Content uploaded by Gary Kleck
Author content
All content in this area was uploaded by Gary Kleck on Sep 10, 2014
Content may be subject to copyright.
Resisting Rape: The Effects of Victim Self-Protection on
Rape Completion and Injury
Jongyeon Tark
Associate Professor of Criminology
Department of Police Administration
Hannam University
133 Ojeong-dong Daejeon, South Korea.
e-mail: crim2@hotmail.com
Gary Kleck (corresponding author)
Professor
College of Criminology and Criminal Justice
Florida State University
Tallahassee, Florida 32306
gkleck@fsu.edu
(850) 644-7651
January 5, 2013
2
Abstract
The impact of victim resistance on rape completion and injury was examined utilizing a
large probability sample of sexual assault incidents, derived from the National Crime
Victimization Survey for 1992-2002, and taking account of whether harm to the victim followed
or preceded self-protection (SP) actions. Additional injuries besides rape, particularly serious
injuries, following victim resistance are rare. Multivariate logistic regression estimates indicate
that most SP actions, both forceful and non-forceful, reduce the risk of rape completion, and do
not significantly affect the risk of additional injury. The effects of SP actions in sexual assaults
are similar to their effects in nonsexual assaults with female victims.
Rape and sexual assault remains widespread in America. It has been estimated that
approximately 20% of all women will be raped at some point through their life course (Koss, 1993).
Rape can cause long-term physical and emotional trauma to victims, including persisting fear,
anxiety, suspicion, confusion, anger, and even suicidal behaviors (Burgess & Holmstrom, 1974;
Kilpatrick, Resick, & Veronen, 1981). Completed rapes cause more psychological problems than
attempted rapes, and those involving physical injury cause even more posttraumatic stress disorder
(Kilpatrick et al., 1981).
Prior Research
Despite the seriousness of the problem, researchers and other authorities have failed to
provide consistent and specific guidance to potential victims concerning the best tactics to use in the
face of a rape attempt (Ullman, 1997). Researchers have generally agreed that victim resistance is
effective for avoiding rape completion (Clay-Warner, 2002; Guerette & Santana, 2010; Kleck &
Sayles, 1990; Marchbanks, Liu & Mercy, 1990; Ullman, 1998; Ullman, 2007; Ullman & Knight,
1992; Zoucha-Jenson & Coyne, 1993), but there is no consensus on which specific victim tactics are
most effective. Further, controversy remains concerning the impact of resistance, especially
forceful resistance, on whether the victim suffers any additional injuries beyond rape itself (Ullman,
1997). Some scholars have argued that victim resistance, especially forceful resistance, is useless
and even dangerous because it provokes rapists into inflicting additional injuries (e.g., Brecklin &
Ullman, 2001, p. 14; Griffin & Griffin, 1981; Marchbanks, Liu, & Mercy, 1990). Others have
concluded that resistance is generally either beneficial or does not increase the risk of additional
injury (Guerette & Santana, 2010; Kleck & Sayles, 1990; Ullman & Knight, 1992; Zoucha-Jensen
& Coyne, 1993).
Variation in findings on this point may be due in part to differences in the samples analyzed.
Some studies have been based on small nonprobability samples of crimes, typically local
2
convenience samples of incidents known to authorities, such as those reported to a single local law
enforcement agency (Ullman, 1998; Ullman & Knight, 1992; Zoucha-Jensen & Coyne, 1993),
incidents involving college students at a single campus (Levine-MacCombie & Koss, 1986), victims
who sought help from a particular rape crisis center (Atkeson, Calhoun, & Morris, 1989; Ruback &
Ivie, 1988), offenders incarcerated in a single institution or handled by a single treatment facility
(Ullman & Knight, 1992, 1993), or self-selected volunteer subjects (Bart 1981; Bart & O’Brien,
1985). Convenience samples of crimes that come to the attention of the authorities or treatment
personnel are afflicted by biases that bear directly on the apparent effectiveness of victim defensive
actions. Victims tend not to report to the police less serious crimes and those in which they suffered
no injuries or property loss (United States Bureau of Justice Statistics, 1985; Rennison, 2002), so
samples of crimes known to the authorities disproportionately exclude cases in which victim actions
were effective in preventing injury or property loss. Incidents reported to victim crisis centers or
treatment facilities would suffer from similar censoring of crimes with better outcomes for victims,
since such victims would be less likely to seek treatment or counseling.
Apparent conflicts in findings of studies may also be attributable to the failure of many
researchers to establish the sequence of protective actions and injury. Researchers who found
positive associations between injury and self-protection actions, and concluded that resistance
provoked offenders into attacking victims, failed to establish whether self-protective (SP) actions
preceded or followed the offender’s inflicting of injury (e.g., Atkeson et al., 1989; Block & Skogan,
1986; Brecklin and Ullman, 2001; Griffin & Griffin, 1981; Marchbanks, Lui, and Mercy, 1990;
Ruback & Ivie, 1988). In these studies, crimes where a victim was injured before doing something
to resist were effectively treated as cases in which resistance provoked injury. In contrast, studies
that established the injury-SP sequence have generally found that all or most types of resistance
either reduce the risk of subsequent injury or have no net effect one way or the other (Guerette &
Santana, 2010; Quinsey & Upfold, 1985; Tark & Kleck, 2004; Ullman & Knight, 1992).
3
Another problem in this research is the use of limited two- or three-category typologies of
resistance actions. Some researchers simply divide victims into those who resisted or did not resist,
or distinguish only physical (“forceful,” “direct,” “combative”) resistance from nonphysical
resistance (e.g., Block & Skogan, 1985; Marchbank et al., 1990; Ullman, 1998). These scholars
have generally concluded that physical resistance increased the risk of injury while it reduced the
likelihood of rape completion. Although the pre-1986 National Crime Victimization Survey
(NCVS) distinguished eight types of self-protective actions, and the post-1986 NCVS provides
information on 16 types, even researchers using this rich source of information have needlessly
grouped different types of victim actions into a few broad categories. In contrast, Kleck and Sayles
(1990) separately assessed all eight distinct categories of self-protection that were coded in the pre-
1986 NCVS and found that some forceful responses appeared to reduce the risk of injury while
others did not, and that some non-forceful responses appeared to be effective while others, such as
attempting to get help, seemed to increase the risk of injury. Thus, important differences in impact
can be lost by combining protection strategies.
Further, some researchers of rape and sexual assault studies have used few or no controls for
potential confounders. The absence of statistical controls is problematic because the choice of
resistance method and injury outcome are both strongly correlated with other variables, such as
types of offender attack or threat (Ullman & Knight, 1992), victim and offender alcohol
consumption (Breklin & Ullman, 2001), and victim-offender relationships (Atkeson et al., 1989;
Levine-MacCombie & Koss, 1986).
We test the following hypotheses with respect to each type of victim self-protective action:
(1) Self-protective action reduces the likelihood of a rape attempt being completed, and (2) Self-
protective action reduces the likelihood of a rape victim suffering nonsexual injury.
Our general expectation was that the more forceful a self-protective action was, the more likely it
was to deter the aggressor from continuing his efforts to rape the victim, and the more likely it was
4
to discourage rather than provoke offender attempts to inflict physical injury on the victim. Thus,
merely verbal resistance would have mild effects, unarmed physical resistance would have stronger
effects, and armed physical resistance would have the strongest effects in preventing rape
completion.
Armed resistance by rape victims appears to be quite rare, especially armed resistance with a
gun. This may be partly a reflection of NCVS victim-respondents’ reluctance to admit possibly
unlawful gun possession to federal government interviewers (Kleck & Gertz, 1995), but is also at
least partly due to the rarity with which women personally own guns or carry them away from the
home. U.S. adult women are less than one third as likely to personally own a handgun as men
(Kleck, 1997, p. 101), and less than half as likely as men to carry a gun on their person away from
home (Kleck & Gertz, 1998, p. 213). Thus, research samples typically include only a handful of
cases of women who used guns or other weapons to resist crime, and conclusions about the effects
of such use are correspondingly tentative. This limited evidence evidence indicates that armed
resistance is effective in preventing rape completion (Kleck & Sayles, 1990, p. 155), and has no
significant net effect on whether victims suffer any additional injury (Kleck & Sayles, p. 156; Tark
& Kleck, 2004, p. 888).
Method
Sample
The data for this study were generated by interviews conducted by the U.S. Census Bureau
in connection with the National Crime Victimization Survey (NCVS), covering crime incidents that
occurred in the United States from 1992 through 2002 (United States Bureau of Justice Statistics,
2000). Information on the sequence of victim harm and resistance actions was not gathered before
1992. Rapes involving female victims were selected for analysis according to the NCVS Type of
Crime (TOC) typology. The sample was composed of 733 unweighted cases of rape or attempted
rape with female victim and 1,278 sexual assaults that were not rapes or attempted rapes. Incidents
5
were weighted using a modified version of the NCVS Incident Weight, which reflects their differing
probabilities of selection into the sample.
Establishing the Occurrence of the Offense
The occurrence of a rape is established in the NCVS through a long and intricate series of
questions. First, “individual screen” questions are asked to establish the possible occurrence of
such a crime. After a series of broad questions inquiring about attacks or threats in general, the key
rape-related screen question was: “Incidents involving forced or unwanted sexual acts are often
difficult to talk about. [Other than any incidents already mentioned] have you been forced or
coerced to engage in unwanted sexual activity by – (a) someone you didn’t know before, (b) a
casual acquaintance, or (c) someone you know well?” Later, a long series of open-ended questions
are asked in the Crime Incident Report section of the interview to establish what sort of assault
occurred, including questions about exactly how the offender threatened or attacked the victim, or
attempted to do so (interviewers could code “unwanted sexual conduct with [or without] force,”
“verbal threat of rape,” “raped,” “tried to rape,” or “sexual assault other than rape or attempted
rape”), and what injuries were inflicted (including “raped” and “attempted” rape”) (see United
States Bureau of Justice Statistics 2000, p. 138-146 for full details).
(Table 1 about here)
Outcome (Dependent) Variables
Table 1 lists the variables included in the analyses. Most variables are binary, indicating the
presence or absence of an attribute. The types of injuries recorded in NCVS are: (1) raped, (2)
attempted rape, (3) sexual assault other than rape or attempted rape, (4) knife or stab wounds, (5)
gunshot, bullet wounds, (6) broken bones or teeth knocked out, (7) internal injuries, (8) knocked
unconscious, (9) bruises, black eyes, cuts, scratches, swelling, chipped teeth, and (10) other injuries.
Rape completion was measured using the contrast between raped vs. attempted rape (categories 1
and 2), while additional injury was coded based on injury categories (3) through (10). The exact
6
cut-off between more serious and less serious injury is necessarily somewhat subjective, but we
used the fairly conventional one adopted in past research using NCVS data: the last two categories
(9 and 10) were treated as less serious, the rest (3-8) as more serious.
As used in the various analyses, the dependent variables measured whether the victim (1)
suffered completion of rape, regardless of whether it occurred before or after self-protective actions
were taken by the victim (RAPED), (2) suffered completion of rape after taking some self-
protective action (POSTRAPE), (3) suffered any nonsexual injury (i.e., injury beyond the rape
itself), regardless of when it occurred (NOSEXINJ), (4) suffered any nonsexual injury after taking
some self-protective action (PONOSEXI), (5) suffered any serious nonsexual injury, regardless of
when it occurred (NOSEXSR), or (6) suffered any serious nonsexual injury after taking some self-
protective action (PONSEXSR). Since our dependent variables were all binary variables, we used
logistic regression to estimate models.
Determination of the Sequence of Injury and Self-Protection Actions
Injury that occurs before the victim took any SP actions obviously cannot be the result of
those actions. Some of our outcome measures, RAPED, NOSEXINJ, NOSEXSR measure whether
injury occurred, regardless of the temporal sequence of SP and injury - intentionally mimicking the
measures used in past research for purposes of replication. We then separately analyzed the
occurrence of rape suffered after taking some self-protective action (POSTRAPE), the experience of
some additional injury (i.e., injury besides rape itself) suffered after taking self-protective actions
(PONOSEXI), and serious additional injury suffered after taking self-protective actions
(PONSEXSR). These variables effectively code an injury as potentially the result of victim SP
actions only if the injury was inflicted after SP actions.
The NCVS does not address the possibility of complex sequences in which multiple
different types of defensive actions are taken, and injury occurs after one victim action but before
another type of action. Rather, all victims who were injured and used protective actions are simply
7
coded as to whether protective actions (in general) were taken before, during, or after suffering
injury. We treated victims who were injured after victim actions as having suffered post-self-
protection (“post-SP”) injury. In some incidents, victims were unable to say whether their protective
actions came before or after injury. We treated these incidents as missing on the post-SP injury
variables. Interviewers could code incidents for as many of these sequences as were appropriate,
and a victim therefore could be coded as having suffered injury before, during, and after taking
defensive actions. For the purpose of coding post-SP injury, we treated victims who were injured
both before and after victim actions, or both during and after SP actions, as having suffered post-SP
injury, thereby intentionally favoring the hypothesis that resistance increases the victim’s risk of
injury.
Self-Protection Variables Measured
The independent variables of primary interest were 16 binary variables denoting whether the
victim took a given type of SP action (1=action was taken, 0=action was not taken). The specific
actions recorded are listed in Table 1. Interviewers recorded as many or few of these strategies as
victims reported, so it is possible for any one victim to be coded 1 on multiple SP variables.
Victims who did nothing to resist would simply be coded 0 on all 16 SP variables. Because there
was no variable included in the models that explicitly denoted that victims did nothing to protect
themselves, “no self-protection” was the omitted SP category, which therefore serves as the point of
comparison for all specific protective actions. Thus the coefficient of each SP variable reflects how
much more or less likely a given outcome was for victims who took that action, compared to
victims who did nothing to resist, other things being equal.
Control Variables
Other independent variables included in the models measure characteristics of the victims,
offenders, and circumstances that might influence the outcomes of the incidences, and that might
also be correlated with the willingness or ability of victims to use each defensive action. Of
8
particular interest, three variables were included to reflect power advantages that offenders had over
victims. ADVAGEOF is coded higher when one or more offenders were in their physical prime ages
(age 15-29) and the victim was not of this age range, i.e. there was likely to be a power advantage to
the offender(s) based on age and associated physical fitness. ADVNUM equaled the number of
offenders minus the number of victims, reflecting any numerical advantage of offenders.
MALEOFDC is coded higher when one or more offenders are male, who are stronger than female
victims. Other variables measure whether offenders possessed weapons during the incident
(OHADGUN, OHADKNIF, OHADSHAP) and whether offenders actually attacked victim
(OFDATCK). A more complete rationale for the inclusion of each control variable may be found in
Tark & Kleck (2004).
To summarize, our study makes a contribution to the literature on self-protection in sexual
assault in the following ways: (1) we examine a large national probability sample of sexual assaults
rather than a small nonprobability sample, (2) we take account of the sequence of victim protective
actions and injury in appropriate ways, (3) we control for many confounding correlates of defensive
actions, (4) we separately assess the impact on the outcomes of crimes of each of the 16 specific
victim actions coded in the post-1992 NCVS, (5) we assess the impact of self-protective actions on
both rape completion and on whether the victim suffers additional injuries. To our knowledge, this
study is the only one to combine these strengths. The few prior studies that took account of the
sequence of self-protective actions and injury (e.g., Quinsey & Upfold, 1985; Ullman, 1998) were
nearly all based on nonprobability local samples of offenses reported to the police, while almost all
of those that analyzed national probability samples did not take account of sequence (e.g. Block &
Skogan, 1986; Brecklin & Ullman, 2001; Lizotte, 1986; Marchbanks et al., 1990). The only prior
study based on a large national sample that took account of sequence was that of Tark and Kleck
(2004), but this study assessed only effects on injury, and did not address rape completion.
9
Results
Frequency, Completion, And Injury Rates Of Protective Actions
Table 2 shows how often NCVS crime victims reported using the various types of victim
protective actions, the share of victims using each method who experienced a completed rape (vs.
an attempted rape), and the share who suffered any additional injury other than rape itself, and rates
of injury that that occurred after the victim took SP actions. The figures show that while many
victims of rape or attempted rape suffer nonrape injuries, few of these injuries were inflicted after
the victim took protective actions. Resisting victims are less likely to be injured after taking some
kind of protective action than non-resisting victims. In 556 rape/attempted rape incidents where
victims resisted in some way, 54% of the rape attempts were completed, but only 19% of rape
attempts were completed after the victim took SP actions; 26% involved the victim suffering some
nonsexual injury after taking SP actions, and 5% involved the victim suffering a more serious
nonsexual injury. In contrast, among the 177 incidents involving victims who did not resist, 88% of
incidents resulted in rape completion, 25% of such incidents resulted in a nonsexual injury, and
2.8% resulted in serious nonsexual injury. Overall, victim SP during rape attempts was associated
with significantly lower risks of rape completion and roughly the same risk of serious nonsexual
injuries as compared to taking no SP actions. These figures suggest that resistance during rape
attempts could have provoked offenders to inflict additional serious injuries on victims in no more
than 2.4% of all rape incidents - the 5.2% post-SP serious injury rate among resisting victims, minus
the 2.8% “baseline” serious injury rate that prevailed even among nonresisting victims.
There were only nine sample cases of rape victims resisting with weapons, only one of
which involved use of a gun. Nevertheless, in these nine cases, none of the victims suffered a rape
completion after resisting with a weapon, and none suffered any injury, serious or not, after doing
so. Among who victims did suffer some additional physical injury, all were injured before taking
self-protective actions with weapons.
10
(Table 2 about here)
Similar patterns were found among 1,278 nonrape sexual assault. Among incidents
involving victims who resisted, only 11.7% of the victims were non-sexually injured after resisting,
and only 2.2% were seriously injured after resisting. Among non-resisting victims, the rate of non-
rape injury was higher. In the 265 incidents with non-resisting victims, 19.2% resulted in victim
injury, and 2.3% resulted in serious victim injury. If we view these injury rates among nonresisting
victims as indications of the “baseline” level of danger prevailing in assaults even without any
victim defensive actions, resistance does not appear to add to this level of risk of injury. Of the 17
sample cases of armed resistance, none resulted in injury after the victim used a weapon.
In sum, resistance, including armed resistance, rarely is followed by the offender inflicting
further injury on the victim, and almost never by the infliction of serious injury. A wide variety of
victim resistance strategies appear to be free of any substantial risk of provoking aggressors into
inflicting further injury on the victim. These conclusions can be drawn even before performing
complex multivariate tests because even if one were to make the extreme assumption that in all
cases of post-SP injury, victim resistance alone caused the offender to hurt the victim, it would still
be valid to conclude that resistance rarely causes the victim to suffer further injury. In reality, it is
unlikely that all crime victims who resisted and then were injured suffered those injuries because
they resisted, since some offenders were probably going to hurt their victims regardless of whether
the victims resisted. Thus, the post-SP injury percentages are properly viewed as upper limits on
the share of crimes in which protective actions provoked offenders into injuring the victim.
These simple injury rates, however, cannot tell us whether resistance actually reduces risk of
injury. Perhaps victims resist only in situations that were already relatively safe, or only resist
offenders who appeared unlikely to hurt them. The figures likewise cannot tell us which protective
actions are relatively more effective or harmful because they do not reflect controls for other factors
influencing assault outcomes. Therefore multivariate controls are needed.
11
Multivariate Results
Rape Completion
Table 3 displays findings from the logistic regression analysis assessing the impact of each
type of SP action on whether rapes are completed versus being unsuccessful rape attempts,
controlling for a extensive set of other possible determinants of rape completion. The first column
presents findings based on an analysis intentionally designed to have the same flaws as most past
research, in that they show the association between protective actions and rape completion without
respect to whether rape completion preceded or followed resistance. Odds ratios below one for an
SP variable indicate that the indicated resistance tactic is associated with a lower likelihood of the
rape attempt being completed, compared to nonresistance, while odds ratios over one indicate that
the tactic is associated with a higher likelihood of rape completion. The number shown in
parentheses below each odds ratio is the ratio of the estimated logistic regression coefficient over its
estimated standard error.
(Table 3 about here)
In this analysis, most forms of resistance seem to reduce the risk of rape completion, though
the association is sometimes not statistically significant. We cannot, however, be sure that
resistance prevents completion if we do not know the temporal order of the two. Perhaps there is a
negative association merely because once the rape is completed, resistance seems to the victim to be
pointless. Or perhaps the appearance, early in the encounter, that completion is inevitable
demoralizes the victim and discourages resistance.
The second column of Table 3 presents findings from an analysis that addresses the problem
of sequence. Here the dependent variable denotes whether the rape was completed after the victim
took protective actions. Victims were coded “1” if they took SP actions and were raped after doing
so, and were coded “0” if they took SP actions and were not raped after doing so, the latter group
including incidents in which rape was completed before the victim took any SP actions. The results
12
in the second column are based on a sample composed only of victims who took some kind of
protective action, since the concept of rape completion happening after SP actions does not apply to
incidents in which no SP actions were taken.
This analysis addresses the question: “Among rape victims who did something for self-
protection, which actions were relatively more effective in averting completion of the rape
attempt?” Only relative effects are estimated in this analysis because estimating “absolute”
effectiveness would require a comparison with incidents in which there were no SP actions of any
kind taken by the victim. We selected “called the police” as the omitted category because it is
sometimes presented as the officially recommend course of action for victims, and thus can serve as
a useful point of comparison. The choice of an excluded category, however, has no effect on
estimates of the relative effectiveness of different SP actions. It is important, however, to
understand that a given SP action may be quite effective even if its coefficient in this analysis was
not significantly different from zero, if one regards calling the police as an effective action.
In this analysis, the effectiveness of most SP actions did not significantly differ from calling
the police, suggesting that all SP actions are roughly equally effective in preventing rape
completion. Resistance in general appears to be effective in preventing the completion of rape
attempts, in light of the Table 2 finding that only 19.1% of attempts were completed after the victim
somehow resisted, compared to an 88.1% completion rate in attempts in which the victim did not
resist in any way. There were three exceptions to this generalization. First, the only SP tactic that
significantly increased the likelihood of rape completion (relative to calling the police) was
‘arguing, reasoning or pleading” with the rapist – this strategy increases the odds of completion by a
factor of 4.5. This is noteworthy because this is the second-most common type of SP action taken
by rape victims (first column, Table 2). Second, cooperating with the rapist, or pretending to do so,
has a similarly large effect increasing the likelihood of completion, though the association is not
quite significant at the .05 level. Third, “trying to attract attention or help/cried out for help”
13
appears to significantly reduce the risk of rape completion below that associated with calling the
police. “Screaming from pain or fear” also appears to be less effective than calling the police,
although the difference was not statistically significant. Results concerning this self-protective
strategy, however, are ambiguous because it is possible that this behavior is the result of the rape
completion itself, or (when it precedes completion) the anticipation of the rape completion.
Finally, we analyzed post-SP rape completion using a sample that included incidents in
which the victim did not take any SP actions. In this analysis, the outcome variable, post-SP rape,
would be coded “1” if either (a) the victim took some SP actions and rape was completed
afterwards, or (b) took no SP and rape was completed. It was coded “0” if (a) the victim took SP
actions and the rape attempt was not completed, (b) took SP action and the attempt was completed,
but before SP actions, or (c) took no SP action and rape was not completed. Cases in which the
victim reported that SP actions and injury occurred simultaneously were treated as missing, since it
was impossible to establish the SP-injury sequence in these incidents. Since “no-SP” was treated as
the excluded SP category in this analysis, the odds ratio for SP variables can be interpreted as
reflecting a comparison between each specific SP tactic and taking no SP actions at all. This
analysis therefore provides estimates of the “absolute” effectiveness of SP strategies as compared to
nonresistance.
The third column in Table 3 presents the results of this analysis. The odds ratios are directly
comparable to those of the first column. This comparison directly establishes the effects of taking
account of the sequence of injury and SP actions, since this is the only difference between the
estimates reported in the first column and those reported in the third column. Most SP actions are
associated with a lower risk of rape completion as compared to nonresistance, many of them
significantly so. Leaving aside the ambiguous findings concerning “screaming from pain or fear,”
no form of victim resistance was associated with a significantly higher risk of rape completion than
nonresistance. The findings support prior research that found that many SP actions were effective
14
in reducing the risk of rape completion. The most effective tactics for avoiding completion were (1)
running or driving away/hiding/locking door, (2) attracting attention/calling for help, (3) physically
struggling, (4) an unspecified miscellany of “other actions,” and (5) unarmed attacks on the rapist.
These SP actions appear to decrease the risk of rape completion about 80-86%, compared to
nonresistance.
There are no estimates of the effects of victims using guns to attack or threaten offenders,
because the NCVS sample included only a single case of a rape victim using a gun. (For similar
reasons, there are no estimates of the effect of chasing or trying to hold the offender on post-SP rape
completion.) Likewise, odds ratios for the use of other (nongun) weapons are uninterpretable and
highly unstable because the odds are based on tiny numbers of incidents in which these tactics were
used and the complete absence of any sample cases in which rape completion followed their use
(see Table 2, “Rape” panel, column labeled “% Completed After SP”). There were no sample cases
of rape attempts being completed after victims used weapons to resist. Thus, although it is
impossible to estimate meaningful multivariate logistic regression coefficients and odds ratios for
the armed resistance variables, the complete absence of rape completion following use of this tactic
certainly suggests that it is effective.
Physical Injury Other than Rape
Some scholars have argued that even though victim resistance may reduce the risk of rape
completion, it increases the risk of other injuries by angering the rapist into further attacks. Table 4
presents findings from analyses of the impact of each SP action on whether the offender inflicted
non-sexual injuries, i.e. those other than rape, attempted rape, or the verbal threat of rape. The
results presented in the first column show associations between SP actions and non-sexual injury in
rape incidents, without respect to whether injury preceded or followed resistance. These results
seem, at first glance, to support the idea that some SP actions increase the risk of physical injury.
“Attacking without weapons,” “struggling” and “screaming from pain/fear” are all significantly
15
associated with higher injury rates compared to nonresistance. The results, however, are misleading
because they do not take into account the temporal sequence between SP action and injury. The
positive associations might reflect victims being injured, and then taking SP actions they might
otherwise not have taken. These results are shown to illustrate the kinds of findings that were
obtained in past research that did nothing to take account of the sequence of victim self-protective
actions and injury.
The estimates in the second column take into account the temporal sequence, because the
dependent variable measures nonsexual injury inflicted after SP. That is, an incident is assigned the
higher code only if the victim first took some SP action and was then injured. Incidents in which
victims took no SP actions at all were excluded from this analysis, and the omitted reference
category is “calling the police.” The effects of most SP variables were not significantly different
from those of calling the police. Only “screaming from pain or fear” was associated with a
significantly higher risk of injury than calling the police. This result, however, is hard to interpret,
even though sequence was taken into account, since the positive association may be due to victims
screaming from fear shortly before the injury was inflicted, e.g. as a blow was about to fall. In such
cases, it would be more accurate to say that imminent injury caused victims to scream, rather than
the victim’s screaming provoking the offender into attacking. More generally, once sequence is
taken into account, the evidence does not support the claims that resistance increases the risk of
nonsexual injury, or that forceful resistance is more likely than non-forceful resistance to provoke
offenders into attacking and injuring rape victims.
It may currently be impossible to find statistically significant effects of many of the SP
actions on injury, because even in the largest available samples of rapes, sample sizes are relatively
small. Keeping this statistical limitation in mind, our findings indicate that victim resistance does
not significantly affect the risk of rape victims suffering additional injuries.
(Table 4 about here)
16
The middle panel of Table 4 displays estimates of models pertaining to injury in “sexual
assaults,” a category that includes victimizations of a sexual nature other than rapes or attempted
rapes. As with the rape analysis, the results indicate that SP tactics generally have no significant
association with injury, and the only SP action that appears to elevate the risk of injury is the victim
screaming from pain or fear – a result whose meaning is ambiguous. Besides this tactic, the only
tactic showing a near-significant injury-elevating effect was arguing, reasoning, or pleading with
the offender – a nonforceful tactic. Again, there is no support for the claim that forceful forms of
resistance are more dangerous than nonforceful forms.
Serious Injury
We conducted an additional analysis addressing the impact of SP actions on whether the
victim suffered more serious non-rape injury, defined as injury (other than rape itself) more serious
than cuts or bruises. In part because of the rarity of serious post-SP injury (Table 2, right-most
column of Rape and Nonrape Sexual Assault panels), no coefficient was significantly associated
with serious injury among either rape incidents or sexual assaults (results not shown). In the larger
sample of all assault incidents, many victim resistance actions were associated with a lower risk of
serious injury than nonresistance, as was found in the analysis of all injury.
An analysis of information provided in interviews with crime victims cannot take account of
the most serious possible injury, death. Tark and Kleck (2004) estimated that, at most, one in
4,208 rapes and sexual assaults results in the death of the victim. Thus, in our sample of 1,278 rapes
and sexual assaults, there probably would not have been a single murdered victim even if we had
used sampling methods that could capture murdered victims (e.g., using proxy reports from
relatives of victims). Our findings that resistance does not provoke the inflicting of nonfatal injury
provide strong reason to expect that it likewise does not provoke the inflicting of fatal injury.
Are the Effects of Protective Actions Contingent on Other Conditions?
Some scholars have suggested that the effectiveness of different defensive actions may
17
depend on a variety of conditions under which they are used. We examined whether the effects of
each SP action differ depending on (a) the victim-offender relationship, particularly, when the
victim and offender(s) are sexual intimates (e.g., Bachman et al., 2002; Ruback & Ivie, 1984), (b)
offender alcohol consumption, (c) location of the incident (at home or not), (d) time (day vs. night),
and (e) the number of offenders. In the post-SP rape completion models, we did not find that the
effectiveness of SP actions depend on these conditions. No more than one out 16 interaction
variables had a significant coefficient in any one model, and one would expect one coefficient to be
“significant” at the .05 level solely as a result of chance, due to the large number of hypothesis tests.
Further, the signs of the coefficients were as likely to be contrary to theoretical expectations as
consistent with them (results not shown). On the whole, the effects of victim actions on rape
completion or injury do not significantly vary depending on these conditions.
Discussion
A number of limitations of these data need to be made explicit. First, many rape
victimizations are not reported. While the problem is not as severe in victim survey samples as in
samples of crimes known to police, it is nevertheless well known that rape victims are often
reluctant to report victimizations, particularly when the offenders are intimates (Bachman, 1998).
The underreporting of sexual assaults aggravates the problem of small sample sizes and contributes
to making standard errors of coefficient estimates so large that fewer estimates of the effects of SP
actions could be significantly different from zero. Further, rape victims may be less likely to report
incidents involving certain kinds of defensive actions. For example, they may not want to report
use of weapons because weapon possession is often unlawful, especially in public places (Kleck &
Gertz, 1995). Victims may be reluctant to report unsuccessful SP actions because it makes them
look foolish or ineffectual, or may fail to report successful SP actions because they do not think
rape attempts that did not result in rape completion or other injury qualify as crimes or merit
reporting them to interviewers (Hindelang & Gottfredson, 1976; Felson, Messner & Hoskin, 1999).
18
Further, the NCVS does not measure various circumstantial factors that can influence victimization
outcomes, such as the victim’s alcohol consumption and the relative physical power of victims and
offenders. Finally, given the impossibility of experimental research on this topic, it should be noted
that our findings are necessarily based on observed associations between victim actions and assault
outcomes, thereby precluding definitive conclusions about causal effects.
None of the nine sample cases of armed resistance resulted in either rape completion or
other injury following the self-protective actions. Given the small number of relevant cases, this
finding must be regarded as merely suggestive, but it is consistent with prior research on rape and
resistance. Kleck and Sayles (1990, p. 155) found that armed resistance was the most effective
victim strategy, of six considered, in avoiding rape completion. Likewise, Guerette (2010, p. 216)
found armed resistance (“threatened or used object, knife, or gun”) to be the most effective tactic in
preventing rape completion. Finally, although Clay-Warner (2002, p. 698) did not separately assess
armed resistance, she found that the broader “physical protective actions” category that included
armed resistance was the most effective category of protective actions, of three considered, in
preventing rape completion.
To summarize, most SP actions, both forceful and non-forceful, either significantly reduce
the risk of rape completion or have no significant effects. In particular, actions such as attacking
without weapons, struggling, running away or hiding, and trying to attract attention or help appear
to reduce the risk of rape completion more than 80% compared to nonresistance. Further, although
there were too few sample cases for associations to be statistically significant, the few sample
instances of rape victims using weapons to resist appear to have been completely effective in
preventing rape completion and avoiding post-self-protection injury, given that none of the nine
sample cases resulted in a completed rape or post-SP injury. Most SP tactics do not significantly
affect the risk of additional physical injury one way or the other. More generally, the findings did
not support the argument that forceful SP actions are less effective or more risky for the victim than
19
nonforceful SP actions. Indeed, the only SP tactics associated (albeit usually nonsignificantly) with
an increased risk of rape completion were nonforceful tactics – cooperating or pretending to
cooperate with the offender, or arguing or pleading with the offender. Overall, our evidence
indicates that rape victims’ self-protection actions generally reduce the probability of rape
completion, without significantly raising the risk of additional injury.
We believe that these findings imply any police officers and rape victim support groups
who counsel against forceful resistance should reconsider these policies. The notion that resistance
increases the victim’s chances of suffering additional injury beyond the rape itself is not supported
by the evidence. Likewise, it is not helpful for those offering advice to prospective victims confine
their suggestions to how to avoid situations that might give rise to sexual assaults. While useful,
such advice does not help those who cannot, despite their best efforts, avoid such situations and are
confronted by a would-be assaulter. As Fisher, Daigle and Cullen concluded, coupling “learning
how to use effective self-protective action when threatened or assaulted” with education on
avoidance of risky situations “is critical to preventing rape” (2008, p. 171-172). In any given
threatening situation, some defensive tactics may be feasible while others are not. Therefore, self-
defense training might in future focus on providing prospective victims with knowledge of a wide
array of tactics, both forceful and nonforceful, to choose from. For their part, future researchers
might explore in greater detail which specific tactics have worked best in which kinds of situations.
20
References
Atkeson, B. M., Calhoun, K. S., & Morris, K. T. (1989). Victim resistance to rape. Archives of
Sexual Behavior, 18, 497-507.
Bachman, R. (1998). The factors related to rape reporting behavior and arrest: new evidence
from the National Crime Victimization Survey. Criminal Justice Behavior, 25, 8-29.
Bachman, R., Saltzman, L. E., Thompson, M. P., & Carmody, D.C. (2002). Disentangling the
effects of self-protective behaviors on the risk of injury in assaults against women.
Journal of Quantitative Criminology, 18, 135-57.
Bart, P. B. (1981). A study of women who both were raped and avoided rape. The Journal of
Social Issues, 37, 123-137.
Bart, P. B., & O'Brien, P. H. (1984). Stopping rape: effective avoidance strategies. Signs,10,
83-101.
Block, R., & Skogan, W. (1986). Resistance and nonfatal outcomes in stranger-to-stranger
predatory crime. Violence and Victims, 4, 241-253.
Brecklin, L.R., & Ullman, S.E. (2001). The role of offender alcohol use in rape attacks.” Journal of
Interpersonal Violence, 16, 3-21.
Burgess, A. W., & Holmstrom, L.L. (Ed.). (1974). Rape: Victims of Crisis, Brady, Bowie. MD:
Robert J. Brady Company.
Clay-Warner, J. (2002). Avoiding rape: the effect of protective actions and situational factors
on rape outcome. Violence and Victims, 17, 691-705.
Felson, R., Messner, S., & Hoskin, A. (1999). The victim-offender relationship and calling
the police in assaults. Criminology, 37, 931-947.
Fisher, B. S., Daigle, L. E., & Cullen, F. T. (2008). Rape against women:
what can research offer to guide the development of prevention programs and risk
reduction interventions? Journal of Contemporary Criminal Justice, 24, 163-177
21
Griffin, B. S., & Griffin, C. T. (1981). Victims in rape confrontation. Victimology,6, 59-75.
Guerette, R. T., & Santana, S. A. (2010). Explaining victim self-protective behavior
effects on crime incident outcomes: a test of opportunity theory. Crime & Delinquency,
56, 198-226.
Hindelang, M. J., & Gottfredson, M. (1976). The victim’s decision not to invoke the criminal justice
process in Criminal Justice and the Victim, (Ed. McDonald, W. F.). Beverly Hills: Sage.
1976.
Kilpatrick, D. G., Resick, P. A., & Veronen, L. J. (1981). Effects of a rape experience: a
longitudinal study. Journal of Social Issues, 37, 105-122.
Kleck, G. (1997). Targeting Guns: Firearms and their Control. Hawthorne, N.Y.: Aldine de
Gruyter.
Kleck, G., & Gertz, M. (1995). Armed resistance to crime: the prevalence and nature of self-
defense with a gun. Journal of Criminal Law and Criminology, 86, 150-187.
Kleck, G., & Gertz, M. (1998). Carrying guns for protection: results from the National Self-
Defense Survey. Journal of Research in Crime and Delinquency, 35, 193-224.
Kleck, G., & Sayles, S. (1990). Rape and resistance. Social Problems, 37, 149-162.
Koss, M. P. (1993). Detecting the scope of rape. Journal of Interpersonal Violence, 8, 198-222.
Levine-MacCombie, J., & Koss, M. P. (1986). Acquaintance rape: effective avoidance strategies.
Psychology of Women Quarterly, 10, 311-320.
Marchbanks, P. A., Lui, K., & Mercy, J. A. (1990). Risk of injury from resisting rape. American
Journal of Epidemiology, 132, 540-549.
Quinsey, V. L., & Upfold, D. (1985). Rape completion and victim injury as a function
of female resistance strategy. Canadian Journal of Behavioural Science, 17,
40-50.
Rennison, C. M. (2002). Rape and Sexual Assault: Reporting to Police and Medical Attention,
22
1992-2000, Bureau of Justice Statistics Selected Findings, Washington, DC: U.S.
Department of Justice, Office of Justice Programs, NCJ 194530.
Ruback, R. B,. & Ivie, D. L. (1988). Prior relationship, resistance, and injury in rapes: an analysis of
crisis center records. Violence and Victims, 3, 99-111.
Tark, J., & Kleck, G. (2004). Resisting crime: the effects of victim action on the
outcomes of crimes. Criminology, 42, 861-909.
Ullman, S. E. (1997). Review and critique of empirical studies of rape avoidance. Criminal
Justice and Behavior, 24, 177-204.
Ullman, S. E. (1998). Does offender violence escalate when rape victims fight back? The Journal
of Interpersonal Violence, 13,179-192.
Ullman, S. E. (2007). A 10-year update of ‘review and critique of empirical studies of rape
avoidance. Criminal Justice and Behavior, 34, 411-429.
Ullman, S. E., & Knight, R.A. (1992). Fighting back: women’s resistance to rape. Journal of
Interpersonal Violence, 7, 31-43.
Ullman, S. E., & Knight, R.A. (1993). The efficacy of women’s resistance strategies in rape
situations. Psychology of Women Quarterly, 17, 23-38.
United States Bureau of Justice Statistics. (1985). Reporting Crimes to the Police. Washington,
D.C.: U.S. Government Printing Office.
United States Bureau of Justice Statistics. (2000). Criminal Victimization in the United States, 1995.
Washington, D.C.: U.S. Government Printing Office.
Zoucha-Jensen, J. M., & Coyne, A. (1993). The effects of resistance strategies on rape. American
Journal of Public Health, 83, 1633-1634.
TABLE 1
Variables in the Analyses*
Variable Description Proportion
Dependent Variables
RAPED V was raped (completed rape) 0.62
POSTRAPE V was raped after taking self-protective actions 0.34
NOSEXINJ V was injured, excluding rape 0.37
PONOSEXI V was injured, excl. rape, after taking self-protective actions 0.26
NOSEXSR V was seriously injured, excluding rape 0.04
PONSEXSR V was seriously injured, excl. rape, after self-protective actions 0.05
Independent Variables
Victim’s Self Protection
GUNATACK V attacked O with gun; fired gun 0.00
GUNTHRET V threatened O with gun 0.00
NOGUNATK V attacked O with other weapons (knife, etc.) 0.00
NOGUNTHR V threatened O with other weapon (knife, etc.) 0.00
NOWEPATK V attacked O without weapon (hit, kicked, etc.) 0.14
NOWEPTHR V threatened without weapon 0.01
STRUGGLE V struggled, ducked, blocked blows, held onto property 0.38
CHASHELD V chased, tried to catch or hold O 0.00
SCAREOFF V yelled at O, turned on lights, threatened to call police 0.18
COPRSTAL V cooperated, or pretended to (stalled, did what they asked) 0.08
ARGUE V argued, reasoned, pleaded, bargained, etc. 0.22
RANHIDE V ran or drove away, or tried; hid, locked door 0.12
CALLPOL V called police or guard 0.04
GETHELP V tried to attract attention or help, warn others (cried out for
help, called children inside)
0.04
SCREAM V screamed from pain or fear 0.12
OTHERS V took other SP actions 0.10
Power Difference between V and O
ADVAGEOF O age 15-29 and V either under 15 or 30 or older 0.09
ADVNUM Number of O – number of V (raw number) -.08
MALEOFDC O was male 0.98
Offender Weapons and Attack
OHADGUN O had gun 0.05
OHADKNIF O had knife 0.06
OHADSHAP O had sharp object 0.00
OFDATCK O attacked V 0.98
24
Note.* For binary variables, 1=attribute is present, 0=attribute is not present
25
TABLE 1
Variables in the Analyses (Continued)
Variable Description Proportion
Victim Characteristics
HADCHILD Child in the victim’s household 0.41
HOUSOWN V owned the house 0.34
EMPLOYED V had a job last week or for 2 weeks in last 6 months 0.60
OLD65 V was 65 or older 0.01
MARRIED V was married 0.09
HIGHDIPL V had high school diploma or higher 0.38
BLACK V was black 0.19
ASIAN V was Asian 0.02
HISPANIC V was Hispanic origin 0.08
NUMVICEX Number of victimization in last 6 months (raw number) 2.27
Offender Characteristics
OFDGANG O was gang member 0.06
OFDSUBST O was under influence of alcohol or drugs 0.48
OFDFAMIL O was V’s family member 0.03
OSEXINTI O was V’s sexual intimate 0.28
OFDACQNT O was V’s acquaintance (not family or work acquaintance) 0.26
OWORKACQ O was V’s work acquaintance 0.03
OFDBLACK O was Black 0.26
OFDWHITE O was White 0.63
Incident Circumstances
RURAL Incident occurred in rural area 0.20
URBAN Incident occurred in urban area 0.41
ATHOME Incident occurred at home 0.44
NEARHOME Incident occurred near home 0.31
SECUPUB Incident occurred in public place which might have security 0.06
OTHRPRES Incident occurred with third parties present 0.20
Other Variables Eliminated in Logistic Analyses
ANYSD16 V took any of 16 types of self-protective action 0.76
TOTALSD Total number of victim actions 0.33
HOMINCOM Income of the household (categorical variable) 6.61
YOUG1529 Victim was 15 to 29 years old 0.62
NUMOFD Number of Offenders (raw number) 0.13
YONGOFDC Offender was 15 to 29 yr old 0.57
NIGHT Incident occurred at night 0.72
AFTERNON Incident occurred in the afternoon 0.09
SOUTH Incident occurred in South 0.25
WEST Incident occurred in West 0.18
26
27
Table 2 Frequency, Rape Completion Rates, and Injury Rates of Self-Protection (SP) Strategies
SP Strategy
Rapes
Frequency % Rapes
Completed
% Com-
pleted
After SP
%
Injured**
% Injured
After SP**
%
Seriously
Injured**
%
Seriously
Injured
After SP**
V attacked O with gun; fired gun 0 - - - - - -
V threatened O with gun 1 100.0 0.0 0.0 0.0 0.0 0.0
V attacked O with other weapons (knife, etc.) 4 25.0 0.0 100.0 0.0 0.0 0.0
V threatened O with other weapon (knife, etc.) 4 25.0 0.0 25.0 0.0 25.0 0.0
V attacked O without weapon (hit, kicked, etc.) 100 49.0 23.8 49.5 33.3 6.0 6.8
V threatened without weapon 7 57.1 33.3 42.9 25.0 25.0 25.0
V struggled, ducked, blocked blows 279 49.8 17.5 45.5 28.1 6.1 6.8
V chased, tried to catch or hold O 2 50.0 0.0 0.0 0.0 0.0 -
V yelled, turned on lights, threatened to call police 135 45.2 16.7 50.4 43.3 8.1 10.6
V cooperated, or pretended
To (stalled, did what they asked) 56 82.1 37.5 39.3 53.8 8.8 27.3
V argued, reasoned, pleaded, bargained, etc. 162 69.8 38.5 44.4 44.6 5.6 16.7
V ran or drove away, or tried; hid, locked door 89 34.8 0.0 38.9 22.0 2.2 0.0
V called police or guard 29 41.4 0.0 62.1 35.7 6.9 0.0
V tried to attract attention or help, warn others
(cried out for help, called children inside) 31 45.2 11.1 67.7 68.8 6.5 20.0
V screamed from pain or fear 90 66.7 22.2 64.0 62.5 10.0 19.0
V took other SP actions 71 63.4 10.0 29.6 9.1 4.2 0.0
Any self-protection action taken 556 54.5 19.1 40.8 26.3 4.3 5.2
No self-protection actions taken 177 88.1 - 24.9 - 2.8 -
Total Incidents* 733 62.9 11.0 36.9 8.0 4.0 1.4
28
Table 2 (Continued) Frequency and Injury Rates of Self-Protection (SP) Strategies
Type of SP
Sexual Assaults**
Fre-
quency
%
Injured
%
Injured
After
SP
% Ser-
iously
Injured
% Ser-
iously
Injured
After
SP
V attacked O with gun; fired gun 0 - - - -
V threatened O with gun 2 0.0 0.0 0.0 0.0
V attacked O with other weapons (knife, etc.) 4 100.0 0.0 0.0 0.0
V threatened O with other weapon (knife, etc.) 11 9.1 0.0 9.1 0.0
V attacked O without weapon (hit, kicked, etc.) 144 36.8 19.8 4.1 3.6
V threatened without weapon 12 33.3 16.7 15.4 16.7
V struggled, ducked, blocked blows, held onto
property 400 37.5 18.8 4.5 3.6
V chased, tried to catch or hold O 5 0.0 0.0 0.0 0.0
V yelled at O, turned on lights, threatened to call
police 251 33.9 19.2 4.4 4.3
V cooperated, or pretended
to (stalled, did what they asked) 68 39.7 54.5 8.7 23.5
V argued, reasoned, pleaded, bargained, etc. 234 37.2 27.7 4.7 10.4
V ran or drove away, or tried; hid, locked door 198 22.7 8.3 1.0 0.0
V called police or guard 58 37.9 12.8 3.4 0.0
V tried to attract attention or help, warn others (cried
out for help, called children inside) 53 49.1 38.7 3.8 8.3
V screamed from pain or fear 105 59.0 52.4 9.4 16.7
V took other SP actions 189 13.8 1.5 1.6 0.0
Any self-protection action taken 1,013 26.8 11.7 2.7 2.2
No self-protection actions taken 265 19.2 - 2.3 -
Total Incidents* 1,278 25.2 5.6 2.5 1.0
Note. * Total Incidents are smaller than the sum of separate counts of SP actions because some victims employed multiple SP actions.
29
Table 3 Self-Protection Effects on Rape Completion
Note. Bold: p<.01 (two-tailed), Italic: 01<p<.05 (two-tailed)
* Not including No-SP group; “Calling the Police” is the omitted category.
** Including No-SP group; “No Self-protection” is the omitted category.
Variable Odds Ratio (Coef./S.E.)
Rape
Completed
Rape
Completed
After SP*
Rape
Completed
After SP**
Victim’s Self Protection
V attacked O with gun; fired gun - - -
V threatened O with gun ---
V attacked O with other weapons (knife, etc.) 0.00
(0.00)
17.70
(0.00)
0.00
(0.00)
V threatened O with other weapon (knife, etc.) 0.23
(-0.98)
29.17
(0.00)
5.31
(0.00)
V attacked O without weapon (hit, kicked, etc.) 0.35
(-3.83)
1.82
(0.99)
0.20
(-3.70)
V threatened without weapon 2.47
(0.73)
4*108
(0.00)
2*109
(0.00)
V struggled, ducked, blocked blows 0.38
(-4.72)
1.11
(0.20)
0.15
(-5.31)
V chased, tried to catch or hold O 0.00
(0.00) - -
V yelled at O, turned on lights, threatened to call police 0.40
(-3.45)
0.72
(-0.57)
0.40
(-1.82)
V cooperated, or pretended to (stalled, did what they
asked)
2.96
(2.42)
4.45
(1.74)
1.41
(0.46)
V argued, reasoned, pleaded, bargained, etc. 1.27
(0.93)
4.53
(2.88)
1.27
(0.55)
V ran or drove away, or tried to; hid, locked door 0.28
(-4.27)
0.39
(-1.30)
0.13
(-3.66)
V called police or guard 0.57
(-1.02) -
0.84
(-0.17)
V tried to attract attention or help, warn others (cried out
for help, called children inside)
0.47
(-1.51)
0.07
(-2.14)
0.14
(-2.07)
V screamed from pain or fear 1.96
(1.98)
4.02
(1.87)
4.61
(2.35)
V took other SP actions 1.05
(0.14)
1.62
(0.59)
0.19
(-2.44)
Power Difference between V and O
O age 15-29 and V either under 15 or 30 or older 1.47
(1.14)
1.94
(0.94)
1.12
(0.23)
Number of O – number of V 1.43
(2.06)
1.70
(1.27)
1.62
(1.70)
O was male, V was female 1.71
(0.55)
2*108
(0.00)
1.35
(0.24)
Offender Weapons and Attack
O had gun
1.04
(0.07)
2.55
(1.00)
1.18
(0.26)
O had knife
1.05
(0.10)
14.76
(2.02)
2.43
(1.10)
O had sharp object
1.09
(0.05)
0.00
(0.00)
0.00
(0.00)
O attacked V 3*109
(0.00)
1*109
(0.00)
5*109
(0.00)
30
Table 3 (continued)
Variable
Rape
Completed
Rape
Completed
After SP*
Rape
Completed
After SP**
Victim Characteristics
Child in the victim’s household 1.04
(0.18)
2.03
(1.35)
1.55
(1.32)
V owned the house 0.79
(-1.11)
0.20
(-2.86)
0.32
(-3.18)
V had a job last week or for 2 weeks last 6 months 0.60
(-2.29)
0.44
(-1.49)
0.61
(-1.43)
V was 65 or older 0.07
(-2.53) -
0.09
(-1.83)
V was married 1.34
(0.80)
3.40
(1.40)
1.46
(0.66)
V had high school diploma or higher 1.24
(0.99)
1.89
(1.32)
1.22
(0.57)
V was black 0.53
(-1.69)
0.09
(-2.28)
0.23
(-2.20)
V was Asian 3.12
(1.39)
5*109
(0.00)
1*109
(0.00)
V was Hispanic origin 0.75
(-0.77)
1.22
(0.28)
0.81
(-0.40)
Number of victimizations in last six months 1.02
(0.79)
0.86
(-1.22)
1.01
(0.60)
Offender Characteristics
O was gang member 0.73
(-0.74)
6.18
(1.44)
1.88
(0.90)
O was on substance (alcohol or drugs) 0.67
(-2.01)
0.67
(-0.84)
0.69
(-1.12)
O was V’ sexual intimate 1.52
(1.43)
1.34
(0.37)
1.67
(1.14)
O was V’s family member 0.74
(-056)
3.99
(1.20)
0.98
(-0.02)
O was V’s acquaintance (no family, work acquaint.) 0.94
(-0.26)
1.57
(0.81)
0.94
(-0.18)
O was V’s work acquaintance 0.80
(-0.40)
0.55
(-0.34)
2.83
(1.06)
O was black 2.39
(2.05)
2.02
(0.67)
1.67
(0.71)
O was white 1.89
(1.86)
0.86
(-0.21)
0.87
(-0.27)
O was repeat O 0.95
(-0.18)
0.42
(-1.15)
0.59
(-1.33)
Incident Circumstances
Incident occurred in rural area 2.00
(2.36)
0.61
(-0.75)
0.83
(-0.41)
Incident occurred in urban area 0.95
(-0.23)
0.57
(-1.04)
0.45
(-2.14)
Incident occurred at home 2.03
(2.37)
6.25
(2.62)
3.62
(2.72)
Incident occurred near home 1.89 4.48 2.53
Table 4. Self
Protection Effects
on Injury Other
than Rape
Odds Ratio (Coef./S.E.)
Self Protection Strategy Rape Sexual Assault
Injury Injury
After
SP2
Injury
After
SP3
Injury Injury
After
SP2
Injury
After
SP3
V attacked O with gun; fired gun - - - - - -
V threatened O with gun - - - 0.00
(0.00)
0.00
(0.00)
0.00
(0.00)
V attacked O with other weapons (knife, etc.) 6*109
(0.00)
0.00
(0.00)
0.00
(0.00)
7*109
(0.00)
0.00
(0.00)
0.00
(0.00)
V threatened O with other weapon (knife, etc.) 0.24
(-0.86)
174.92
(0.00)
6.11
(0.00)
0.18
(-1.15)
0.00
(0.00)
0.00
(0.00)
V attacked O without weapon (hit, kicked, etc.) 1.77
(2.16)
2.95
(1.68)
1.85
(1.33)
1.46
(1.69)
1.89
(1.42)
1.27
(0.67)
V threatened without weapon 0.49
(-0.67)
0.04
(-0.64)
0.15
(-0.69)
0.52
(-0.71)
0.11
(-0.53)
0.29
(-0.51)
V struggled, ducked, blocked blows 1.86
(3.19)
1.69
(0.99)
1.05
(0.14)
1.97
(4.10)
2.36
(2.22)
1.12
(0.39)
V chased, tried to catch or hold O 0.00
(0.00)
- - 0.00
(0.00)
0.00
(0.00)
0.00
(0.00)
V yelled at O, turned on lights, threatened to call police 1.22
(0.80)
2.34
(1.54)
1.58
(1.02)
0.92
(-0.43)
1.35
(0.76)
0.96
(-0.13)
V cooperated, or pretended to (stalled, did what they
asked)
0.63
(-1.32)
3.79
(1.27)
0.97
(-0.04)
0.98
(-0.06)
5.46
(2.30)
2.38
(1.42)
V argued, reasoned, pleaded, bargained, etc. 1.21
(0.82)
2.07
(1.28)
1.63
(1.13)
1.38
(1.66)
2.80
(2.52)
1.78
(1.74)
V ran or drove away, or tried; hid, locked door 1.16
(0.52)
1.16
(0.24)
0.78
(-0.47)
0.92
(-0.37)
1.06
(0.13)
0.55
(-1.42)
V called police or guard 2.62
(1.82)
- 0.31
(-1.26)
3.15
(2.60)
- 0.57
(-.68)
V tried to attract attention or help, warn others (cried out
for help, called children inside)
1.27
(0.51)
3.22
(1.31)
2.72
(1.37)
1.27
(0.62)
1.97
(0.95)
1.98
(1.17)
V screamed from pain or fear 3.31
(3.98)
5.31
(2.16)
6.29
(3.01)
3.40
(4.56)
4.41
(2.66)
5.62
(3.55)
V took other SP actions 0.90
(-0.33)
0.86
(-0.13)
0.40
(-0.94)
0.5
(-2.33)
0.24
(-1.60)
0.09
(-2.90)
Sample Size 673 201 351 1,178 569 798
-2 Log Likelihood 747 139 314 1,054 254 476
31
32
Note. Bold p<0.01 (two-tailed), Italic 0.01<0.05 (two-tailed)
1. Coefficient and standard error estimates for variables other than self-protective actions are not shown, to save space.
2. Not including No-SP group; “calling the police” is the omitted category. 3. Including “no-SP incidents; “no self-protection” is the omitted category
Biographical Statement
Jongyeon Tark is Associate Professor of Criminology at Hannam University in South Korea. He received his doctorate in
criminology from the Florida State in 2005. His research has focused on the topics of victimization, criminology and law enforcement. His
articles have appeared in Criminology, Journal of Criminal Justice and the Journal of the Korean Association of Police Study.
Gary Kleck is Professor of Criminology and Criminal Justice at Florida State University. He received his doctorate in Sociology
from the University of Illinois in 1979. He is the author of Point Blank: Guns and Violence in America, which won the 1993 Michael J.
Hindelang Award of the American Society of Criminology, Targeting Guns (1997) and, with Don B. Kates, Jr., The Great American Gun
Debate (1997) and Armed (2001).
