact rape case processing has
focused on these issues of evidentiary completeness and there has only been one study
that examined attrition of rape cases before and after the implementation of a SANE
program. Crandall and Helitzer (2003) examined 247 rape cases that were processed by
the hospital and noted that 38% of the cases were referred for warranting. They also
examined 715 rape cases processed by the SANE program and revealed that 50 percent of
the rape cases were referred. In addition, they ascertained through interviews that
prosecution and law enforcement believed that the SAN E program had significantly
improved evidence collection, storage, and chain of custody. Law enforcement indicated
that communication and collaboration had improved with medical personnel after the
advent of SANE. Law enforcement noted that SANE contributed to an increased
efficiency of their work environment which allowed law enforcement more time to
investigate cases. For example, law enforcement obtained medical records quicker and
less time was spent waiting for the completion of the exam so they could transport the
rape kit. Prosecution personnel believed that the evidence collection by the SAN E
program was more reliable, easier to read, and more effective in court. Furthermore, they
felt that the SANE program was more conducive to cooperation with law enforcement.
That is, victims examined at SAN B were more willing to talk to law enforcement because
they had been treated with sensitivity and respect. In fact, the study found that more
victims reported to police when examined at a SANE compared to the hospital (67% vs.

46%, p<0.0001). This preliminary evidence suggests that SANE programs may be an

25

effective intervention in addressing the problems of rape case attrition However, it
remains unknown if SANE programs are an effective intervention for all rape cases or

those with injuries or certain types of injuries.

26

CURRENT STUDY

Project Aims and Research Questions

T he first aim of the proposed project was to examine the rates of referral,
warranting, and case outcomes for rape cases examined in a SANE program. Most prior
research on rape case attrition in rape cases was conducted in the 19803 and reported that
only 16% (Frazier & Haney, 1996) to 18 % (Chandler & Tomey, 1981) of all reported
cases were prosecuted. Previous studies focusing on case attrition have largely examined
later stages (6. g. warranting) or late stages (e. g. case outcomes) of case processing. This
study offers a more refined analysis by focusing on earlier stages of processing, such as
police referring cases. This step in the process of rape prosecution is important to
understand because it represents one of the first instances for cases to be filtered out of
the system. Since the earlier studies were conducted, the advent of sexual assault nurse
examiner (SANE) programs has created significant changes in the way many
communities throughout the United States respond to rape. Therefore, new research is
needed that re-examines the issue of case attrition within the context of these alternative
reform programs. Anecdotal evidence suggests that SAN E programs may prevent case
attrition and may increase the likelihood that cases will move further in the legal system
(Littel, 2001). One empirical study of a SANE program provided empirical evidence that
cases were significantly more likely to be warranted and prosecuted after implementation
of the SANE program (Crandall & Helitzer, 2003).
Question #1: To what extent do rape cases processed by a SAN E program move forward

in the legal system?

27

Objective #1 a: Patient/victim names, and assault date were obtained from one large
SAN E program in Michigan (Turning Point in Macomb County). This information was
needed to track whether cases examined at Turning Point’s SANE program were later
pursued by the legal system. Program records were obtained for the period of time from
1999 to 2002.

Objective #1 b: Obtained the rate at which rape cases examined at Turning Point’s SANE
program were actually referred by law enforcement, warranted by prosecutors, and the
final legal outcomes. The Macomb County Prosecutor’s Office maintains a database of
all cases referred to the prosecutor’s office and another database of all legal outcomes for
referred cases (e. g. warranting decisions, final disposition) , which can be searched by
victim name, assault date, and complaint numbers. Descriptive data was collected to
document how many cases were referred by the police, warranted by the prosecutor’s
office, prosecuted and resulted in conviction.

The second aim of this study was to examine whether the presence and type of
forensic evidence collected by the SAN E nurses predicts case attrition. Previous studies
on the factors that predict case attrition have not examined the nuances of forensic
evidence (6. g. type of injury). Because SANE programs are particularly detailed in the
collection and documentation of evidence, this study can examine the role of specific
types of physical and anogenital injury and other forensic evidence (e. g. trace evidence)
on cases moving forward. Previous literature has also suggested that forensic medical
evidence is one of many factors that influence whether cases move forward. Drawing on
previous literature, this study also examined whether victim characteristics (e.g. age,

race) and case characteristics (e. g. victims’ use of alcohol at time of assault) predict

28

whether rape cases move forward in each step of the legal system (warranting,
prosecution, and case outcome). This aspect of the study compared the relative predictive
value of forensic medical evidence in cases being referred and warranted compared to
victim characteristics and case characteristics. This is important because it is unlikely that
cases move forward based on only one particular evidentiary, victim, or case
characteristic.

Question #2: What factors predict case attrition? What is the predictive value of forensic
medical evidence in charging decisions relative to victim characteristics and case
characteristics?

Objective #21:: Obtained injury findings of the forensic exams such as the type and
number of specific physical and anogenital injuries from Turning Point’s SANE program
patient records (see Appendix B for a complete list of all evidence/injury types coded).
Objective #2c: Obtained descriptive data about the victims (age, gender, race/ethnicity,
disability status, address [proxy for household income]) and the rape (type of rape such as
vaginal, oral or anal, victim-offender relationship, tactic used by the offender to control
or coerce the victim, alcohol or drug use by the victim before or during the rape) (see
Appendix B for a complete list of all victim and case characteristics coded). All of these
variables are consistently documented in the SAN E patient record except victim-offender
relationship and race of the victim. Victim/offender relationship and victim race were
obtained from Turning Point’s “Sexual Assault Survivor Profile” database. This database
was searched using the victim’s name, case number and date of assault.

Objective #2c: Conducted a series of logistic regression analyses predicting investigator

referral decisions (referred/not referred), prosecutor warranting decisions and

29

(warranted/not warranted) from three sets of independent variables: victim
characteristics, case characteristics, and forensic medical evidence. Odds ratios were
examined to determine relative importance of each predictor within each model.
Research Setting and Context

The setting for this study was Turning Point’s SAN E program located in
Macomb County. This program was established in 1997 through a community task force
that sought to address the problem of low reporting and conviction rates for rape as well
as inadequate forensic evidence collection and a victim blaming response by emergency
room personnel. The task force determined that a SANE program was needed in the
community and Turning Point was selected as the host for the program because of their
10 year history of providing comprehensive services to sexual assault victims. In
September 1999, Turning Point opened the first SAN E program in Southeastern
Michigan with established agreements with all Macomb County hospitals and law
enforcement jurisdictions to transfer rape victims for evidence collection unless they need
urgent medical care. A complete forensic exam is conducted by specially trained SANE’s
for victims within 96 hours of the assault. Services are accessible 24 hours a day 7 days a
week through Turning Point’s 24-hour crisis line. A team comprised of a nurse and an
advocate is dispatched to the site by request of law enforcement, hospital or the victim.
Medical treatment, crisis intervention and follow-up services are also offered. The
program setting, based at a medical office facility, aims to meet the needs of victims by
providing a separate interview and examine rooms, a waiting area for family and friends,
and a private bathroom with shower. The facility is also equipped with state-of the-art

medical forensic equipment not generally available at hospitals.

30

The program has provided over 800 forensic exams for victims since the program
began. Analysis of agency records revealed that 97% of the victims were female and 3%
were male; 68% of the victims were Caucasian, 25% were Afi'ican American, 2% were
multiracial, 1% was Hispanic, and 4% were unknown; 8% of the victims were 0-12 years
old, 34% were 13-18 years old, 36% were 19-29 years old, 16% were 30-44 years old,
5% were 45-64 years old, and 1% were 65 years old or older. The racial demographics
of the sample are consistent with the 2000 Macomb County census data. Most victims
were residents of Macomb County (71%) and 78% of the rape cases occurred in Macomb
County.

Turning Point’s SANE program was an ideal site for addressing the research
questions for several reasons. First, Turning Point’s SAN E program is consistent with
national standards for training and supervising forensic nurse examiners, which makes
them typical among other SANE programs (IAFN, 2004). Therefore, the findings are
expected to be generalizable to other communities throughout the country. Second, the
program has been in existence long enough so that examining a substantial number of
closed cases is now possible. Third, Turning Point’s SAN B program examines the highest
number of victims annually in Michigan, providing a large sample size (Patterson, 2004).
Finally, Turning Point conducts multiple quality assurance procedures to assure that the
nurses are collecting and documenting evidence in a standard manner, providing high
reliability of records.

To assess further the appropriateness of this site for research purposes, key
informant interviews were conducted with legal personnel in Macomb County to

determine how rape cases have been processed during the timeframe of the study. There

31

have been no significant policy or personnel changes within the Macomb County
Prosecutors’ Office from 1999 to 2002 that would directly impact the processing of rape
cases. Furthermore, there have been no significant changes in the way the State of
Michigan Crime Lab processes evidence kits during the timeframe of the study.
Additionally, the training and technology used by Turning Point’s SANE program did not

significantly change during the study timeframe.

32

METHODS
Sample

Adult rape victims treated in Turning Point’s SANE program from September
1999 to December 2002 were the target sample for this study. Cases were sampled that fit
the following criteria: 1) the victim is at least 18 years old currently; 2) the victim was
age 18 or older at the time of the assault; 3) the victim was victimized in Macomb
County; 4) the victim reported the rape to a law enforcement agency in Macomb County;
and 5) the victim received a complete exam by Turning Point’s SANE program. The
focal program provided forensic services to 514 people during the timeframe of the study,
and 205 cases met the study’s eligibility criteria (see Figure 1 for reasons of case
exclusion). Of the 205 cases sampled, 20 more cases were eliminated from the sample for
multiple reasons (see Figure 2) resulting in a final sample size of 185 cases.

Descriptive information about the sample is provided in Table 1-3. In each of
these tables, descriptive information is presented for: 1) the entire sample (all reported
cases, on which police made referring decisions) (N=185); 2) the subsample of cases
police referred to the prosecutors, on which prosecutors made warranting decisions
(N =90); and 3) the subsample of cases that were warranted by prosecutors (N=57). Table
1 summarizes victim demographics. Of all reported cases, the mean age was 29.72 years,
with a range of 18 years to 89 years old (no missing data). The majority of the sample
was Caucasian (83%) (3% of missing data), and 80% did not have a disability (no

missing data).

33

Figure 1

Case Excluded from Sample Based on the Eligibility Criteria

 

514 cases examined by the SANE Program during 1999 to 2002

 

 

 

7 cases excluded because the victims were murdered by the offenders

 

 

 

32 cases excluded because the examinations were of suspects of rape cases

 

 

 

110 cases were excluded because rape did not occur in Macomb County

 

 

130 cases were excluded because the victims were under 18 years old

 

 

 

20 cases were excluded because the victims declined an exam

 

 

9 cases were excluded because the victims did not report the rape to police

 

 

 

1 case was excluded because the victim was not penetrated

 

 

 

f'__\ f—fi f \ r N f 1 f \ f 1 f W f N

205 cases met the eligibility criteria

]
l
J
J
l
l
l
l
]

 

34

Figure 2

Reasons Cases were Excluded from Eligibility Sample

 

205 eligible cases

 

 

 

5 victims were charged with false reporting

 

 

 

4 victims recanted their police report

 

 

 

7 stranger cases had unidentified offenders

 

 

2 medical records had missing sheets describing physical injuries

 

 

 

2 cases involve male victims“

 

 

 

Final sample equals 185 cases

r—fir—flr—xr—fif—jr—fir—fi

QQQQHPQ

 

‘ Charging decisions for cases involving male victims may different than cases with female victims.
Because only two cases involve male victims, analyses exploring this issue would be impossible and may
serve as a confounding variable if left in the sample. Therefore, the two male cases were removed.

35

Of the victims with a disability, 58% had a cognitive or psychiatric disability, 11% had a
physical disability (e. g. hearing impaired), 8% had both a physical disability and a
cognitive or psychiatric disability, and 22% were unknown. The mean household income
was $45,710, with a range of no income (e.g. homeless) to $111,114 (four percent of data

missing).

Table 1

Demographics of Victims (Percentage or Mean)

 

 

Demographics of Victims All Reported Referred Warranted
Cases Cases Cases
(N=185) (N=90) (N=57)
Age in years (mean) 28.72 (11.70) 29 (12.61) 28.82 (10.20)
Race/Ethnicity
Caucasian 83% 82% 78%
Minority 1 7% l 8% 22%
Has a disability 20% 19% 16%
Income (mean in thousands) $45.71 (17.46) $45.64 (16.44) $43.28 (15.9)

 

Note: Standard deviation in parentheses

Table 2 displays case (i.e. assault) characteristics for all reported cases, cases
referred by the police, and cases warranted by the prosecutors. The majority of victims
were raped by someone they knew, with 23% being raped by their partners (e. g. dating
partner, spouse, life-time partner) or family member (e. g. sibling, uncle), and 55% being
raped by acquaintances (e. g. fiiends, coworker) (no missing data). Fifty-four percent of

the victims had consumed drugs or alcohol before or during the rape (11% of data

36

missing). Interestingly, 16% of the victims were unaware of what happened to them, or
whether they were penetrated (no missing data on type of penetration). For example,
some victims woke up with sore genitals and their underwear removed from their bodies.
These cases may be related to the increased use of “date-rape” drugs in the community.
Force was used by the offender (e. g. physically controlled victim such as holding victim
down) in 72% of the cases (16% of data missing). The time between the assault and the
victim receiving an exam ranged from 2.17 hours to 97.62 hours, with an average of

17.02 hours (13% of the cases missing).

37

Table 2

Demographics of Case Characteristics (Percentages and Means)

 

 

Demographics of Case All Reported Referred Cases Warranted
Cases (N=90) Cases

Characteristics (N =1 85) (N =57)

Consumed drugs or alcohol before/ 54% 42% 36%

during assault

Type of penetration
Single 43% 40% 46%
Multiple 41% 54% 53%
Unknown by victim 16% 6% 2%

Victim/offender relationship

Stranger 22% 17% 21%

Intimate/Familial 23% 32% 35%

Acquaintance 55% 5 1 % 44%
Force was used by the offender 72% 81% 86%
Time between assault and exam 17.02 (18.20) 15.54 (17.37) 15.82 (17.67)
(hours)

 

Note: Standard deviation in parentheses

Table 3 presents the characteristics of the forensic medical evidence documented
in the medical records. The majority (71%) of victims had documented physical injury,
with a mean of 4.72 injuries, and a range of 0 to 32 physical injuries. Of the subtypes of
physical injuries, more victims had bruises (54%), while fewer had tearing (3%). F ifiy-

four percent of victims had documented anogenital injury, with a mean of 1.35 anogenital

38

injuries, and a range of 0 to 11 anogenital injuries. Of the subtypes of anogenital injuries,
more victims had redness (29%), followed by abrasions (22%), while fewer had
anogenital bruises (10%) (no data was missing on physical or anogenital injuries).
Twelve percent of the victims had documented trace evidence found on their body (no

missing data).

39

Table3

Demographics of Forensic Medical Evidence (Percentages and Means)

 

Demographics of Forensic Medical All Reported Referred Cases Warranted

 

Cases (N=90) Cases
Evidence (N =1 85) (N =5 7)
Number of physical injuries 4.72 (6.27) 4.99 (6.03) 4.91 (6.00)
Type of physical injury
Redness 22% 23% 28%
Tears 3% 2% 2%
Bruises 54% 58% 53%
Abrasions 43% 46% 49%
Other 1 8% 20% 25%
Number of anogenital injuries 1.35 (2.04) 1.66 (2.27) 1.49 (2.32)
Type of anogenital injury
Redness 29% 33% i 30%
Tears 14% 14% 12%
Bruises 10% 13% 12%
Abrasions 22% 33% 28%
Other 8% 9% 9%
Trace evidence 12% 14% 11%

Note: Standard deviation in parentheses

40

Procedures

To assess the reliability of the sampling procedures, 30% of the cases from
September 1999 to December 2002 were randomly selected and reviewed by a Turning
Point employee to determine if the same cases were selected for inclusion based on the
five criteria for eligibility (see Sample section for discussion on the five eligibility criteria
for sample inclusion). The program’s employee was asked to conduct the reliability
assessment in order to limit access of patients’ identities exclusively to the principal
investigator. The program employee and investigator had 100% agreement on cases that
were sampled for inclusion based on the five eligibility criteria. For the sampled cases,
the victims’ names, complaint numbers, and date of assault were recorded; this
information was required to search the Macomb County Prosecutors’ Office’s database
for information regarding referring and warranting decisions. A new and unique ID
number was also assigned to the case for the researcher’s records. The list linking the
research ID number and victims’ names and complaint numbers was destroyed after all
data were collected and reliability and validity assessment were complete.

The prosecutor’s database for referring was searched using complaint numbers. In
this database, a record was created for all cases referred by law enforcement to the
prosecutor’s office. In order to assess the reliability of these data, 30% of cases that were
referred, but not warranted were randomly selected and verified against law enforcement
records to verify that the case was referred. Kappa was computed to assess the degree of
agreement between the “referral” database, and law enforcement records: kappa=.52.
Because this low kappa value calls into question the reliability of the data, all cases were

checked for accuracy using law enforcement records. Kappa was computed again to

41

assess the degree of agreement between the “referral” database, and law enforcement
records for all cases: kappa=.60. To resolve the discrepancy between the prosecutor’s
database and the law enforcement records, it was decided to use the information from the
law enforcement records as they were likely more accurate for several reasons. First, the
“referring” database could only be searched using complaint numbers. It may be possible
that data entry errors occurred while complaint numbers were entered into the “referring”
database. Second, the prosecutor’s office does not have a mechanism to track cases that
were not referred to their office. Thus, the “referral” database only contained cases that
were referred to the prosecutor’s office. So the absence of cases in the “referral” database
indicates that they were not referred. The types of disagreements between the “referring”
database and law enforcement records were examined. All but one disagreement
involved cases in which there was no record of referral in the “referring” databases but
the law enforcement record indicated that the case had been referred. Based on the
limitations of the “referring” database, law enforcement records were used to code the
referral variable.

The Macomb County Prosecutor’s database was searched for warranting and final
case outcomes using victims’ names and date of assault. In this database, a record is
created on whether the cases were warranted and its final outcome. In order to assess the
reliability of the information from this database, warranting and final outcome data were
compared against court case files for 30% of the warranted cases. There was 100%
agreement on warranting and case outcomes between the “warranting/ outcome” database

and court records.

42

To assess the reliability of coding for referral, warranting, and final case outcome,
30% of the cases were randomly selected and independently coded from the law
enforcement records and prosecutor’s databases by a second coder. An employee from
the prosecutor’s office was chosen as a second coder for the prosecutor’s database, and a
Turning Point employee was chosen as a second coder for the law enforcement records in
order to limit access of victims’ identities exclusively to the principal investigator. The
investigator and second coders had 100% agreement on referral, warranting, and the final
outcome. That is, the investigator and second coders agreed if a cases was referred or not
(based on law enforcement records), if a case was warranted or not, and on the final
outcome of the case (based on prosecutor database).

Turning Point patient records were searched by victim name and date of assault to
obtain information about victim characteristics, case characteristics, and forensic medical
evidence. To assess the reliability of the coding, the investigator and a research assistant
independently coded victim characteristics, case characteristics, and forensic medical
evidence documented in the patient record for 67% of the total cases. Coding was
consistently monitored throughout the project to maintain kappas >.80. Kappas were
computed to assess the degree of agreement between the two coders on each variable:
trace evidence, kappa=.98; time of the exam, kappa=.92; alcohol/drug use, kappa=.98;
physical bruises, kappa=.98; anogenital abrasions, kappa=.98; physical injuries,
kappa=.96; and other types of physical injury, kappa=.98. All other variables had 100%
agreement. Turning Point patient database was searched by victim name and date of
assault to obtain information for victim and case characteristics (e.g. race, disability, and

victim/offender relationship) that were not systematically documented in the patient

43

records. To assess the reliability of the coding, the investigator and a Turning Point
employee independently coded victim race, disability, and victim/offender relationship
documented in the database for 30% of the total cases. Coding was consistently
monitored throughout the project to maintain kappas >.80. Kappas were computed to
assess the degree of agreement between the two coders on each variable: race, kappa=.98.
Disability, and victim/offender relationship had 100% agreement. There were no
secondary sources against which to compare the accuracy of the forensic medical
evidence, victim characteristics and case characteristics, so no validity assessments were
conducted for these variables.

Measures

The first research question focused on rape case attrition rates, therefore, law
enforcement referral rates, prosecutors’ warranting rates, and case outcomes were coded
from law enforcement records, and the prosecutor’s database (Operational definitions and
coding for each variable are located in Appendix C). In order to determine attrition rates,
law enforcement referral outcomes were coded (was the case referred, yes/no),
warranting outcomes were coded (was the case warranted, yes/no); and final case
outcomes were coded (case charged, but later dropped, plea bargain/trial resulting in
conviction, trial resulting in acquittal) for each case.

The second research question focused on predicting rape case attrition. In order
to determine factors that predict case attrition, information regarding victim
characteristics, case characteristics, and forensic medical evidence were coded for each
case. Specifically, these independent variables included: 1) Victim characteristics: a) age;

b) race/ethnicity (coding: Caucasian/Minority); c) disability status (coding: no disability/

44

disability); (1) median household income; 2) Case characteristics: a) alcohol or drug use
by victim (coding: no/yes); b) type of rape/ penetration (coding: single/multiple/unknown
penetration); c) victim-offender relationship (coding: stranger/nonstranger); d) offender
tactics (coding: unconscious/coercion/force)l; the time between the assault and exam
(coding: actual time in hours); and 3) Forensic medical evidence: a) physical injury
subtypes2 (coding: no/yes) including redness, bruises, abrasions, and other types of
injuries; b) anogenital injury subtypes (coding: no/yes) including redness, tears, bruises,
abrasions, and other types of anogenital injury; trace evidence (coding: no/yes) (see
Appendix B for a complete list of all injuries/evidence coded, and Appendix D for a
complete list of the refined coding used for the analyses). Table 4 presents the
intercorrelations among the predictor variables using all 185 cases in the study (i.e. all
reported cases); and Table 5 presents the intercorrelations among variables in the
subsampled cases that were referred to the prosecutor for a warranting decision (N=90).
The predictors were not highly correlated with each other for either model, with the
exception of the dummy coded variables (type of penetration, and victirrr/offender
relationship). Because these dummy coded variables were entered into the model as a set,

the correlations among the variables within the set should not pose a problem.

 

' Sixteen percent of the data for offender tactics was missing, therefore not included in the multivariate
analyses.

2 Only 3% of the sample had tears (physical injury) and thus, were not included in the multivariate
analyses.

45

 

 

 

Table 4

Correlations of the Predictor Variables and Outcome for the Referring Model

 

VaflaNe 1 2 3 4 5 6 7 8 9 10 11 12 3 14 15 16 17 18 19 70 ”I ’3 7‘
1. Refer 1.00

2 Age .02 100

3. Race ~04 15* 1.00

4.DBaany -rn .22* 15 100

5.1ncome —.01 .05 .16’?‘ 4.02 1.00

6. Drug/alcoh 7.23 —.20* 22* —.12 .03 1.00

7.8inghipen :08 —i() :03 —.12 .13 :03 1.00

8.14unipen .28* .02 :06 .10 :16* :10 :71* 1rr)

9.Unknpen :26 .11 .12 .04 .04 .17* :39* :37* 1.00

10.Stranger -.11 .07 r)2 —r)9 ~ll8 —r)5 —r>7 —r18 .21=r 1.1x)

lLInUnVFani .20* .12 :21* :04 :05 :25* :02 .13 :15’ -29* 1 00

12.Acqumn :09 :16* .13 .10 .09 .26* .09 :05 —r6 :56* :62* 1.00

l3.Deby :10 .02 .10 .08 .21* .07 :01 :09 .13 .13 :10 :03 11m

14.PPlRedness .04 .13 :03 .04 :21* :01 :03 .07 :06 .19* .13 :27*= :02 1.00

15.P1131umes .08 .10 .06 .08 :03 :04 :03 :03 .07 :08 .04 .04 .11 .10 100 \
16.P11Abraaon .06 .13 .01 .07 :03 .n) :08 .04 .05 .03 .17 :12 :00 .34* .37* 11X)

17 Pricnhcr .04 .06 :10 :09 :17* —.n) in ri1 :03 .10 .15* :22* .01 .47* .05 .15* 100

18.A£}Redness .11 i1 .21* .02 :07 .02 :02 .01 .01 .03 .73 -rn .07 .n) .06 .03 .04 100

19.AC}Team .01 .03 :03 .08 :03 :15 :04 .08 :06 .06 .05 :10 :06 .20* .00 .060 .05 .02 100
20.AC}Brumes .10 .10 .05 :08 :06 :09 :08 .05 .04 .19* :05 .11 .08 :01 .03 .140 .07 .14 .07 100

.12 100
21.AriAbragons .18* .01 :08 .03 :07 :07 .04 :02 :03 .03 .10 :12 .08 .04 .11 .14 .08 .12 in

r .16* .22* 1.00
22.ACicnher .03 .14 .09 .00 .17* :06 .03 :05 .02 :01 .14 :12 .01 :06 .07 .08 :09 .12 .22

— -.()2 .08 .13 1.00
23~Traceevidence .06 —.01 .07 —.06 .07 .12 .01 .06 —.08 :06 —.()3 .07 :05 .00 —.04 .07 —.01 .02 .08

 

. - , .. 4 ‘ - ' e enetration 0=n0 multiple penetration); 9.
1. Warrant (0=no). 3. Race (OzMinority); 4. Disability (Ozno disability); 6. Drugs/alcohol (02m) drugs/alcohol used): 7. Single penetration (0=no smglL pmtlrflliamiéj’efifiglfiitirliiate/familizil (Oznot raped by intimate/family);
Penetration unknown to victim (used as reference in dummy code with single & multiple penetration as a set); 10. Raped by stranger (Oznot raped. by :strang::))_sn0 .in_ul:y for all subtypes of injury); 23' Trace evidence (0:no
12- Raped by acquaintance (reference in dummy code with stranger & intimate/familial as a set); 14—17. PH :physical injury; 18—22. AGzanogenital injury — .1 .
trace evidence)

46

 

 

Table 5

Correlations of the Predictor Variables and Outcome for the Warranting Model

 

 

Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 33
l . Warrant 1 .00

2. Age .02 1.00

3. Race :12 .19 1.00

4. Disability —. 10 :01 .16 1.00

5. Income :18 .17 .12 -.11 1.00

6. Drug/alcohol —.17 —.20 28* 7.03 .15 1.00

7. Single pen .18 .05 :06 :04 .19 .03 1.00

8. M11111 pen :05 :07 .05 .04 :26* : 10 :87* 1.00

9. Unknown pen :26* .04 .01 :02 .15 .14 :2 1 :29* 1.00

10. Stranger .15 .05 :01 :06 :06 : 10 :11 .04 15 1.00

11. Intim/Famil .10 .11 :01 :07 : 10 :17 .06 :08 .04 :3 1 1.00

12. Acquaint :20 :14 .01 .11 .13 24* .03 .04 : 15 :46* :70* 1.00

13. ExamDelay :02 :02 :09 :01 25* 25* .05 :13 .16 .09 :06 —.02 1.00

14PHReaness .15 .15 .04 .07 :22* :05 .05 :02 :04 .19 .10 2:3 :19 1.00

15.PHBruises :14 .14 .13 .07 :10 :14 :06 :01 .14 :15 .15 41.3 28* .15 '00

16.PHAbI‘aSion .09 .18 .06 .07 :06 .11 .05 :06 .02 .01 .16 :16 :01 39* 47* 1.00

l7.PHOther .15 .04 :14 :10 :20 :13 .00 .01 :02 .08 .10 :16 .06 25* .15 27* 1.00

18. AG Redness :10 .02 .33 :22* :03 28* .02 :06 .09 .06 :08 .03 :09 .06 :06 .02 .00 1.00

19.AGTears :08 .18 .03 :04 :04 :37* :07 :01 .14 .17 .08 --.21 :19 22* .10 .13 .19 .11 1.00

20AGBruises :04 23* .02 :11 :16 :14 :05 .03 .03 26* :13 :08 .02 .02 .07 .17 .05 .14 .12 1.00
21.AGAbrasion :06 :04 :13 :13 :01 .09 .03 :13 21* .12 .06 :14 .12 :02 .12 .18 .10 <10 08 ~17 1'00
22.AGOther .00 22* .05 :06 .08 :11 .00 :12 23* .17 .12 :24* .01. .02 .10 27* :06 .11 43* 22* .13 1.00
23.Traceevid :15 .08 .12 .05 .06 .10 :07 .12 :11 .08 -.()6 :01 :06 :08 :16 :06 :05 :02 .10 :07 .08 .09 100

   

 

' ' - ‘ ' ‘ 9.

' ' .- , .. . ‘ '8. M in le enetration (O—no multiple penetration), .

1. Warrant (Ozno), 3. Race (OzMinority); 4. Disability (0=no disability); 6. Drugs/alcohol (Ozno drugs/alcohol used); 7. Slngle penetration (Ozno single pezzgatlirlmfiapw lay Fmiriate/familial (0:n0t raped by intimate/famfly);
Penetration unknown to victim (used as reference in dummy code with single & multiple penetration as a set); 10. Raped by stranger (Oznot raped by strang . ~

' ‘ ' ' ' ' O=no
' ' - 1 ‘ ' ' : ' r for all subtypes ofinjury). 23. Trace evtdence(
12' Raped by acquaintance (reference in dummy code with stranger & intimate/familial as a set); 14—17. PH :physrcal injury; 18—22. AGzanogenital 1n‘1ury(0 no 1n]u Y
trace evidence

47

Data Analyses Plan

Raw data were statistically and graphically explored for quality of the data,
potential outliers, and distributional problems prior to conducting data analysis (see
Results section). In addition, potential threats to validity were assessed (see Results
section). For Question #1, descriptive analyses were used to examine the extent that rape
cases move forward in the legal system. For Question #2, given that the dependent
variables were dichotomous (referred/not referred; warranted/not warranted), logistic
regression was used to examine the strength and direction of relationships between
processing decisions, and victim characteristics, case characteristics, and forensic medical
evidence (Hosmer & Lemeshow, 2000). The analysis was approached from a hierarchical
perspective, with predictors organized into conceptually meaningful blocks to be entered
sequentially, with the order planned to facilitate examination of the contribution of
variables in later blocks, controlling for the effects of variables in earlier blocks. This
analysis allowed tests of: 1) the effects of victim and case characteristics on rape case
processing decisions, controlling for the effects of law enforcement agencies; 2) the
influence of forensic evidence on rape case processing decisions, controlling for the
effects of victim (e.g., race) and case characteristics (e.g., victim/offender relationship);
3) the significance and relative unique contribution of variables in each conceptual block
to the prediction of rape case processing decisions; and 4) the overall predictability of
rape case processing decisions. Hierarchical approaches to logistic regression have been
utilized effectively with comparable conditions. Power was estimated, using PASS
software (Heintze, 2001) for the referring and warranting models. For the analysis of

referrals, power to detect a single predictor with an odds ratio of 2.50 was .99 for a

48

continuous predictor, and .70 for a dichotomous predictor with a 50% baserate. These
estimates assume N=14l, a baserate of .49 for referrals, p < .05, and that the analysis
controls for the effects of other independent variables that accounted for an estimated
10% of the variance in the predictor of interest (Hosmer & Lemeshow, 2000). For the
analysis of warranting, power to detect a single predictor with an odds ratio of 2.50 was
.96 for a continuous predictor, and .39 for a dichotomous predictor with a 50% baserate.
These estimates assume N=79, a baserate of .63 for warranting, p < .05, and that the
analysis controls for the effects of other independent variables that accounted for an
estimated 10% of the variance in the predictor of interest Hosmer & Lemeshow, 2000).
The number of predictors included in these logistic regression equations was
evaluated relative to the available sample size. Green (1991) recommends a sample size
equal to or greater than 104 plus the number of predictors for linear regression. To
examine the predictive utility of victim characteristics, case characteristics, and forensic
medical evidence (plus controlling for the effects of law enforcement agencies), 21
predictors were selected for the model. The sample size for the current study (N=185) is
greater than 104 plus 21 predictors. Based on Green’s recommendation (1991), the
sample size, and number of predictors used in the model should be able to detect a

medium to large effect.

49

RESULTS

Univariate and Multivariate Data Screening

Prior to conducting the substantive quantitative data analysis, raw data were
statistically and graphically examined to verify data quality, potential outliers, and
distributional problems that may require transformations or alternative methods. First,
the data were assessed for multivariate outliers. By using Mahalanobis distance with
p<.001, two cases were identified as multivariate outliers. Analyses were conducted for
the referring and warranting models with and without the multivariate outliers included in
the model. Including the multivariate outlier cases did not impact the results; and thus,
they were included in the models. Second, while missing data were not a problem for
most of the variables, there were a few variables that had proportionally more data
missing. The exact time between the assault and exam had missing information for 24
cases. While the date of the assault and exam were available for all cases except one, the
time of the assault was either unknown or missing in the file for 23 cases. An imputation
was created for these cases by using a mid-cutoff point of the days between the assault
and exam. For example, if the victim came in for an exam on the same day as the assault,
the missing cases would be coded as 12 hours. If the victim came in one day after the
assault, the missing cases would be coded as 36 hours. Additionally, income was missing
for seven cases. An imputation was created for the missing data by substituting the seven
cases with the median of household income ($41,859). Offender tactics (e.g. weapon use)
had missing data for 30 cases. Imputations were not possible for this variable, and it was

not included in the model.

50

Distributions for each variable were assessed for any potential problems. First,
lowess graphs were used to explore whether there were curvilinear relationships between
the continuous independent variables (e.g. age, time between exam and assault, and
income) and the dependent variables. The lowess graphs showed that curvilinear
relationships did not exist between the continuous independent and the dependent
variables. Second, histograms and skew statistics were examined, and indicated
distributional problems for two variables: age, and time between assault and exam. The
skew for age was 2.317, and had kurtosis of 7.795. A log transformation was applied to
age, which improved the skew (1.035), and kurtosis (1.001). The skew for time between
assault and exam was 2.082, and had kurtosis of 4.294. A log transformation was applied
to this variable, which improved the skew (.503) and kurtosis (-537). Histograms and
lowess graphs of the transformed variables were examined, and also indicated
improvement in the distributions. After transforming the two variables, the data were
assessed again for multivariate outliers. Using Mahalanobis distance with p<.001, one
case was no longer a multivariate outlier, while one case remained a multivariate outlier.
Again, analyses were conducted for the referring and warranting models with and without
the remaining multivariate outlier case included in the models. Including the multivariate
outlier case did not impact the results, and thus, the case remained in the models.
Research Question #1: Rates of Case Attrition

The first research question in this study was: To what extent do rape cases
processed by a SANE program move forward in the legal system? Specifically, this study
examined the rates of referral, warranting, and case outcomes for rape cases examined in

a SANE program. Descriptive data were collected to document how many cases were 1)

51

referred by the police; 2) warranted by prosecutors; 3) prosecuted; and 4) resulted in
conviction.

Of the 185 reported cases examined in this study, less than half were referred by
the police, and only 25% were prosecuted (see Figure 3 for case attrition at each stage of

the legal system).

52

Figure 3

Case Attrition at Each Stage of the Legal System

 

 

 

 

 

1

185 reported cases

i

 

90 referred by the police to the prosecutor
(49% of total cases)

i

 

57 warranted by the prosecutor

(63% of 90 referred cases)

1
l

 

l

 

(25% of total cases)

47 prosecuted
(83% of 57 warranted cases)

 

l

 

[ (31% oftotal cases)

 

 

 

 

44 resulted in conviction
(24% of total cases)

(94% of 47 prosecuted cases)

 

 

 

 

 

 

l

 

 

 

 

*The average prison sentences ranged from 18 to 34 years.

53

W \ W V F \ K W F N
3 13 2 jail l9 7
deferred probation sentences prison sentences
sentences sentences (under 12 sentences“ unknown
months)

2% of 7% of 1% of 10% of 4% of
total cases total cases total cases total cases total cases
7% of 29.5% of 4.5% of 43% of 16% of
convicted convicted convicted convicted convicted
cases cases cases cases cases
x J K J K J k J \ J

 

 

Research Question #2: Factors Predicting Case Attrition

The second question in this study was: What factors predict case attrition?
Specifically, this study explored the extent to which the victim characteristics, case
characteristics, or the presence and type of forensic evidence collected by the SAN E
nurses predicted case attrition at each stage of the legal system. Furthermore, this phase
of the study examined the predictive value of forensic medical evidence in charging
decisions relative to victim characteristics and case characteristics. Specifically, this
research compared the predictive value of forensic medical evidence relative to victim
characteristics and case characteristics in whether a case moved forward in the legal
system. To answer these questions, a series of logistic regression analyses were
conducted predicting two dependent variables (police referral decisions, and prosecutor
warranting decisions) from three sets of independent variables: victim characteristics;
case characteristics; and forensic medical evidence. Odds ratios were examined to
determine relative importance of each predictor within each model.

Referring Model. Prior to analyzing the police referring model, three potential
threats to validity were evaluated. In other words, are there other factors besides victim,
case, and forensic medical evidence characteristics that influence whether the case is
referred? For example, do cases have the same probability of being referred if a nurse
with little experience collects evidence compared to a nurse with more experience? Do
cases have the same probability of being referred regardless of the law enforcement
agency handling the case? To assess whether maturation of nursing skills was a potential
confounding variable, the nurse assigned to each case was coded by the number of exams

performed during the timeframe of the study. Case experience ranged from one to 129

54

examinations conducted, with a median of 49 exams. A t-test was used to examine the
relationship between referring cases (yes/no), and the number of exams the nurse
conducted during the timeframe of the study. The t-test was non-significant indicating
there was no relationship between police referring and examination experience of the
nurses (t = [183] -.75, p=.46). Thus, maturation of nurses’ skills was not included in the
model. Second, there has been one high profile rape case in Macomb County that
involved a sheriff as the offender and garnered substantial media attention. This may
introduce a historical threat that could influence referral decisions. This possibility was
assessed by examining police referring decisions before and after the high profile case
occurred. A dichotomous variable (before high profile case/after high profile case) was
created. Chi-Square analyses were used to assess whether there was a significant
relationship between case referral and time of the high profile case; this test was non-
significant ()6: [1] .549, p=.46). The third potential threat to validity was examined by
exploring differences between individual law enforcement agencies and case referral.
Nineteen law enforcement agencies made referral decisions for rape cases, ranging from
1 to 39 reported cases per agency. Three characteristics of law enforcement agencies
were examined: reported rape case volume; reported crime volume; and the communities’
median household income of the law enforcement agencies. Agencies with fewer
reported rape cases may have less experience to hone their investigation skills. On the
other hand, agencies with a high volume of rape cases may have more barriers to
providing quality investigations, such as investigator burnout, or lack of resources

(Kerstetter & Van Winkle, 1990). To assess rape case volume, the law enforcement

agencies were dichotomized by the median number of rapes reported to Macomb County

55

law enforcement agencies. This information was obtained from the 2001 Uniform Crime
Report from Michigan State Police. Reported rapes ranged from 0 to 143, with a median
of 36. Law enforcement agencies that had low volume of reported rapes (36 or less) were
coded zero, and those that had high volume of reported rapes (more than 36) were coded
one. Chi-Square analysis was used to examine the relationship between police referring,
and rape case volume. No significant relationships were found among these variables (x =
[1, N=183] .01, p=.92). Similarly, reported crime volume of the law enforcement
agencies was also examined. To assess the volume of reported crimes, the law
enforcement agencies were dichotomized by the median number of crimes reported in
Macomb County law enforcement agencies. This information was obtained from the
2001 Uniform Crime Report from Michigan State Police. Reported crimes ranged from
21 to 5346, with a median of 866. Law enforcement agencies that had low volume of
reported crimes (866 or less) were coded as zero, and those that had high volume of
reported crimes (more than 866) were coded one. Chi-Square analysis was used to
examine the relationship between police referring, and reported crime volume. No
significant relationships were found among these variables (x2 [1, N=183] .08, p=.78).
Case attrition may also be influenced by the available resources within the law
enforcement agency (Frazier & Haney, 1996). Because law enforcement agency budgets
were unavailable, resources for the law enforcement agencies were characterized by the
median household income for their city/ township. The median household income for
each city or township was obtained from the United States 2000 Census. Median
household income ranged from $32,000 to $69,000, with a median of $49,000. Law

enforcement agencies that had communities with low household income ($49,000 or less)

56

were coded as zero, and those that had communities with high household income (more
$49,000) were coded one. Chi-Square analysis was used to examine the relationship
between police referring cases, and the communities’ median household of the law
enforcement agencies. The Chi-Square Statistic was significant, indicating that cases
from communities with lower household incomes were more likely to be referred (12 [1 ,
N=184] = 4.823, p<.05). Thus, law enforcement agencies (as defined by their
community’s median household income) may predict police referrals, and was included
in the referring model.

Table 6 features the results summary for the referring model. The predictor
variables were entered into blocks organized by victim characteristics, case
characteristics, and forensic medical evidence. The first column displays the outcome
variable and the predictor variables. The second column (with the heading B) features the
logistic coefficient, which represents the change in the logit (natural log of the odds) of
the outcome variable associated with a one-unit difference in the predictor variable
(Fields, 2000; Tabachnick & Fidell, 2001). The third column represents the Wald
statistic, which is a z-test on the regression weight(B), and has a Chi-Square distribution.
The fourth column displays the odds ratio, which can be interpreted as the multiplicative
change in the odds of being in one outcome category verses the other when the value of
the predictor increases by one unit (Tabachnick & Fidel], 2001). The fifth column shows
the 95% confidence interval of the odds ratio. The last line of each block features the
Chi-Square Difference test, which evaluates if significant improvement in equation fit to

the data occurred in the present block compared to the last block.

57

Predictors were organized into conceptually meaningful blocks including victim
characteristics, case characteristics, and forensic medical evidence. However, univariate
analyses showed a potential threat to validity existed between individual law enforcement
agencies and case referral. Thus, the first block only included law enforcement agencies
(from low—income communities/high—income communities) in order to control for the
effect of this variable in future blocks. The Wald Statistic indicated that law enforcement
agencies did not significantly predict police referring cases. The Model-Chi-Square was
not significant in this equation, meaning that overall the model was not predicting police
referral significantly better than it was with only the constant included.

The second block includes all victim characteristics expected to predict police
referring cases. Controlling for the effects of law enforcement agencies, the Wald
Statistic indicates that victims’ age, race, disability status, and median household income
do not significantly predict police referring cases. The Model-Chi-Square was not
significant in this equation meaning this set of predictors did not significantly predict
police referral.

The third block includes all case characteristics expected to predict police
referring cases. Controlling for the effects of law enforcement agencies and victim
characteristics, the Wald Statistic indicates that some case characteristic significantly
predicted police referring cases including alcohol/drug use by victims, and the type of
penetration. The results of the odds ratio indicated that cases in which victims were
consuming alcohol or drugs before or during the assault decreased the odds of the case
being referred by 34%. Cases in which the offender penetrated more than one orifice of

the victim were 5.56 times more likely to be referred than cases in which the victim did

58

not know what occurred during the rape (e. g. victim was unconscious). In addition, cases
in which the offender was an intimate partner or family member were somewhat more
likely to be referred than when the offender was an acquaintance (2.41 times as likely,
p=.08). The Model-Chi-Square was significant (Model x2=35.977, p<.01) in this block,
meaning that this set of predictors was able to significantly predict police referral. In
addition, this group of predictors was overall able to correctly classify police referral for
65% of the cases.

The fourth block includes all forensic medical evidence expected to predict police
referring cases. Controlling for the effects of law enforcement agencies, and victim and
case characteristics, the Wald Statistic indicates two types of forensic medical evidence
significantly predicted police referring cases: anogenital bruises and abrasions. The
results of the odds ratio indicated that cases with documented anogenital bruises were
6.27 times more likely to be referred. Cases with documented anogenital abrasions were
5.11 times more likely to be referred. The Model-Chi-Square was significant (Model
x2=60.484, p<.01) in this block meaning that this set of predictors was able to
significantly predict police referral. With the addition of the forensic medical evidence
predictors, the model was overall able to classify correctly police referral for 74% of the

C3865.

59

 

Table 6

Logistic Regression Analyses Predicting Police Referring Cases to Prosecutor. N:|41

 

 

 

 

Block One Block Two Block Three Block Four
Predictor ()dds Odds Odds Odds
B Wald Ratio 95% Cl [3 Wald Ratio 9512 C1 B Wald Ratio 05% (‘1 [3 Wald Ratio 95% C1
Law Enforcement Agency 0.52 2.43 0.59 0.31— 40.57 2.66 0.56 0.28-1.12 —().83 4.34:“ 0.44 0.20-0.95 0.77 2.84“? .46 .19—1.14
1 . 14
Victim Characteristics
A80 ().2| .15 1.23 0.43—3.58 —0. 16 0.06 085 0.25—2.95 «.48 .44 .62 .15-2.54
Race - -0.42 .86 0.66 0.27—1.60 0.24 0.21 1.28 0.45—3.61 .14 .05 1.15 .35—3.76
Median household income _0.40 .81 0.67 0.28—1.60 —().69 1.83 0.50 0.19—1.36 :75 1.70 .47 .15—1.46
Disability status 0.00 .12 1.00 098.102 0.01 1.22 1.01 0.99—1.04 .03 3.7513 1.03 1.00—1.06
Case characteristics
Victim used drugs/alcohol —1.08 0.34 0.15—0.77 - 1.40 8.55:“: .25 .10—.63
Single 13011011311011 0.60 607.33 1.82 0.54-61 l .38 .30 1.46 .38-563
Multiple penetration 1.72 5.56 1.64— 1.89 6.83M 6.62 1.60—
7.623“? 18.81 27.33
RaPEd by stranger —0. 14 0.08 0.87 0.34—2.23 -. 15 .06 .86 .28—2.71
Raped by intimate/familial 0.88 3.091’ 2.41 0.90—6.42 1.00 3.28% 2.73 928.09
Delay in exam -0.22 1.07 0.80 0.52—1.22 —.61 553* .54 .33—.90
Physical injury
Redness .04 .01 1.04 .30—3.65
Bruises .41 .71 1.50 .58—3.89
Abrasions .30 .36 1.35 .50—366
Other .03 .00 1.03 .32-3.31
Anogenital injury
Redness .80 2.60 2.23 .84—5.89
Tears :94 1.98 .39 .11—1.45
. * -
Bru1ses 1.84 5.11 6.27 3102:1
- *4: _
Abrasrons 1.63 8.19 5.1 1 1156g0
Other -1.07 1.83 .34 .07—1.62
Trace evidence 1.25 2.35 3.51 .71—17.43 ‘x.
)6 Difference 2.45 1.95 31 .58M 24.507M

 

 

*p<.05, **p<.01. 1‘p<. 10

60

 

 

Chi-Square differences tests were used to evaluate the predictive value of forensic
medical evidence in charging decisions relative to victim characteristics and case
characteristics. Do police place more emphasis on forensic medical evidence in their
decisions to refer cases than victim and case characteristics? Table 7 displays the value of
-2 log likelihood the differences in the Chi-Square among each set of predictors entered
into logistic regression (Fields, 2000). Lower values of -2 log-likelihood indicate that the
model is predicting the outcome variable more accurately. The Chi-Square difference
statistic indicates the improvement in the predictive power of the model since the last
stage. As shown in Table 7, the value of the -2 log-likelihood for the constant model was
only 206.55, which was reduced to 204.102 when law enforcement agencies was added,
then reduced to 202.149 when victim characteristics were introduced, then reduced to
170.574 when case characteristics were added, and finally reduced to 146.068 with the
addition of forensic medical evidence in the model. Based on the -2 log-likelihood test,
the model is better at predicting police referral with the addition of forensic medical
evidence predictors. In evaluating the Chi-Square difference, the addition of law
enforcement agencies, and victim characteristics did not significantly improve the
predictive power of the model. The addition of case characteristics significantly improved
the predictive power of the model ()6 difference = 31.575, p<.01). Furthermore, the
addition of forensic medical evidence also significantly improved the predictive power of
the model (3;2 difference = 24.507, p<.01). Thus, forensic medical evidence provides more

predictive value in referring cases than victim and case characteristics.

61

Table 7

Chi-Square Difference Test for Types of Predictors for Police Decisions
Predictors -2 Log likelihood )6 Difference

 

Law Enforcement Agencies 204.102 2.449

Victims Characteristics 202.149 1.953

Age

Race

Disability status

Median household income

Case Characteristics 170.574 3 1 .575 * *

Alcohol/drugs used before/after the
rape

Type of penetration
Victim/offender relationship

Delay in exam

Forensic Medical Evidence 146.068 24.507"

Physical redness
Physical bruises
Physical abrasions
Physical other
Anogenital redness
Anogenital tears
Anogenital bruises
Anogenital abrasions
Anogenital other
Trace evidence

 

* * p<.01

In hierarchical regression, the significance of each predictor is typically evaluated
in the block in which it was entered into the equation. However, when new predictors are
added in subsequent blocks, the significance of previous predictors may change (i.e.
some variables that are non—significant become significant and vice versa). Therefore, it

is necessary to examine the previous predictors in greater detail as new predictors are

62

 

added to the model. For example, when case characteristics were added to the model, the
significance of law enforcement agencies changed. While law enforcement agencies were
not significant in earlier blocks, adding case characteristics changed law enforcement
agencies to become significant. Specifically, higher income law enforcement agencies
had 44% lower odds of being referred relative to cases from lower income agencies. To
examine why the addition of case characteristics would change the significance of the
law enforcement agencies, univariate analyses were conducted to examine the
relationship between law enforcement agencies and case characteristics. These analyses
did not find any significant relationships between these predictors (see Table 8). Yet,
when forensic medical evidence was added to the model, the significance of law
enforcement agencies changed to a trend (p=.09). The addition of forensic medical
evidence also changed the significance of two other predictors. First, the addition of
forensic medical evidence influenced the median household of victims to become a trend,
which was originally non-significant. However, the results of the odds ratio indicated that
there was virtually no effect (odds ratio = 1.027). Second, the addition of forensic
medical evidence influenced time between the assault and exam to become significant.
Specifically, for every unit (transformed hour) between the assault and exam, the odds of
the case being referred decreased by 54%. To examine why the addition of forensic
medical evidence would change the significance of time between the assault and exam, t-
tests were conducted to examine the relationship between injuries and time between the
assault and the exam. These analyses did not find any significant relationships between

these predictors.

63

Table 8

Relationship between Law Enforcement Agencies and Case Characteristics of Referred

 

 

Cases
All Referred Referred Test of Significance
Referred Cases in Cases in “Low Income LE
Cases Low High Agency” vs. “High
Income Income Income LE Agency”
LE LE for Referred Cases
Agencies Agencies
n = 90 n = 49 n = 41
Victim used drugs/alcohol 42% 45% 39% NS
Single Penetration 39% 39% 39% NS
Multiple Penetration 54% 53% 56% NS
Unknown Penetration 7% 8% 5% NS
Raped by stranger 17% 21% 12% NS
Raped by intimate/familial 32% 28% 37% NS
Raped by acquaintance 51% 51% 51% NS
Time between assault and 2.38 2.33 2.42 NS
exam

64

Further analyses were conducted to examine if a combination of case
characteristic predictors changed the significance of law enforcement agencies. The
logistic regression equation was reanalyzed by adding different combinations of case
characteristic into the block to determine which combination of predictors changed the
significance level of law enforcement agencies. The effect of law enforcement agencies
increased and became significant (p=.02) when type of penetration and victim/offender
relationship were entered into the case characteristic block. Analyses were also conducted
to examine if a combination of forensic medical evidence predictors changed the
significance of law enforcement agencies. Again, several different combinations of
forensic medical evidence predictors were entered into the forensic medical evidence
block to assess which combination of predictors changed the significance level of law
enforcement agencies. The analyses showed that the effect of law enforcement agencies
became significant (p=.04) when only physical redness, other types of physical injury,
anogenital tears, anogenital bruises, other types of anogenital injury, and trace evidence
were added to the forensic medical evidence block. To examine why the addition of
forensic medical evidence would change the significance of time between the assault and
exam, several combinations of forensic medical evidence predictors were entered into the
forensic medical evidence block. The effect of time between assault and exam became
significant (p=.04) when only physical abrasion, physical redness, anogenital redness,
and anogenital abrasion were entered into the forensic medical evidence block.

Warranting model. Another series of logistic regression equations were performed
to explore whether victim characteristics, case characteristics, or the presence and type of

forensic medical evidence collected by the SANE nurses predicts prosecutors warranting

65

cases. Only 90 cases were referred to the prosecutors, which drastically reduced the
sample size of the warranting model. The sample size of the warranting model exceeds
the rules for events (i.e. predictors) to subjects ratio. With a sample size of 90, the
number of predictors needed to be decreased to 8 to achieve stability for detecting a large
effect (Green, 1991). Hosmer and Lemshow (2000) suggest dropping variables that have
no predictive value. Therefore, variables that did not show a significant relationship with
police referral were dropped (age, race, disability status, median household income,
physical redness, other types of physical injury, anogenital redness, other types of
anogenital injury, and trace evidence found on victims’ body) with one exception.
Discussions with key informants reported that prosecutors were more likely to warrant
cases in which the victim had documented physical injury. Although none of the physical
injury subtypes were significant in the police referring model, two of them were included
in the warranting model. Because the key informants suggested that physical bruises and
abrasions were important in prosecution, these two physical injury subtypes were
included in the warranting model. One final predictor was dropped from the warranting
model. Dummy coding was used to examine type of penetration: single penetration,
multiple penetration, and penetration unknown by the victim due to being unconscious
(served as reference). In the referring model, 17% of cases involved penetration unknown
by the victim. However, the warranting model only contained 7% of cases involving
unknown penetration. A preliminary analysis showed that type of penetration had large
unstable parameters when added to the warranting model due to the low number of cases
involving unknown penetration, and therefore this variable was dropped from the model.

The final predictors included in the warranting model included: alcohol/drugs used before

66

or during the assault, victim offender relationship, time between assault and exam,
physical bruises and abrasions, and anogenital bruises and abrasions.

Prior to analyses, three potential threats to validity were evaluated for the
warranting model. The purpose of this exploration was to determine if there are other
factors besides victim characteristics, case characteristics, and forensic medical evidence
that influenced warranting decisions. For example, do cases have the same probability of
being warranted regardless of the prosecutor handling the case? First, to assess validity of
maturation of nursing skills as a confounding variable, the nurse assigned to each case
was coded by the number of exams performed during the timeframe of the study. A t-test
was used to examine the relationship between warranting cases, and the number of exams
conducted by the nurses. The t-test did not show a significant relationship between
warranting, and examination experience of the nurses (t= [88] -.98, p=.33). Thus,
maturation of nursing skills was not included in the model. Second, to examine whether
the high profile rape case in Macomb County introduced a historical threat, warranting
decisions before and after the high profile case were compared. A dichotomous variable
(before high profile case/after high profile case) was created. Chi-Square analysis was
used to assess this threat, and did not find a significant relationship between the high
profile case with warranting cases ()8 [1, N=90] = 1.188, p=.28).

The third potential threat to validity existed between individual prosecutors and
warranting decisions. Three prosecutors made warranting decisions for all referred rape
cases in the county during the timeframe of the study. The county is divided into three
sections, and each prosecutor makes warranting decisions for rapes occurring in their

assigned section. Thus, investigators within the same law enforcement agency refer cases

67

to the same prosecutor with one exception. One large law enforcement agency is assigned
to two prosecutors due to a high volume of cases. Unfortunately, it was not possible to
determine the assigned prosecutor for this law enforcement agency. Thus, 24% of the
cases for this variable were missing. A code with three categories was created to
distinguish cases processed in these three sections (e.g. area one, area two, area three).
Cramér’s V was used to examine the relationship between the three prosecutors, and
warranting cases. Cramér’s V is useful in exploring the strength of association between
two variables when one of the variables has more than two categorical levels (Pett, 1997).
Cramér’s V did not show a significant relationship between assigned prosecutors and
warranting cases (x,2 = 2.366, p=.19, Crame'r’s V =.19).

An additional analysis was conducted examining the relationship between
prosecutors and warranting cases. Key informants reported that they expected one
particular prosecutor to warrant more cases, and to make warranting decisions based on
different criteria than the other two prosecutors. In particular, they noted that this
prosecutor was more likely to warrant cases even if there was no documented injury. Key
informants believed that this prosecutor made decisions unique to the other prosecutors
because of having more experience prosecuting rape cases, receiving more specialized
training, and engaging in collaborations with the local rape crisis center as well as the
local sexual assault nurse examiner program. Therefore, a dichotomous variable
(experienced prosecutor/less experienced prosecutors) was created to examine the
relationship between the experienced prosecutor and warranting cases. Fortunately, the
section assigned to the experienced prosecutor does not involve the large law

enforcement agency split among two prosecutors. Thus, this new variable did not have

68

missing data. Using Chi-Square analysis, a significant relationship between the
experience of the prosecutors and warranting was not found. Based on these univariate
analyses, assigned prosecutors and the prosecutor experience variable were not included
in the model (712 [1, N=90] = .627, p=.43).

Predictors were organized into conceptually meaningful blocks including case
characteristics, and forensic medical evidence. The first block included case
characteristics expected to predict prosecutors warranting cases (see Table 9). The Wald
Statistic indicates that none of the case characteristic significantly predicted prosecutors
warranting cases. The Model-Chi-Square was not significant in this equation meaning

that this set of predictors was not able to significantly predict warranting.

69

cfiVnfi. JoVPI. .mo.VQ*

 

 

 

.53: end oosocobmu «x
modémd owd Ed mmd- 2285.6:
cc. 736 $3 2: oz. mommam
SE 33822
3.5 ~ -o ~ ._ m: ... I; 21 338.252
No.o-mo.o S6 434.0 cod- mommam
has samba
3.23 of we; :3 me. :2 86 8o :3. use 5 6:5
8.6-23 8; S... Rd 8.33 of 23 3o Ezsateséea 3 Beam
8.8-98 8a :6 $6 8.2-63 42 on; 3o 85:... 3 685
68.986 moo .88 9.. _- em. :3 :8 e3 26. 65838.5 ea: 83>
mowmtoaowbfio ommo
03mm 2me
5 $3 $60 263 m 5 $3 £60 255 m
88%on
93 82m 25 82m

 

om“ Z £25885 c8 mommu meta—«53 cousoomoiwcwfiwocm mowing commmocwom otmmwoq

o 2an

7O

The second block includes forensic medical evidence expected to predict
prosecutors warranting cases. The Wald Statistic indicates two types of forensic medical
evidence significantly predicting prosecutors warranting cases: physical bruises and
abrasions. The results of the odds ratio indicated that documented physical bruises
decreased the odds of the case being warranted by 12%. Cases with documented physical
abrasions were 4.45 times more likely to be warranted. The Model-Chi-Square was
marginally significant (Model x =15.357, p=.05) in this block meaning that this set of
predictors was somewhat able to predict prosecutors warranting cases. Furthermore, this
group of predictors was overall able to correctly classify prosecutors warranting for 75%
of the cases.

Chi-Square differences among each set of predictors was used to evaluate the
predictive value of forensic medical evidence in warranting decisions relative to case .
characteristics. Do prosecutors place more emphasis on forensic medical evidence in their
decisions to warrant cases than case characteristics? Table 10 displays the value of -2 log
likelihood and the differences in the Chi-Square among each set of predictors entered into
logistic regression. As shown in Table 10, the value of the -2 log-likelihood for the
constant was 100.101, which was reduced to 94.742 when case characteristics were
added, and finally reduced to 84.744 with the addition of forensic medical evidence in the
model. Based on the -2 log-likelihood test, the model is better at predicting prosecutors
warranting with the addition of forensic medical evidence predictors. In evaluating the
Chi-Square difference, the addition of case characteristics did not significantly improve
the predictive power of the model. The addition of forensic medical evidence

significantly improved the predictive power of the model (x2 difference = 10.00, p<.05).

71

Thus, forensic medical evidence provided more predictive value for warranting than did
case characteristics.
Table 10

Chi-Square Difference Test for Types of Predictors for Prosecutor Decisions

 

 

Predictors -2 Log likelihood )6 Difference
Case Characteristics 94.742 5.359
Alcohol/drugs used before/after the
rape

Victim/offender relationship
Delay in exam

Forensic Medical Evidence 84.744 9998*

Physical bruises
Physical abrasions
Anogenital bruises
Anogenital abrasions

 

* p<.05

As previously mentioned, the significance of predictors is typically evaluated in
the block in which they were entered into the equation. However, when new predictors
are added in subsequent blocks, the significance of previous predictors may change.
Therefore, previous predictors were examined in greater detail as new predictors were
added to the model. The addition of forensic medical evidence to the model influenced
one case characteristic to become significant. While alcohol/drugs used by the victim was
not a significant predictor in the first block (case characteristics), it became significant
when forensic medical evidence was added to the model. The results of the odds ratio
indicated that victim’s consumption of alcohol/drugs before or during the assault
decreased the odds of the case being warranted by 23%. Univariate analyses were

conducted to examine the relationship between alcohol and forensic medical evidence

72

(e. g. physical and anogenital injuries) to help explain this finding. Chi-Square Statistic
found a significant relationship between physical bruises, and alcohol/drug use. Cases in
which the victim was using alcohol/drugs had significantly less injury than cases in
which the victim was not using alcohol/drugs (x2 [1, N=53] = 4.44, p<.05). This
relationship may have influenced alcohol/drug use to become significant when forensic
medical evidence was added to the model. Further analyses were conducted to examine if
a combination of forensic medical evidence predictors changed the significance of
alcohol/drug use. The logistic regression equation was reanalyzed by adding various
combinations of forensic medical evidence into the block to determine which
combination of predictors changed the significance level of alcohol and drug use by the
victim. Alcohol/drug use became significant (p=.02) when type of physical bruise,
physical abrasion, anogenital bruise, and anogenital abrasion were entered into the
forensic medical evidence block.

One unexpected finding also required further analyses. Previous research
indicates that cases are more likely to be warranted if the victim sustained injury
(Chandler & Tomey, 1981; Spears et al., 2001). However, cases with documented
physical bruises were significantly less likely to be warranted in the model. Interviews
with key informants reported that one particular prosecutor was more likely to warrant
cases even when there was no documented injury. Therefore, univariate analyses were
conducted to examine the relationship between physical bruises, and experience of
prosecutors. Did the experienced prosecutor warrant cases with significantly more/less
physical bruises? Chi-Square Statistic did not find any significant relationships or trends

between these variables (x’= [1, N=90] 1.688, p=. l 9).

73

Further analyses were conducted to explore the relationship between all three
prosecutors, and physical bruises. First, univariate analyses were conducted to examine
the relationships between the three prosecutors and warranted cases with physical bruises.
Did any of the prosecutors warrant cases with significantly more/less physical bruises?
Cramér’s V was chosen for this analysis because of its useful in exploring the strength of
association between two variables when one of the variables has more than two
categorical levels (Pett, 1997). A trend (p=.11) was found between the three prosecutors
and physical bruises. In particular, 80% of the cases warranted by one prosecutor had

physical bruises but only 41% and 44% of cases warranted by the other prosecutors had
physical bruises (x2=4.442, p=.11, Cramér’s V=.33). A second univariate analysis was

conducted to examine the relationships between prosecutors and physical bruises for a_ll
m cases received by the prosecutors. The purpose of this analysis was to evaluate
significant difference in the number of documented physical bruises among the gag
meg by the three prosecutors. Did any of the individual prosecutors receive
significantly more cases with physical bruises than their colleagues? The analysis showed

no significant difference in the number of cases with physical bruises received by the
three prosecutors (x2=3.946, p=.14, Cramér’s V=. 16). This suggests that no significant

differences existed in the number of received cases with physical bruises among the three
prosecutors. However, there was a slight difference in the number of cases warranted
with physical bruises among the three prosecutors. Overall, one prosecutor tended to
warrant a smaller proportion of assigned cases than the other prosecutors, and the

majority of the warranted cases by this prosecutor had documented bruises. Thus, it may

74

be possible that this combination of the lower warrant rate and greater reliance on bruises

may have produced the decreased odds in cases with bruises being warranted.

75

DISCUSSION

The first goal of this research was to examine the rates of referral, warranting, and
case outcomes for rape cases examined in a SANE program. Most prior research on rape
case attrition was conducted in the 19805, and has primarily examined later stages (e. g.
warranting) of case processing. This study offered a more refined analysis by focusing on
earlier stages of processing when cases are likely to be filtered out of the legal system.
Furthermore, this research expanded the literature by focusing on cases processed
through a SAN E program, which in other studies have been found to create significant
changes in the way communities respond to rape.

The second goal of this research was to examine whether the presence and type of
forensic medical evidence collected by the SANE nurses predicts case attrition. Previous
studies on the factors that predict case attrition have not examined the nuances of forensic
evidence (e. g. type of injury). This study expanded the literature by examining how the
presence of specific types of physical and anogenital injury predicted whether cases move
forward in the legal system. Previous literature has also found that victim and case
characteristics predict whether rape cases move forward in each step of the legal system.
This study expanded the literature by comparing the predictive value of forensic medical
evidence in cases being referred or warranted relative to victim characteristics and case
characteristics. This is important because it is unlikely that cases are referred or
warranted based on only one particular evidentiary, victim, or case characteristic.
Summary of Major Findings

Case attrition. The first focus of the study explored the extent to which rape cases

processed by a SANE program moved forward in the legal system. The results showed

76

that 49% of reported cases processed by the SAN E program were referred by the police.
This finding is somewhat higher than previous studies examining rape case attrition in
cases not processed by SAN E programs (see Appendix E for comparison between
published attrition rates and attrition rates of current study). The literature suggests that
only 41% (Chandler & Tomey, 1981) to 44% (Galvin & Polk, 1982) of rape cases were
referred in the 19803, while more recent studies found that only 18% (Bouffard, 2000) to
22% (Frazier & Haney, 1996) of rape cases were referred. In this study of the reported
cases processed by the SAN E program, 25% were prosecuted. This result is also slightly
higher than past studies examining rape case attrition of non-SAN E cases. For example,
previous literature found that only 16% (Frazier & Haney, 1996) to 18% (Chandler &
Tomey, 1981) of all reported cases were prosecuted. In addition, the findings of this
study indicated that 24% of the reported cases resulted in conviction (i.e. plea-bargain or
guilty verdict). This rate is higher than past studies that found that only 7% (Galvin &
Polk, 1982) to 17% (Chandler & Tomey, 1981) of reported (non-SANE) cases resulted in
conviction. The current study also showed that only 11% of reported cases ended with
prison sentences. This sentencing rate is slightly greater than previous studies, which
found that only 6% (Galvin & Polk, 1982) to 9% (LaFree, 1980) of the reported cases
resulted in prison sentences. The results of this study suggest that referral, prosecution,
conviction, and prison sentencing rates are somewhat higher in the focal SAN E program
compared to published rates in non-SANE cases.

There have been few studies that have examined rape case attrition rates in cases
processed by SAN E programs. Ledray (1992) documented attrition rates for SAN E cases

processed in Minneapolis, and found attrition rates similar to the current study. Of all

77

reported cases processed in the Minneapolis SAN E program, 46% were referred by
police, and 32% were prosecuted. The referral rate in the current study was also similar to
the referral rate found in the only empirical study to date that has examined attrition rates
before and after the implementation of a SANE program. Crandall and Helitzer (2003)
found that only 38% of the rapes cases were referred prior to implementing the SANE
program, while 50% of the rape cases were referred for warranting after the
implementation of the SANE program. However, the prosecution rates in Crandall and
Helitzer’s (2003) study were higher than the current study. They found that 25% of the
reported cases were prosecuted before the implementation of the SAN E program, while
34% of the reported SAN E cases were prosecuted afier the implementation of the SANE
program. The referral rates for the current study (49%) are similar to the published rates
for referral of cases processed by SAN E programs (42% & 50%). However, the
prosecution rates of the current study (25%) are less than the published rates for
prosecution of SANE cases (32% & 34%).

Police referring. The second focus of the study examined the factors that predict
police referring rape cases to prosecutors, and the relative predictive value of forensic
medical evidence, victim characteristics, and case characteristics in police referral
decisions. With respect to victim characteristics, this study found that age, race, and
household income did not predict police referral. This is consistent with previous studies
that did not find a relationship between race of victims and cases being referred (Frazier
& Haney, 1996; Bouffard, 2000), but is inconsistent with previous findings regarding
victim age and income. Previous studies suggested that police were less likely to refer

cases with younger women because they view them as more suspicious and believed that

78

younger victims were likely to fabricate or instigate the rape (Rose & Randall, 1982).
Furthermore, prior literature indicated that police display class prejudice for victims who
hold lower class status as well as victims who hold upper class status and hold
stereotypes towards those who live in particular economically homogenous areas (Rose
& Randall, 1982). Prior research to date has never examined the role of the victim’s
disability on legal outcomes. The current study showed that victims with disabilities were
no more, or less, likely to have their cases referred by the police.

The second set of predictors examined was case characteristics. The length of
time between the assault and exam did not predict police referral. However, some case
characteristics significantly predicted police referral. First, drug or alcohol use by the
victim before or during the assault significantly decreased police referral by 34%. This
finding was consistent with the majority of studies examining the relationship between
alcohol and drug use and police referral (Kerstetter, 1990; Campbell, 1998). Second,
cases in which offenders penetrated more than one orifice of the victim were six times
more likely to be referred than cases in which victims were unaware of what occurred
due to being unconscious or blacking out. While previous literature has focused on the
presence or absence of penetration, this study is the first to date that explored how the
type of penetration may affect case attrition. Third, cases in which the offender was an
intimate partner or family member were somewhat more likely to be referred than when
the offender was an acquaintance. To date, prior studies have not examined referral rates
between intimate/familial relationships and acquaintances. Past research has focused
mainly on comparing rapes committed by strangers to rapes committed by all offenders

known to victims (i.e. combining intimate, familial, and acquaintances into one group).

79

These studies had mixed results with some studies indicating that stranger cases were
more likely to be referred (Kerstetter, 1990; Frazier & Haney, 1996), while other studies
indicated that cases in which the offenders knew the victims were more likely to be
referred (Bouffard, 2000). The current study did not find stranger cases referred
significantly more or less often than acquaintance cases.

The third set of predictors examined was forensic medical evidence. The findings
suggest that physical injuries do not predict police referral. However, two types of
anogenital injuries did predict police referring cases. First, cases with documented
anogenital bruises were six times more likely to be referred. Second, cases with
documented anogenital abrasions were five times more likely to be referred. While other
research has examined the relationship between the presence of injury and referral, those
studies did not examine the nature of the injuries (e.g. physical, anogenital). Prior
research has suggested that police refer cases more often if victims were injured (Galton,
1976; Rose & Randall, 1982); but only one study has explored anogenital injury and
referring. Kerstetter (1990) interviewed twenty detectives and found that they were more
likely to refer cases if the victim endured injuries to her sex organs. The current study
also supported the finding that anogenital injury was an important factor in police
offrcers’ decision-making on referring cases. Furthermore, the current study expanded
this finding by showing that the subtypes of anogenital injury (e. g. abrasion) also were
important factors in referring cases. The current study also explored the predictive value
of forensic medical evidence in police referral decisions relative to victim and case
characteristics. The findings show that forensic medical evidence provided more

predictive value in referring cases than victim and case characteristics. This finding is

80

similar to that of a prior study that indicated that cases with physical evidence had a
higher probability of referral even when the victim’s credibility was questioned (Rose &
Randall, 1982).

Warranting. The third focus of the study examined the factors that predict
whether cases are warranted for prosecution. The first set of predictors examined was
case characteristics. The findings suggest alcohol or drug use by the victim, the
relationship between the victim and offender, and the length of time between the assault
and exam did not predict prosecutors warranting rape cases. These findings were
inconsistent with prior studies that showed that the role of victim alcohol and drug use
before or during the rape decreases the probability that cases are warranted (Chandler &
Tomey, 1981; Spears & Spohn, 1996; Frohmann, 1997). Prior studies have yielded
mixed findings on warranting cases and the relationship between the victim and offender.
This study is consistent with the majority of studies that did not find a significant
relationship between warranting and the victim offender relationship (Bachman, 1998;
Spohn & Homey, 1993; Spears & Spohn, 1996, 1997). However, other studies have
indicated that prosecutors were less likely to warrant cases when the victim was a relative
of the offender (Bradmiller & Walters, 1985), and more likely to warrant cases when the
victim and offender were acquaintances (Spears, Beichner, and Davis-Frenzel, 2001).

The final set of predictors examined was forensic medical evidence. The findings
suggest that anogenital injuries do not predict warranting. However, two types of physical
injuries did predict warranting. First, cases with documented physical abrasions were four
times more likely to be warranted. This finding is consistent with prior research that

suggest that documented injuries is predictive of cases moving firrther through the

81

criminal justice system (Chandler & Tomey, 1981; Feldman-Summers & Palmers, 1980;
Frazier & Haney, 1996; Martin & Powell, 1995; Spohn & Spears, 1997; Spohn, Beichner,
& Davis-Frenzal, 2001). However, the current study found that documented physical
bruises decreased the odds of the case being warranted by 12%, which is not consistent
with prior research. Further univariate analyses were conducted to examine the
relationship between physical bruises, and the three prosecutors assigned to the cases.
Findings suggested that while one prosecutor tended to warrant mostly cases with
documented physical bruises, the other two prosecutors warranted cases both with and
without physical bruises. This finding suggests that warranting may be influenced by
characteristics of the prosecutors making the decisions in addition to evidentiary
characteristics. This study also explored the predictive value of forensic medical evidence
in warranting decisions relative to case characteristics. The findings show that forensic
medical evidence significantly provides more value in predicting warranting than case
characteristics. That is, prosecutors base their warranting decisions more on forensic
medical evidence than the characteristics of the case.
Implication of Findings

Results from this study indicate that rates of referral, prosecution, conviction, and
prison sentences were somewhat higher in the focal SAN E program compared to
published rates in non-SANE cases. These findings imply that SANE programs may
reduce case attrition. There are several components of SAN E programs that may
contribute to cases moving further in the system. First, sexual assault nurse examiners
receive specialized training in assessment of injuries, identifying patterned injury,

documentation of injuries, and using specialized equipment to assess injuries (Cohen et

82

 

 

col.

al., 1996; Ledray, 1998). This training may increase the probability of injuries being
detected and properly documented, and thus, increases the probability of cases moving
forward in the legal system. In addition, the quality of the medical report may affect case
attrition. Crandall and Helitzer (2003) interviewed law enforcement and prosecutors and
found that the SANE forensic medical reports were more reliable, easier to read, more
efficiently transferred to their office, and more effective in court than the forensic
medical records prior to the implementation of SANE. Thus, increased accessibility of
the forensic medical report and trust of the documented evidentiary findings may have
reduced case attrition.

Second, nurse examiners are trained to provide expert witness testimony to
explain the documented injuries or normalize the absence of injury during the trial. It
may be possible that police and prosecutors are more willing to forward cases when
expert witness testimony is available to explain the medical findings to jurors and judges.
This may be particularly true in cases in which it is anticipated that jurors may not deem
the victim to be a credible witness.

Third, previous research has suggested that SANE programs improve
collaboration between the legal and medical system (Crandall & Helitzer, 2003).
Specifically, nurse examiners provide training and case consultation for police and
prosecutors to increase their understanding of medical evidence findings (Littel, 2001).
Furthermore, training and consultation is also an opportunity to normalize the absence of
injuries. That is, the absence of injuries does not indicate that the rape was fabricated.

Fourth, prior research suggests that the time police spend waiting for the evidence

collection to be completed was greatly reduced because SAN E programs only provide

83

services to rape victims (Crandall & Helitzer, 2003). Reduced waiting time provides
officers with more time to investigate cases. This is important because previous studies
have found that time constraints may influence police to refer only cases most likely to
result in conviction (Frazier & Haney, 1996; Kerstetter & Van Winkle 1990; Madigan &
Gamble, 1991; Martin & Powell, 1995). For example, one study found that officers
sympathized with victims but their obligation to build a case, establish probable cause,
prepare a victim for court along with their time and energy constraints discouraged them
from referring the case (Martin & Powell, 1995). Thus, this extra time to investigate
cases may have increased the number of cases referred.

Fifih, the goal of SANE programs is to provide supportive care to victims (Amdt,
1988). Anecdotal evidence suggests that SAN Es provide more compassionate treatment
of victims than traditional emergency room personnel. Providing compassionate care to
victims may increase the number of victims continuing to participate in the prosecution
of their case. However, it remains unknown if SANE programs truly provide more
compassionate care of victims than traditional emergency room personnel, and how this
compassionate care influences attrition rates. Further research is needed to understand if
and how these components of SAN E programs affect case attrition and the decision-
making process of police and prosecutors.

Prior research has indicated that police base their referring decisions in part on
characteristics of the victim. In particular, police have rejected cases if they did not think
the victim would make a credible witness, even if they believed the rape occurred
(F rohmann, 1997; Spears & Spohn, 1997). The literature further suggests that the

perception of victim credibility is partially based on characteristics of the victim such as

84

age, race, and social status. However, findings from the current study suggest that victim
characteristics do not predict police referring cases. These findings imply that police are
determining their referral decisions based on factors other than victim characteristics.
While these indices of victim credibility did not predict police referral, some of the case
characteristics that predicted police referral could be considered indicators of victim
credibility. For example, alcohol and drug use by the victim significantly predicted police
referral. Schuller and Stewart (2000) surveyed law enforcement officers, and found that
they perceived intoxicated victims as less credible, more interested in sexual intercourse,
more likely to communicate interest in sex, and more responsible for how far things had
progressed. It appears that alcohol and drug use still diminishes the victim’s credibility in
the eyes of many law enforcement personnel. Furthermore, police were somewhat more
likely to refer cases in which the offender and victim were intimate partners or family
members than when the offender and victim were acquaintances (e. g. friend, date). In
other words, police are referring cases in which the victim is/was engaged in a serious
relationship with or related to the offender but not when the offender is a friend, co-
worker, date, etc. The literature suggests that decisions to refer acquaintance rape cases
are determined largely by assessing the credibility of victims (Kerstetter & Van Winkle,
1990). Through interviews and observations, Kerstetter (1990) found that police were
less likely to refer cases when there was sexually discrediting information about the
victim. Furthermore, that study noted that police were less likely to refer the case if the
victim violated sex-role norms prior to the rape (e.g. being at home alone with a date;
leaving a party with the offender). Thus, it may be possible that police focus more on the

behavior of the victim rather than the offender for acquaintance rape cases. Taken

85

together, the findings on victim and case characteristics for this study imply that while
police are focusing less on types of victims (e.g. young, lower economic social status),
they are still basing their decisions in part on the credibility of the victim. However, case
referral was primarily predicted by forensic medical evidence (e. g. anogenital abrasion)
rather than victim or case characteristics. Overall, these results imply that police are
relying less on indices of credibility (e.g. age, relationship to offender), and more on
evidentiary factors (e.g. injuries).

Prior research has indicated that prosecutors base their warranting decisions in
part on case characteristics, specifically those that judge the credibility of the victim (e. g.
alcohol or drug use by the victim) (Spears & Spohn, 1996; Frohmann, 1997). However,
the results from this study indicated that case characteristics did not influence warranting.
These findings imply that prosecutors are making warranting decisions on factors other
than case characteristics. Previous studies indicate that injuries increase the probability of
cases being warranting. The current study also supported that finding. In fact, warranting
cases was primarily predicted by forensic medical evidence (e. g. physical abrasion) rather
than case characteristics. Similar to police, these results suggest that prosecutors are
focusing less on indices of credibility (e. g. alcohol or drug use, relationship to offender),
and more on evidentiary factors (e. g. injuries).

There was one unexpected finding when examining the predictors of prosecutors
warranting cases. The study found documented bruises only influenced one prosecutor to
warrant cases. This finding suggests that warranting decisions may be influenced by traits
of the individual prosecutors making the decisions in addition to evidentiary

characteristics. For example, the prosecutor who warranted a high volume of cases with

86

bruises may believe that the absence of injury indicates fabrication, or perhaps feels less
confident in “winning” the cases without documented bruises. On the other hand, the
prosecutors who warranted cases with less bruises may have more experience prosecuting
cases without injuries, and more willing to risk “losing” cases in trial. Furthermore, this
differential decision-making could have affected the number of SANE cases being
prosecuted. While the prosecution rates of the current study were higher than the
published rates of non-SANE cases, the rates were less than the published norms for
prosecution of SANE cases. Thus, it may be possible that having one prosecutor base
decisions heavily on the presence of injury may have led to a smaller number of cases
being prosecuted. This finding illustrates that a great deal of control rests in the hands of
individual prosecutors without clear mechanisms to provide oversight of their decisions.
Further studies should examine potential factors that may or may not influence individual
prosecutors in their warranting decisions such as training, case experience, knowledge of
forensic medical evidence, pressure to have high conviction rates, and organizational
policies.

To date, this is the first study focusing on predictors of case attrition specifically
with assaults processed by a SAN E program. In addition to reducing case attrition, the
findings of this study also imply that police and prosecutors are focusing more on
evidentiary factors of the case than factors regarding the victim’s credibility. It is also
important to note that the improved case attrition or a decreased focus on victim’s
credibility may result from historical effects not accounted for in this study. The majority
of case attrition studies occurred one to two decades ago. During the past decade, the

Violence against Women Act was instituted that changed laws regarding prosecution of

87

rape, and increased funding to change the legal system’s response to rape throughout the
United States. This funding was aimed to increase training of law enforcement officers
and prosecutors to more effectively identify and respond to rape, develop or expand units
of law enforcement and prosecutors specifically focusing on rape, and develop and
implement more effective police and prosecution policies and protocols in handling rape
cases (0.1 P, 2005). These efforts were specifically aimed to improve the response to rape,
increase reporting, and reduce case attrition. Therefore, the Violence against Women Act
may have contributed to the improvement in case attrition, and may have impacted police
and prosecutors focusing less on the credibility of victims and more on evidentiary
factors.
Limitations of this study

Several methodological limitations of this study may mitigate the strength of the
conclusions that can be drawn from this work. First, this study did not have a comparison
group of cases not processed by a SANE program to examine the unique contribution of
SANE programs in improving case attrition. Therefore, there may be other factors related
to improving case attrition not accounted for in this study. Comparative studies should be
conducted to examine the unique contribution of SAN E programs on case attrition
relative to non-SANE cases (e. g. those examined by traditional emergency room
personnel). Second, attrition rates and the predictors of case attrition can only be
generalized to rape cases that were processed through a SAN E program. Attrition rates
and predictors of case attrition may be different for those cases processed in traditional

emergency rooms, or those cases in which the victim did not seek medical assistance.

88

Third, there may be some methodological reasons why some of the variables did
not significantly predict charging decisions. In the analyses for referral decisions, the
power for a continuous predictor was adequate to detect a small effect, but the power for
a dichotomous predictor was only adequate to detect a medium effect. Therefore,
dichotomous variables with small effects on referring would not have been found in this
study. In the analyses for warranting decisions, the power for a continuous predictor was
adequate to detect even a small effect, but the power for a dichotomous predictor was
only sufficient to detect a large effect. Therefore, dichotomous variables with small to
medium effects on warranting would not have been discovered in this study. Fourth,
because secondary data analysis was used in this study, the conditions under which the
data were originally collected cannot be controlled. Assessing the consistency of the
documentation and data entry among the original data collectors (e. g. SANE nurses)
cannot be assessed. Although selective recording is a threat to secondary data analysis,
the SANE patient chart requires responses to all parts of their medical forms. For
example, leaving the section on injury blank is unacceptable in SANE practice; the nurses
are required to document that no injury is noted. However, a limitation of their database
existed in consistently tracking offender tactics. Because 30 cases had missing data on
offender tactics, it was excluded from the study. Thus, this study was not able to examine
the role of offender tactics on cases moving forward in the legal system, which previous
studies have shown to impact case attrition (Kerstetter 1990; Campbell et al., 2001). A
final limitation of this study is that only a small group of all the possible factors that
could affect charging decisions were studied. There may be other variables that were not

examined in this model that may impact charging decisions. For example, organizational

89

characteristics (e. g. available resources) of law enforcement agencies may influence
referring decisions but were not included in this study. Furthermore, characteristics of the
offenders may have influenced cases being referred and warranted.

Implications for Research and Practice

This study can serve as a catalyst for several research projects. First, this study
indicates that case attrition of rape case has improved in the last 25 years. However, the
reasons for attrition improvement remain unknown. Comparative studies should be
conducted to examine the unique contribution of SANE programs on case attrition
relative to traditional emergency rooms, as well as cases in which the victim did not seek
medical assistance. If future studies do find that SANE programs significantly improve
case attrition, a second line of research would examine the components of SANE
programs that improve rape case attrition. For example, does collaboration between the
SANE program and law enforcement agencies increase cases being referred by the
police? If so, why does collaboration increase case referral? Another key research issue
raised by these results is why injuries play a more important role in decisions of some
prosecutors, and less of a role for others. A third line of research that stems from these
findings would involve exploring individual factors (e.g. experience) of prosecutors that
influence their decision-making process for warranting rape cases.

The findings of this study also have several practice implications for SANE
programs. First, specific types of injury increased the probability of cases being
forwarded through the legal system. Thus, SANE programs should continue to place
strong emphasis in training nurses in accurate assessment and documentation of injury.

However, some types of injury did not predict cases going forward in the legal system.

90

For example, anogenital tearing and the injuries listed under “other types of injuries”
such as petechiae did not predict referring. It may be possible that law enforcement
officers did not understand the nature of these injuries. Thus, consultation with police
may be necessary when cases have these types of injuries. It is important to note that
while forensic medical evidence primarily predicted referral, some indicators of victim
credibility also predicted referral. Therefore, SAN E programs that provide training to law
enforcement should include topics on the dynamics of rape (e. g. victim/offender
relationships; role of alcohol in rape) to decrease any harmful preconceived beliefs about
rape and rape victims. Furthermore, because police also base their decisions on other
factors regarding the victim’s credibility (e. g. alcohol use), SAN E programs should be
cautious in basing the success of their programs on their rates of prosecution. Instead,
SAN E programs should evaluate program impact in domains such as service delivery to
victims and quality of forensic evidence collection. These aspects of SANE programs are
more directly within programmatic control and reflect the work of the nurses. By
contrast, the results of this study suggest that legal case outcomes reflect the decisions

and actions of multiple stakeholders, both in and outside SANE programs.

91

APPENDIX A

A Comparison of SANE and Hospital Evidence Collection Methods

 

Training

 

Traditional Hospital Care

Sexual Assault Nurse Examiner (SANE)
Programs

 

 

No training/educational requirements or
guidelines exist for hospital personnel (see
APRI, 2004; Giardino et al., 2003; Littel,
2001)

 

International Association of Forensic
Nurses (IAFN) Educational Guidelines:

Part I : Didactic content:

Multidisciplinary team concept
Forensic nursing

SANE roles and responsibilities
Dynamics of sexual assault
Sexual assault forensic evaluation
Communication skills
History taking

Physical assessment

Detailed genital assessment
Physical evidence collection

Forensic photo documentation
Documentation

Evidence Evaluation
Nursing Management
Criminal Justice System
Ethics

Evaluation

Part II: Clinical Preceptorship
Component

Clinical skills to be completed with
instruction by a Registered Nurse or
physician:

0 Detailed genital inspection

 

92

 

 

 

 

o Speculum examination
0 Visualization with techniques and
equipment

Sexual assault forensic examinations
including evidence kit collection to be
reviewed by a practicing SANE or
physician. The SAN E candidate must
perform the required clinical skills until
proficiency is demonstrated.

Observation of criminal trial
proceedings.

 

 

Specialized Techniques and Equipment

 

Traditional Hospital Care

Sexual Assault Nurse Examiner (SANE)
Programs

 

 

Hospitals usually do not have access to
specialized forensic equipment and lack
training in the use of the equipment. (see
APRI, 2004; Giardino et al., 2003; Littel,
2001)

 

Colmscope—A lighted magnifying
instrument used for examination of the
anogenital area. A camera is attached to the
colposcope to photodocument anogenital
injuries. Using a colposcope when
assessing for genital trauma greatly
increases the identification of genital
trauma (Slaughter & Brown, 1992). With a
colposcope, a FNE is three times as likely
to detect microlacerations, bruises and
other injuries in sexual assault victims

(V oelker, 1996).

F olev catheter technique—A foley catheter
may be used with pubertal victims to
examine the hymenal structure by inserting
the catheter into the vagina, and inflating
the catheter balloon with air and gently
pulling back to delineate the hymenal
structure (Giardino et al., 2003).

Toulidine Blue dve—A nuclear stain that
adheres to cellular material and enhances
the identification of injury. The dye causes
the injured area to darken significantly,

 

93

 

 

 

 

making visualization of an injury much
easier (Giardino, 2003). McCauley,
Guzinski , Welch, Gorman, and Osmers
(1987) found tears in only 4% using gross
visualization, but after toluidine blue was
applied, the rate increased to 58%.

Alternative light source—A special light
that can detect bodily fluids that may not

be detected with visual inspection alone.
All stains that are detected on the victim
are swabbed by rolling a saline-dampened
swab over the area (Giardino, 2003).

 

 

Forensic Medical Exam and Evidence Collection Procedures

 

Traditional Hospital Care

Sexual Assault Nurse Examiner (SAN E)
Programs

 

 

Emergency room physicians follow the
instructions in the rape evidence kit as they
perform the exam. During 1999 to 2002,
hospitals in Michigan used the following
instructions:

The instructions direct the nurse/physician
to document pertinent medical history and
history of the assault in the victim’s own
words.

The itemized clothing is placed in
containers separately and tagged for
evidence.

 

Turning Point’s FNE Program follows the
guidelines of the American College of
Emergency Physicians in conducting
exams:

An advocate meets with the victim to
attend to the victim’s emotional needs by
providing crisis intervention and emotional
support and the victim’s physical needs.

When the victim is ready, the nurse begins
taking a patient history. This history
includes relevant medical information, an
open-ended question that documents an
account of the rape in the victim’s words
and direct questions about the types of
contact (e. g. vaginal penetration).

After the history taking, the victim is
escorted to the exam room. The FNE nurse
places a barrier (e.g. sheet) on the floor or
exam table and then places paper on top of
the barrier. In order to preserve any
evidence that may fall off the victim’s
clothing, the victim undresses while on the
paper. The FNE nurse bags each piece of
the victim’s clothing into separate bags and

 

94

 

 

 

Not included in the instructions
Traditional emergency room care does not
have access to an alternative light source.

Not included in the instructions

The victim’s oral cavity is swabbed using
two swabs. Using both swabs, the nurse
makes two smears. Allow swabs and
smears to dry.

The victim’s head hair is combed and the
combings are placed in a labeled envelope.
Then 12 head hairs are pulled at the root
using gloved hands and placed in a labeled
envelope. If applicable, the nurse combs
the pubic hairs and places the combings in
a labeled envelope and 6-8 pulled pubic
hairs are saved in another envelope.

The victim’s vaginal tract is swabbed using
two swabs. Then using two more swabs,
the nurse swabs the vaginal tract for DNA.
Using both swabs, the nurse makes two
smears. Allow swabs and smears to dry.

 

seals the bags properly with evidence tape
to preserve the chain of custody. The paper
is also bagged and sealed. The clothing
serves an important source for the suspect’s
hair, fibers, semen and other items that may
have been transferred during the assault.

Depending on the history and consent of
the victim, the nurse scans the alternate
light source over the victim’s body looking
for blood, saliva or semen. The alternative
light source is a special light that can detect
bodily fluids that may not be detected with
visual inspection alone. The FN E nurse
swabs all stains that are detected or based
on victim history by rolling a sterile water-
dampened swab over the area.

The nurse does a head-to-toe exam
assessing for tenderness, redness, bruises,
cuts or abrasions. If injury is found, the
nurse measures, photographs and
documents the injury on a body diagram.
The victim’s oral cavity is inspected for
signs of trauma, which includes bruises
around the mouth, torn frenulum of the
lower lip or beneath the tongue. Two oral
swabs are taken and air-dried.

The nurse checks the victim’s hair for any
trace evidence. The victim’s head hair is
combed and the combings are placed in a
labeled envelope. Then 25-30 head hairs
are pulled at the root using gloved hands
and placed in a labeled envelope. If
applicable, the nurse combs the pubic hairs
and places the combings in a labeled
envelope and 25-30 pulled pubic hairs are
saved in another envelope.

The nurse then does a visual exam of the
vulva including the inner and outer labia
with gentle labia traction. The nurse looks
for any signs of tearing, lacerations or other
trauma to the area sometimes employing
Toluidine blue dye to the area to detect

 

95

 

 

 

The victim’s rectal area is swabbed using
two swabs. Using both swabs, the nurse
makes two smears. Allow swabs and
smears to dry.

Documentation of injury: the kit contains a
body diagram with instructions to “describe
presence of trauma”.

For DNA, a blood sample is drawn. To
determine the victim’s blood group and
secretor status, a paper disk is placed under
the victim’s mouth and thoroughly
saturated with salvia. The paper disk is air
dried and placed in the envelope.

The nurse packs all of the evidence into the
rape kit, and seals it. The kit may be turned
over to the police department immediately.
The instructions do not provide guidelines
for kit storage if the kit is not picked up
immediately.

The victim is referred to their physician for
follow-up and handed a list of rape crisis
centers in Michigan for counseling. An
advocate may speak with the victim, family
or significant others if the hospital requests
an advocate from the local rape crisis
center.

 

microlacerations. The colposcope may be
used at this point in the exam. A
colposcope is a lighted magnifying
instrument used for examination of the
vulva. A camera is attached to the
colposc0pe to digitally document genital
injuries. The posterior fourchette is
swabbed twice with a cotton swab and
smeared on a glass slide. If the victim is
menarche, the vaginal orifice is swabbed
with four swabs at one time. All swabs are
air dried before placing them in the proper
envelopes.

The nurse then examines the rectal area.
There are four swabs of the anal opening
and then two slides are prepared and then
two slides are prepared.

Documentation - if injury is found, the
nurse measures and photographs the injury
and documents the injury on a body
diagram.

For DNA, the nurse uses buccal swabs. The
nurse brushes the buccal mucosa in the
victim’s mouth using the collector end of
the swabs for 5-10 seconds. The swabs is
air dried before placing it in the proper
envelopes

The nurse packs all of the evidence into the
rape kit, and seals it. The kit is either
turned over to the police department
immediately or locked in a fiidge until the
police pick the kit up.

The advocate meets with the victim and
significant others if present to discuss their
concerns, provide information about next
steps in the legal system and help prepare
them for possible trauma reactions. The
victim is referred to their physician for

 

96

 

 

 

(Michigan State Police Rape Kit
Instructions, 1994)

 

follow-up and to Turning Point for free
counseling.

(Turning Point’s Forensic Nurse Training
Manual)

 

97

 

APPENDIX B

Adult Sexual Assault Case Coding Sheet

ID Number

 

Data Collector

 

Data Sources

 

 

 

 

 

 

 

 

 

 

 

Victim Characteristics: Circle Code/Enter Comments
Data
1a Age (Actual age in
years)
99=unknown
1b Gender 1=male
0=female
99=unknown
1c Race/ethnicity 1=White/Caucasian
2=African American
3=Latino/a
4=Native American
5=Multi-racial
6=Other
99=unknown
1d Disability 1=yes
2=no
99=unknown
1e Income
99999=unknown
Case Characteristics: Circle Code/Enter Comments
Data
2a Time between assault
and medical exam
(hours)
9.99=unknown
2b Offender tactics l= Fear
2= Verbal

 

 

3= Alcohol/drugs
4= Authority figure
5=Physical

6= Weapon used

7= Tied up/Bondage
8= Kidnapped

9= Unconscious

 

 

98

 

 

1 0=Other
99=unknown

 

2c

Victim relationship to
offender

1=stranger
2=acquaintance
3=friend

4=authority figure
5=parent/ guardian
6=step-parent
7=sibling
8=non-related care
giver

9=other relative
10=partner
11=ex-partner

1 2=spouse/life-time
partner

1 3=ex-spouse/ex-LTP
14=employer/coworker
15=other (specify)
16=multiple (specify)
99=unknown

 

2d

Victim consumed

alcohol before or during

the sexual assault

1 =yes
0=no
99=unknown

 

2e

Victim consumed drugs
before or during the
sexual assault

1 =yes
=no
99=unknown

 

2f

Type of sexual assault

l=vaginal penetration
2=oral penetration
3=anal penetration
4=multiple penetration
99=unknown

 

 

28

 

Nurse Examiner

 

1 =Ballinger
2=Barton
3=Bohach
4=Boni
5=Brooks
6=Burton
7=Creger
8=Dean-Mahan
9=Deboer

1 0=Diegel

1 1=Geiman

1 2=Gentile/pena

 

99

 

 

 

l 3=Glover
14=Halleck
15=Hejza
16=Henley

1 7=Hovan

1 8=Hunwick
1 9=Hurst
20=Johnson
21=Khalife
22=Korenek
23=Lamb
24=Lippert
25=Marchesi
26=Matheny-Lane
27=Meshinski
28=Moore
29=Scott
30=Spears

3 1=Starke
32=Tobin
33=Troszak
34=Webster
3 5=Zawacki

 

 

2h

 

Law Enforcement
Agency

 

1 =Centerline
2=Chesterfield Twp
3=Clay Twp
4=Clinton Twp
5=Eastpointe
6=Fraser

7=Mount Clemens
8=Macomb County
Sheriff
9=Michigan State
Police

10=New Baltimore
1 1=New Haven
12=Richmond
13=Romeo
14=Roseville
15=St. Clair Shores
16=Selfiidge
17=Shelby
18=Sterling Heights
19=Utica
20=Warren
99=Unknown

 

100

 

 

 

 

 

 

 

 

 

 

Evidence: Circle Number of Comments
Code/Enter injuries
' Data
3a Physical injury 1=yes
0=no
99=unknown
3b Redness (erythema) 1=yes
0=no
3c Tear/laceration 1=yes
0=no
3d Bruising/hematoma 1=yes
0=no
3e Abrasions 1=yes
0=no
3f Bleeding 1=yes
0=no
3g Oozing injury 1=yes

 

   

2—110
‘ F! 4.71:! ,9: - ,_ ~ *3

 

 

 

 

 

 

 

 

 

 

=no

99=unknown
4b Redness (erythema) 1=yes
0=no
4c Tear/laceration 1=yes
=no
4d Bruising/hematoma 1=yes
0=no
4e Abrasions 1=yes
=no
4f Bleeding 1=yes
0=no
4g Oozing injury 1=yes
2=no
4h Other injury (write type 1=yes
of injury in comment =no

 

box)

 

 

 

 

101

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Evidence continued: Circle Code/Enter Comments
Data
5a Trace evidence 1=yes
0=no
99=unknown
5b Type of trace evidence
Case decisions: Circle Code/Enter Comments
Data
6a Police Referred case 1=yes, referred
0=no, not referred
6b Prosecutor warranting 1=yes, warranted
decision 0=no, not warranted
6c Case outcome l=case warranted, but
later dropped
2=plea bargain
3=trial, conviction
4=trial, acquittal
99=unknown
Non-referred Stranger Circle Code/Enter Comments
Cases Data
9a Was the stranger in this 1=yes, offender was

 

case identified?

 

identified

0=no, offender was not
identified

99-not applicable

 

 

102

 

 

 

APPENDIX C

Operational and Coding Definitions

 

 

 

 

 

 

serve as a proxy for the victim income.

tract block. Income was searched by an
employee of the focal program to limit
access to the victims’ addresses. The
median household income will then

 

f. | . .. . 1
.7 AL . ‘l
1 _ a

 

Victim Operational Definition: Coded as:
Characteristics:
Age Age of the victim on the day of the Actual age in years
exam
Race/ethnicity Racial/ethnic group that the victim Caucasian — the victim
identifies as their own. identifies as Caucasian
Non-Caucasfin — the victim
identifies as anything
besides Caucasian including
bi-racial and multi-racial.
Disability The victim has at least one of the M: the victim does not
following disabilities: physical (e. g. have at least one disability.
hearing impaired), psychiatric (e. g.
bipolar) and cognitive disabilities (e.g. E — the victim has at least
Down syndrome). one disability
Median Address of where the victim resides Median household income
household was used to search the median fiom 2000 census data
income household income in the 2000 census

 

.

 

Case

”Operational Definition:

’ Coded as: i i

 

 

 

offender before or during the assault to
carry out the assault.

Characteristic
Offender tactics The most severe tactic used by the Fear — the victim

acquiesced out of fear

Verbal - the offender made
threats to the victim

Alcohol/drugs - offender
intentionally drugged them
or coerced them into
drinking in order to rape the
victim

 

 

103

 

 

Authority figge - the
offender had an

authoritative role to the
victim (e.g. teacher, boss,
and police officers)

Physical — the offender uses
force such as hitting,
holding victim down,
pushing the victim, etc.

Weapgn used — A weapon
(e.g. gun, knife, club, etc)
was present or used before
or during the assault

Tied up/Bondage - offender
tied up the victim or used
other means to constrain the
victim

Kidnapped - offender
abducted the victim

Unconscious — the victim is
not conscious due to being
asleep, or passed out

Other - any other type of
tactic but a list isn’t
available

Unknown — SANE did not
get this information from
the history

 

Victim
relationship to
offender

The relationship between the victim

and offender at the time of the assault.

Stranger — victim did not
know the offender at the
time of the rape

Nonstranger — victim knew
the offender at the time of
the rape

 

 

Alcohol or drugs
use by victim

 

Victim consumed any amount of
alcohol or drugs before or during the
sexual assault.

 

M — the victim did not
consume alcohol or drugs
(except prescribed drugs)

 

104

 

 

before or during the assault

fig — the victim did
consume alcohol or
nonprescribed drugs before
or during the assault

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

epidermis or superficial dermis and

 

Type of sexual The orifice of the victim that was Vaginal penetrzmon - the
assault penetrated by the offender. offender penetrated the
victim’s vagina
OraLrenetration — the
offender penetrated the
victim’s mouth with
offender’s genitals
Aggl penetration — the
offender penetrated the
victim’s anus
Multiple penetration — the
offender penetrated more
than one orifice of the
victim
Medical Operational definition: Coded as: Number of
Forensic injuries
Evidence:
Physical injury An injury (redness, tear/laceration, Total number of physical
bruising/hematoma, abrasion) on the injuries
victim’s body except the anogenital
area.
Redness An area that is only red without Total number of areas noted
(erythema) additional injury located on a as red
nongenital part of the body..
Tear/laceration Continuity of skin is broken and Total number of tears
disrupted by force and located on a
nongenital part of the body.
Bruising/hemato Collection of blood below the intact Total number of bruises
ma epidermis that leaked from ruptured
capillaries or blood vessels and located
on a nongenital part of the body
(Giardino et al., 2003).
Abrasions Superficial injuries limited to the Total number of abrasions

 

105

 

 

located on a nongenital part of the
body (Giardino et al., 2003).

 

 

 

 

 

 

 

 

 

Bleeding Any injury described as bleeding Total number of injuries
located on a nongenital part of the described as bleeding
body.

Oozing injury Any injury described as oozing located Total number of injuries
on a non enital art of the bod . described as oozin

Anogenital An injury (redness, tear/laceration, Total number of anogenital

injury bruising/hematoma, abrasion) on the injuries
anogenital area.

Redness An area that is only red without Total number of areas noted

(erythema) additional injury located on an as red
anogenital part of the body.

Tear/laceration Continuity of skin is broken and Total number of tears
disrupted by force and located on an
anogenital part of the body (Giardino
et al., 2003).

Bruising/hemato Collection of blood below the intact Total number of bruises

ma epidermis that leaked from ruptured
capillaries or blood vessels and located
an anogenital part of the body
(Giardino et al., 2003).

Abrasions Superficial injuries limited to the Total number of abrasions
epidermis or superficial dermis and
located on an anogenital part of the
body (Giardino et al., 2003).

Bleeding Any injury described as bleeding Total number of injuries
located on an anogenital part of the described as bleeding
body.

Oozing injury Any injury described as oozing located Total number of injuries
on an anogenital part of the body. described as oozing

Other injury Any other injury not listed. Total number of other

injuries

 

 

Trace evidence

 

A material (e. g. condom) not normally
found on the body is identified by the
nurse examiner.

 

Yes — trace evidence was
noted on the victim’s body

M — trace evidence was not
noted on the victim’s body

Type of trace evidence will
be recorded

 

106

 

 

 

Case Processing

Coded as:

 

 

Operational Definition
Decisions:
Police Referral Law enforcement referred the case to Les — case was referred to
decision the prosecutor’s office. the prosecutor’s office
No — case was not referred
to the prosecutor’s office
Prosecutor The prosecutor warranted the case. Yes — case was warranted
warranting by the prosecutor’s office
decision

Np — case was not
warranted by the
prosecutor’s office

 

 

Case outcome

 

The final disposition of the case in the
legal system.

 

Case charged, but later
dropped — case was
originally warranted by the
prosecutor but subsequently
dropped before ending in a
disposition

Plea bargain — the offender
pled guilty to the original
charges or less charges

Trial, conviction- the case
went to trial and the
offender was found guilty

by ajudge orjury

Trial, acquittal — the case
went to trial and the
offender was found not

guilty by a judge or jury

 

107

 

APPENDIX D

Comparison of Original Coding of Variables vs. Coding of Variables for Analyses

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Victim Characteristics: Orfla] Coding CodingUsed for Analyses
1a Age Actual age in years Age was log transformed
1b Gender 1=male 1=male
0=female 0=female
1c Race/ethnicity l=White/Caucasian 1=White/Caucasian
2=African American 2=Minority
3=Latino/a
4=Native American
5=Multi-racial
6=Other
1d Disability 1=yes 1=yes
=no 2=no
1e Income Median household Median household income
income arresented in thousands)
Case Characteristics: Orijifil Coding Coding Used for Analyses
2a Time between assault Time between assault Time between assault and
and medical exam and medical exam in medical exam was log
hours transformed
2b Offender tactics 1= Fear =unconscious (originally 9)
2= Verbal 1=coercion (originally 1-4)
3= Alcohol/drugs 2=force (originally 5-8)
4= Authority figure
5=Physical -other and unknown coded as
6= Weapon used missing
7= Tied up/Bondage -because 16% of the data was
8= Kidnapped missing, offender tactics was
9= Unconscious excluded from the model
10=Other
99=unknown
2c Victim relationship to 1=stranger 0=acquaintance
offender 2=acquaintance (originally 2-4;8, 14)
3=friend 1=stranger (originally 1)
4=authority figure 2=intimate/familial
5=parent/guardian (originally 5-7; 9-13)

 

108

 

 

6=step-parent
7=sibling
8=non-related care
giver

9=other relative

1 0=partner
11=ex-partner

1 2=spouse/life-time
partner

1 3=ex-spouse/ex-LTP
14=employer/coworker
15=other (specify)
16=multiple (specify)

 

2d

Victim consumed

alcohol before or during

the sexual assault

1 =yes
0=no

 

2e

Victim consumed drugs

before or during the
sexual assault

1 =yes
0=no

Victim consumed alcohol/
drugs before or during the
sexual assault

1 =yes
0=no

 

2f

Type of sexual assault

1=vaginal penetration
2=oral penetration
3=anal penetration
4=multiple penetration
99=unknown
penetration

1=single penetration
2=multiple penetration
0=unknown penetration

 

 

28

 

Nurse Examiner

 

1 =Ballinger
2=Barton
3=Bohach
4=Boni
5=Brooks
6=Burton
7=Creger
8=Dean-Mahan
9=Deboer

1 0=Diegel

1 1=Geiman

1 2=Gentile/pena
l 3=Glover
14=Halleck

1 5=Hejza

1 6=Henley

1 7=Hovan

l 8=Hunwick

1 9=Hurst
20=Johnson

 

Each nurse was reassigned a
code of the number of exams
they conducted

 

109

 

 

2 1 =Khalife
22=Korenek
23=Lamb
24=Lippert
25=Marchesi
26=Matheny—Lane
27=Meshinski
28=Moore
29=Scott
30=Spears

3 l=Starke
32=Tobin
33=Troszak
34=Webster
35=Zawacki

 

2h

Law Enforcement
Agency

1=Centerline
2=Chesterfield TWp
3=Clay Twp
4=Clinton Twp
5=Eastpointe
6=Fraser

7=Mount Clemens
8=Macomb County
Sheriff
9=Michigan State
Police

10=New Baltimore
11=New Haven
12=Richmond
13=Romeo
14=Roseville
15=St. Clair Shores
16=Selfridge
17=Shelby
18=Sterling Heights
19=Utica
20=Warren

0=law enforcement agencies
located in lower income
communities

1=law enforcement agencies
located in higher income
communities

 

 

 

Law Enforcement
Agency (as proxy for
Prosecutor area)

 

1=Centerline
2=Chesterfield TWp
3=Clay TWp
4=Clinton TWp
5=Eastpointe
6=Fraser

7=Mount Clemens
8=Macomb County

 

0=area/prosecutor one
Clinton TWp
Mt. Clemens
Macomb Sheriff
Sterling Heights
MSP
Selfridge

 

110

 

 

 

 

 

Sheriff
9=Michigan State
Police

10=New Baltimore
11=New Haven

1=area/prosecutor two
1/2 of Warren
Fraser
Roseville
St. Clair Shores

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

 

12=Richmond New Baltimore
13=Romeo Richmond
14=Roseville
15=St. Clair Shores 2=area/prosecutor three
16=Selfridge ‘/2 of Warren
17=Shelby Centerline
18=Sterling Heights Eastpointe
19=Utica Chesterfield
20=Warren New Haven
Shelby
Utica
Romeo
Evidence: Original Coding Used for Analyses
Coding
3a Physical injury 1=yes 1=yes
0=no =no
99=unknown
3b Redness (erythema) lwes 1=yes
0=no 0=no
3c Tear/laceration 1=yes Only 3% of sample had tearing,
0=no so it was not used in the
analyses
3d Bruising/hematoma 1=yes 1=yes
0=no 0=no
3e Abrasions 1=yes 1=yes
0=no 0=no
3f Bleeding 1=yes 1=yes
=no 0=no
3g Oozing injury 1=yes 1=yes
2=no 0=no
Anogenital injury 1=yes 1=yes
0=no 0=no
4b Redness (erythema) 1=yes 1=yes
0=no =no
4c Tear/laceration 1=yes 1=yes
0=no 0=no
4d Bruising/hematoma 1=yes 1=yes
0=no 0=no

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

4e Abrasions 1=yes 1=yes
0=no 0=no
4f Bleeding 1=yes 1=yes
0=no 0=no
4g Oozing injury 1=yes 1=yes
2=no 0=no
4h Other injury (write type of 1=yes 1=yes
injury in comment box) 0=no 0=no
Evidence continued: Orignal Coding Codingflsed for Analyses
5a Trace evidence 1=yes 1=yes
0=no 0=no
5b Type of trace evidence
Case decisions: Original Coding Coding Used for Analyses
6a Police Referred case 1=yes, referred
0=no, not referred
6b Prosecutor warranting 1=yes, warranted
decision 0=no, not warranted
6c Case outcome l=case warranted, but
later dropped
2=plea bargain
3=frial, conviction
4=trial, acquittal
99=unknown
Non-referred Stranger Circle Code/Enter Comments
Cases Data
9a Was the stranger in this 1=yes, offender was 1=yes, offender was identified

 

case identified?

 

identified

0=no, offender was not
identified

99-not applicable

0=no, offender was not
identified
99-not applicable

-cases were removed from
sample if stranger was not
identified

 

 

112

 

 

APPENDIX E

Comparison of Published Attrition Rates of Cases and Current Study

 

 

Stage in the Legal System Attrition Rates Attrition Rates of Previous
of Current Study Studies“
Cases referred by the police 49% (N=90) 41% (Chandler & Tomey, 1981)

44% (Galvin & Polk, 1982)
22% (Frazier & Haney, 1996)
18% (Bouffard, 2000)
38% (Crandall & Helitzer,
2003).

All reported cases prosecuted 25% (N=47) 17% (Galvin & Polk, 1982)
18% (Chandler & Tomey, 1981)

16% (Frazier &Haney, 1996)

All reported cases that resulted in 24% (N=44) 12% (LaFree, 1980)
conviction (plea-bargain/guilty 17% (Chandler & Tomey, 1981)
verdict) 7% (Galvin & Polk, 1982)

12% (Frazier & Haney, 1996)
All reported cases ending in prison 11% (N=21) 9% (LaFree, 1980)
sentence 6% (Galvin & Polk, 1982)

7% (Frazier & Haney,l996)

*Samples from prior studies were obtained from police departments and/or hospitals not
Sexual Assault Nurse Examiner Programs.

113

REFERENCES

Amdt, S. (1988). Nurses help Santa Cruz sexual assault survivors. California
Nurse, 84.

Bachar & Koss (2001). From prevalence to prevention: closing the gap between
what we know about rape and what we do. In Renzetti, C., Edelson, J. & Bergen,
R.K. (eds.), Sourcebook on Violence Against Women. Thousand Oaks, CA: Sage
Publications, Inc.

Bachman, R. (1998). The factors related to rape reporting behavior and arrest:
New evidence from the National Crime Victimization Survey. Criminal
Justice and Behavior, 25(1), 8-29.

Bouffard, J. (2000). Predicting type of sexual assault case closure from victim, suspect
and case characteristics. Journal of Criminal Justice, 28, 527-542.

Bradmiller, L.L. & Walters, W.S. (1985). Seriousness of sexual assault charges:
Influence factors. Criminal Justice and Behavior, 12(4), 463-484.

Burgess, A.W. & Holmstrom, LL. (1974). Rape Trauma Syndrome. American
Journal of Psychiatry, 131, 981-986.

Campbell, R. (1998). The community response to rape: Victims’ experiences with
the legal, medical and mental health systems. American Journal of Community
Psychology, 25, 355-379.

Campbell, R. & Raja, S. (1999). Secondary victimization of rape victims: Insights
from mental health professionals who treat survivors of violence. Violence
and Victims, 14(3), 261-275.

Campbell, R., Wasco, S., Ahrens, C., Sefl, T., & Barnes, H. (2001). Preventing
the “second rape”: Rape survivors’ experiences with community service
providers. Journal of Interpersonal Violence, 16(12), 1239-1259.

Chandler, S.M. & Tomey, M. (1981). The decisions and the processing of rape
victims through the criminal justice system. California Sociologist, 4, 155-169.

Chappell, D. Geis, R. & Geis, G. (Eds) (1977). Forcible rape: the crime, the victim, and
the oflender. New York: Columbia University Press.

Cleveland, H., Koss, M & Lyons, J. (1999). Rape tactics from the survivors’ perspective:

Contextual dependence and within-event independence. Journal of Interpersonal
Violence, 14(5), 532-547.

114

Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and
Psychological Measurement, 20, 37-46.

Cohen, L., Donahue, J. & Kovener, M. (1996). A SANE approach to treating sexual
assault survivors in Colorado: Utilizing sexual assault nurse examiners.
Denver,CO: Colorado Coalition against Sexual Assault.

Crandall, C. & Helitzer, D. (2003). Impact evaluation of a Sexual Assault Nurse
Examiner (SANE) Program. NIJ Document No: 203276; Award Number 98-WT-
VX-0027.

Derhammer, F., Lucente, V., Reed, J. & Young, M. (2000). Using a SANE
interdisciplinary approach to care of sexual assault victims. The Journal on
Quality Improvement, 26(8), 488-495.

Feldman-Summers, S. & Palmer, G.C. (1980). Rape as viewed by judges, prosecutors
and police officers. Criminal Justice and Behavior, 7(1), 19-40.

Field, A. (2000). Discovering statistics using SPSS for Windows. Thousand Oaks,
California: Sage.

Frazier, P., & Haney, B. (1996). Sexual assault cases in the legal system: Police,
prosecutor and victim perspectives. Law and Human Behavior, 20(6), 607-628.

F rohmann, L. (1997). Convictability and discordant locales: Reproducing
race, class, and gender ideologies in prosecutorial decision making. Law and
Society Review, 31 , 531-556.

Frohmann, L. (1998). Constituting power in sexual assault cases: Prosecutorial
strategies for victim management. Social Problems, 45(3), 393-407.

Galton, B. (1975-1976). Police processing of rape complaints: A case study. American
Journal of Criminal Law, 4, 15-30.

Galvin, J. & Polk, K. (1983). Attrition in case processing: Is rape unique? Journal of
Research in Crime and Delinquency, 20, 126-153.

Giardino, A., Datner, E., & Asher, J. (2003). Sexual assault: Victimization across the life
span: a clinical guide. St. Louis, Mo.: G.W. Medical Publishing.

Green, SB. (1991). How many subjects does it take to do a regression analysis?
Multivariate Behavioral Research, 26(3), 499-510.

Hazelwood, R.R., & Burgess, A.W. (Eds.). (2001). Practical aspects of rape

investigation: A multidisciplinary approach (3'd ed). Boca Raton, FL.: CRC Press
LLC.

115

Heintze, J. (2001). NCSS and PASS 2002. Kaysville, UT: Number Cruncher Statistical
Systems.

Herman, J. (1992). Trauma and Recovery. New York: Basic Books.

Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression (2nd ed.). New
York: Wiley.

International Association of Forensic Nurses (2004). Sexual Assault Nurse Examiner
educational guidelines. Pitrnan, NJ: Author.

Kerstetter, W. (1990). Gateway to justice: police and prosecutor response to sexual
assault against women. Journal of Criminal Law & Criminology, 81(2) 267-313.

Kerstetter, W. & Van Winkle (1990). Who decides? A study of the complainant’s
decision to prosecute in rape cases. Criminal justice and behavior, 1 7(3),
268-283.

King, E. N., & Ryan, T. P. (2002). A preliminary investigation of maximum likelihood
logistic regression versus exact logistic regression. American Statistician, 56, 163-
170.

Koss, M.P., Gidycz, C.A. & Wisniewski, N. (1987). The scope of rape: Incidence
and prevalence of sexual aggression and victimization in a national sample of
higher education students. Journal of Consulting and Clinical Psychology, 55,
162-170.

Koss, MP. (1993). Rape: Scope, impact, interventions, and public policy
responses. American Psychologist, 48, 1062-1069.

Koss, M.P. & Heslet, L. (1992). Somatic consequences of violence against
women. Archives of Family Medicine, 1(1), 53-59.

LaFree, G. (1980). The effect of sexual stratification by race on official reactions
to rape. American Sociological Review, 45(5), 842-854.

LaFree, G. (1981). Official reactions to social problems: police decisions in sexual
assault cases. Social problems, 28(5), 582-594.

Lang, K. (1999). Sexual assault nurse examiner resource guide for Michigan
communities. Okemos, MI: Michigan Coalition against Domestic and Sexual
Violence.

Ledray, L. (1992). The sexual assault nurse clinician: A fifteen-year experience in
Minneapolis. Journal of Emergency Nursing, 18, 217-222.

116

Ledray, LE. (1998) Sexual assault: Clinical issues SANE expert and factual testimony.
Journal of Emergency Nursing, 24, 284-287.

Ledray, 1., & Simmelink, K. (1997). Efficacy of SANE Evidence Collection: a Minnesota
Study. Journal of Emergency Nursing, 23(1), 75-77.

LeGrand, C. (1975). Rape and rape laws: sexism in society and law. California Law
Review, 61, 919-941.

Littel, K. (2001). Sexual assault nurse examiner programs: Improving the community
response to sexual assault victims. Oflice for Victims of Crime Bulletin, April,
1-19.

Madigan, L. & Gamble, N. (1991). The second rape: society’s continued betrayal of the
victim. New York: Lexington Books.

Martin, P.Y. & Powell, M. (1995). Accounting for the second assault: legal
organization’s’ framing of rape victims. Law and Social Inquiry, 20, 853-890.

McCauley J ., Gyzinski, G., Welch, R., Gorman, R., & Osmers, F. (1987). Toulidine blue
in the corroboration of rape in the adult victim. American Journal of Emergency
Medicine, 5, 105-108.

Mehta, C., & Patel, N. (2002). LogXact 5 for Windows user manual. Cambridge, MA:
Cytel Software.

Mehta, C., & Patel, N. (2003). StatXact 6 for Windows user manual. Cambridge, MA:
Cytel Software.

Michigan State Police (1994). Michigan rape kit instructions. Lansing, MI: Author.

Patterson, D. (2004). The Current Practices of Michigan SANE Programs. Okemos, MI:
Michigan Coalition against Domestic and Sexual Violence.

Pett, MA. (1997). Nonparametric statistics for health care research. Thousand Oaks,
CA: Sage.

Rambow, B., Adkinson, C., Frost, T., & Peterson, G. (1992). Female sexual assault:
medical and legal implications. Annals of Emergency Medicine, 21(6), 727-731.

Rose, V, and Randall, S. (1984). The impact of investigator perceptions of victim
legitimacy on the processing of rape/sexual assault cases. Symbolic Interaction,
5(1), 23-36.

Russel, D. (1984). Sexual exploitation: Rape, child sexual abuse and workplace
harassment. Beverly Hills, CA: Sage.

117

Schuller, R. and Stewart, A (2000). Police Responses to Sexual Assault Complaints: The
role of perpetrator/complainant intoxication. Law and Human Behavior 24(5)
535-551

Sobsey, D. & Doe, T. (1991). Patterns of sexual abuse and assault. Sexuality and
Disability, 9(3), 243-259.

Sievers, V., Murphy, S., & Miller, J. (2003). Sexual assault evidence collection more
accurate when completed by sexual assault nurse examiners: Colorado’s
experience. Journal of Emergency Nursing, 29(6), 51 1-514.

Slaughter, L., & Brown, C. (1992). Colposcopy to establish physical findings in rape
victims. American Journal of Obstetrician Gynecology, 166(1), 83-86.

Spears, J., & Spohn, C. (1996). The genuine victim and prosecutors’ charging decisions
in sexual assault cases. American Journal of Criminal Justice, 20(2), 183-205.

Spears, J. & Spohn, C. (1997). The effect of evidence factors and victim characteristics
on prosecutors’ charging decisions in sexual assault cases. Justice Quarterly, 1 4
(3), 501-524.

Spohn, C., Beichner, D., & Davis-Frenzel, E. (2001). Prosecutorial justifications for
sexual assault case rejection: Guarding the “Gateway to Justice”. Social
Problems, 48(2), 206-235.

Spohn, C., & Homey, J. (1993). Rape law reform and the effect of victim characteristics
on case processing. Journal of Quantitative Criminology ,9(4), 383-409.

Tabachnick, B.G. & F idell, LS. (2001). Using Multivariate Statistics (4th ed.). Boston:
Allyn & Bacon.

Tjaden, P. & Thosennes, N. (1998). Prevalence, incidence and consequences of
violence against women. The National Violence Against Women Survey.

Washington, DC: Department of Justice.

Turning Point (2004). Turning Point ’s Forensic Nurse Training Manual. Mt. Clemens,
Michigan: Author.

Voelker, R. (1996). Experts hope team approach will improve the quality of rape exams.
Journal of the_American Medical Association, 275, 973-974.

Wriggins, J. (1983). Rape, racism, and the law. Harvard Women 's Law Journal, 6,
140-141.

118

          

MMMMM

111111111

. .1

 

[11111111111111111 .1

02736