To Resist or Not to Resist?

Abstract

Circumstances under which a sexual assault takes place and how these circumstances affect offenders’ reactions to victim resistance are not well understood. Previous studies have not thoroughly examined the interactions that take place between situational factors and resistance. Using a combination of logistic regression and Chi-square Automatic Interaction Detection (CHAID) analyses, we examine victim, situational, and crime characteristics of 426 sexual assaults involving victim resistance to determine which of these factors increase the likelihood of offender violence. Findings suggest that violence is affected by both the attack strategy employed by the offender and the type of resistance by the victim, along with several other qualifying factors.

Keywords

victim resistance criminal event sex offenders CHAID offending process

Introduction

Sexual assault is especially problematic for society, if not in the prevalence itself, at least insofar as the detrimental effects on the victims, and the societal view of such offenders is anything but favorable (Quinn, Forsyth, & Mullen-Quinn, 2004). Although a great deal of research has been conducted on various levels of these crimes, more research is needed with respect to the perspective of the offender and what factors are associated with the decision to incorporate greater degrees of violence in a response to victim resistance. The current study examines this particular aspect of sexual assault (the assault itself at a level that resulted in a federal prison term for the offender). Particularly, the question that is addressed asks under what scenarios offenders increase their level of violence on encountering passive, verbal, or physical resistance from their victim.

From the outset, it should be noted that, although this study examines the effects on the offender’s perpetration of violence within the assault, this is additive to injuries that occur due to an attempted or completed rape in and of itself. There are not only physical effects of attempted or successful penetration during an attack, but many psychological ramifications as well, which may be related to the degree of victim resistance (Rozee & Koss, 2001). These injuries are not the focus of this study, but they cannot be ignored within a discussion of the effects of victim resistance. Thus, any findings that arise from the analyses herein should be weighed against previous literature that has specifically examined the injurious effects of rape and how these relate to the level of resistance employed by the victim during an attack. Only when all of the information is combined will a lucid and useful picture of the degree of violence incurred during sexual assaults begin to emerge.

Victim–Offender Interactions

In the course of any personal crime, the behavior of one actor is shaped by the behavior of the other (Luckenbill, 1977; Tedeschi & Felson, 1994). Within the context of a sexual crime, this could mean that the victim’s behavior depends—in whole or in part—on the offender’s level of violence and coercion. Similarly, the offender may change his behavior depending on the victim’s perceived willingness or resistance.

Tedeschi and Felson (1994) further developed this calculative aspect through their decision-making theory of coercive action, wherein harmdoing is goal-oriented behavior arising from social interactionist processes. Borrowing from the rational choice perspective, choices regarding one’s actions are made based on the perceived value of rewards (positive outcomes), the perceived value of costs (negative outcomes), and the estimated probabilities or likelihoods of the positive and negative outcomes achieving fruition.

Luckenbill (1977) discusses violence in terms of a “working agreement” between offender and victim. Essentially, he suggests that each participant develops a role within the criminal interchange; each role is shaped by the other player and ultimately contributes to the resulting violent outcome. Because the victim’s actions antagonize the offender in some way—even if this is completely unintentional, as in the case of a child victim who refuses to stop crying—this is perceived by the offender as entering into an agreement where violence and force are acceptable tools that may be utilized to settle the dispute (Luckenbill, 1977). Thus, the escalation of offender violence depends on some antagonizing action made by the victim and is simply a reaction to this “provocation.”

Resistance by the victim would be interpreted by Luckenbill (1977) as such an act of provocation.¹ Although resistance is a natural reaction under certain abusive conditions, this may be viewed by the offender as a valid indication of agreement to the use of violence within the interpersonal exchange. Although the victim is not actually necessarily agreeing to the use of violence (especially in sexual assault cases, where the victim is often simply attempting to escape or stop the assault, as opposed to punishing the offender), the offender may perceive victim resistance as an invitation for further violence and coercion.

The factors that influence a victim’s decision making with regard to resistance affects the entire coercive interchange, as offender actions are often based on the actions of others in the interaction (Luckenbill, 1977). Thus, an awareness of factors that increase the magnitude of victim resistance is crucial to the understanding of the criminal event. However, Luckenbill’s focus on the target’s decision making eludes the factors relevant to the offender’s decision making. Whether or not an offender views coercion as a warranted or necessitated response to victim resistance or perceived antagonization is an important aspect of an abusive interchange. Thus, this study proposes to examine the offender’s point of view to determine when the victim’s actions were perceived as resistance and the response to that resistance.

Victim Resistance in the Criminal Event

As supported previously, an essential situational component of sexual crime is the spectrum of possible reactions from the victim, ranging from resistance to capitulation. We are especially interested in how this victim reaction influences the subsequent coerciveness used by the offender. Research has shown that the level of victim resistance is specifically influenced by, among other factors, time of day, the presence of a weapon, and the presence of alcohol (Clay-Warner, 2003). In reviewing the literature on rape avoidance, Ullman (2007) found that rape completion is related to the amount and immediacy of resistance, with more forceful and more immediate resistance increasing rape avoidance. Furthermore, the level and type of resistance has been shown to match the offender’s initial level of violence during the assault (Ullman, 2007; Ullman & Knight, 1992).

However, other situational variables appear to be important in their effects on offender coercion as well. Two situational variables frequently reported to affect the level of violence in a sexual assault are the presence of a weapon and the use of alcohol and/or drugs. Overall, the presence of a weapon has been found to significantly increase victim injury and violence levels, as well as the proportion of completed rapes (Coker, Walls, & Johnson, 1998; Ullman, 2007; Weaver et al., 2004). The use of alcohol or drugs has been reported to increase the level of violence and likelihood of rape completion, effects potentially due to reduced victim resistance (Carr & VanDeusen, 2004; Coker et al., 1998; Scott & Beaman, 2004; Ullman, 2007).

The level and type of relationship between offender and victim has also been consistently supported as a contributing factor in the amount of violence and victim injury during a sexual assault. The most substantial differences have been found to lie between stranger-perpetrated versus acquaintance-perpetrated attacks (e.g., Weaver et al., 2004). However, sexual assaults between members of an intimate partnership (or courtship rape) have been shown to be among the most common types of sexual assaults (Ullman, 2007). Despite the obvious agreement that offender–victim relationship is an important dimension to consider in any analysis of sexual assault, the literature has been inconclusive, and even contradictory, in how this relationship affects violence levels. Risk of violence has been suggested to be highest when the offender is known to the victim (Weaver et al., 2004), when the offender is a stranger (Scott & Beaman, 2004), and when the offender is a stranger or a romantic partner (Coker et al., 1998). This may be different still for child victims, who appear to be victims of more severe abuse when the abuse is perpetrated by a relative (Ullman, 2007). Evidently, this area requires more research to determine the true nature of this relationship.

The effect of various victim characteristics on offender behavior and violence has been examined as well. The main indicator of the importance of victim characteristics is shown in the abundance of studies that intentionally limit their offender sample to either child molesters (e.g., Ullman, 2007) or sex offenders against adults (e.g., Brecklin & Ullman, 2005; Carr & VanDeusen, 2004; Clay-Warner, 2003; Hartwick, Desmarais, & Hennig, 2007; Scott & Beaman, 2004; Ullman & Knight, 1992) or compare these groups (e.g., Beauregard, Stone, Proulx, & Michaud, 2008). This observed age dichotomy demonstrates the substantial expected difference between offenders who target young victims and those who target adults. The limited number of studies of situational predictors of sexual violence that incorporate child and adult victims have found that offenders are more likely to escalate to greater degrees of violence when offending against adult victims (Scott & Beaman, 2004; Weaver et al., 2004). Furthermore, there has been some research regarding the home environment of the victim, with findings that a victim with a criminogenic background is less likely to encounter an escalation of violence resulting in homicide during a sexual assault (Mieczkowski & Beauregard, 2010).

As can be seen from previous studies, victim resistance has been examined in its effects on two outcomes: victim injury and/or sexual assault completion. Although these are pertinent factors to investigate, this research shifts the focus from these more terminal outcomes to the more direct relationship between the victim’s resistance and the offender’s immediate reaction to that resistance. There could very well be a multitude of factors that come into play between when the victim resists and his or her later injuries (e.g., victim frailty) or whether the crime results in its completion (e.g., interference of a bystander). However, the dependent variable throughout the present analyses questions the offender (rather than the victim) as to his direct response to that resistance. Thus, not only is the variable relationship more direct with little to no moderating effects between victim resistance and offender reaction (whether coercive or not), but furthermore, the information is elicited from the offender’s perspective, as opposed to the victim’s point of view. This allows an examination into the decision-making process of the offender based on his situational perceptions, which ultimately direct his intent and behavior.

This study proposes to identify and assess situational and crime variables that affect the offender’s reaction to victim resistance during the course of a sexual assault. Although supplemented with official data, the current examination maintains focus on the offender’s perceptions to determine how offenders react to perceived victim resistance. Specifically, the event characteristics of sexual assaults involving victim resistance will be examined to determine which variables increase the likelihood of offender violence. On the basis of the research conducted by Luckenbill (1977) and Tedeschi and Felson (1994), it is expected that victim characteristics as well as situational variables will significantly influence how an offender reacts to victim resistance. For example, the age of the victim has been shown to affect the level of overall offender violence, with offenders generally most likely to escalate the degree of violence when offending against adult victims (Scott & Beaman, 2004; Weaver et al., 2004). This difference could be due to the strategies employed by the offender to begin an assault, which appear to differ in accordance with the age of the victim (Hartwick et al., 2007). This is presumed to subsequently affect the crime progression and escalation because of interactions that take place between victim and offender.

Furthermore, factors over which the offender has direct control are expected to interact with victim and situational variables and further affect the offender’s reaction to victim resistance. For instance, the offender’s decision to use a weapon during the assault is a factor he controls and is a behavioral component that would directly affect the victim. Previously, it has been found that weapon use increases the likelihood of offender violence, victim injury, and rape completion (Coker et al., 1998; Ullman, 2007; Weaver et al., 2004). Perhaps this effect is due to the increased resistance by the victim, which further results in increased offender coercion. If previous findings (Luckenbill, 1977; Tedeschi & Felson, 1994) hold true for the current sample and analyses, the variables most strongly affecting the offender’s reaction to resistance are expected to be those that directly assess the victim’s behavior (such as resistance level) and the offender’s behavior (such as weapon use or humiliation of the victim) during the crime. If these variables are found to be the most influential, this would support the importance of victim behavior as it affects offender behavior and vice versa: the social interaction cycle.

Methods

All adult males serving a sentence of at least 2 years in a Canadian federal penitentiary for a sexual crime were recruited for a survey between April 1994 and June 2000. The participation rate was 93%, 624 inmates agreeing to collaborate with the research team. All participants signed a consent form indicating that the information gathered would be used for research purposes only. Participants included in this study were mostly Caucasian (87.7%). On average, they were 39 years old (SD = 12.0) and serving a prison sentence of 4.2 years (SD = 3.6). Only cases involving victim resistance were included in the analyses (n = 426).²

Data were collected during semistructured interviews as well as from police records, victim statements, and institutional case files. All questions regarding specific variables were asked of the offender within the interview³; however, information provided by the offender was subsequently corroborated through various official sources (a process known as information triangulation). In cases of discrepancies between self-reported data gathered during the semistructured interview and official data, the latter were always used.

Dependant variable

Offender reaction to victim resistance was a measure of the level of violence used by the offender in response to resistance from the victim during the sexual assault. This variable was dichotomized (0 = nonviolent reaction, 1 = violent reaction). A violent reaction was considered to have occurred when the offender used threats or physical force when faced with resistance from the victim during the commission of the sexual assault. A nonviolent reaction by the offender consisted of stopping, running away, or using negotiation.⁴ Among the 426 sex offenders included in the study, 250 (58.7%) reacted violently and 176 (41.3%) reacted nonviolently.

Victim characteristics

Three victim characteristics were selected: victim gender (0 = female, 1 = male),⁵ victim age (0 = under 12 years old, 1 = 13-15 years old, 2 = over 16 years old),⁶ and victim from poor/dysfunctional background (0 = no, 1 = yes).⁷

Situational variables

Six situational variables were included in the analyses: alcohol and/or drug use before the crime (0 = no, 1 = yes), pornography consumption prior to the crime (0 = no, 1 = yes), premeditation of the crime (0 = no, 1 = yes),⁸ time of the crime (0 = day, 1 = night), level of offender–victim intimacy (0 = stranger, 1 = known/friendly, 2 = romantic relationship),⁹ and type of victim resistance (0 = low level/passive, 1 = verbal, 2 = physical). Alcohol or drug use and pornography consumption prior to the crime referred to use of these materials within a few hours before the commission of the assault.

Crime characteristics

Five crime-characteristic variables were included in the analyses: strategies to commit the crime (0 = no specific strategy, 1 = nonviolent persuasion, 2 = violent persuasion),¹⁰ weapon use (0 = no, 1 = yes), nature of the sexual acts (0 = nonintrusive, 1 = intrusive),¹¹ humiliation of the victim (0 = no, 1 = yes), and time spent with the victim (0 = less than 30 minutes, 1 = more than 30 minutes). In general, crime variables describe factors over which the offender has direct control during the sexual assault. The offender must therefore make a decision with respect to each of these factors while the criminal event is taking place, and interactions with the victim and the victim’s responses may affect some or all of these decisions. The frequencies for these variables can be found in Table 1; as can be seen, in no case did the percentage of a single category in any variable exceed 90%.

Table 1.

Univariate Descriptors of Victim, Situational, and Crime Variables

Variable	n	%
Victim characteristics
Victim gender
Female	364	85.4
Male	62	14.6
Victim age
≤ 12 years old	143	33.6
13-15 years old	57	13.4
≥ 16 years old	226	53.1
Poor/dysfunctional background
No	276	64.8
Yes	150	35.2
Crime variables
Strategies to commit crime
No specific strategy	80	18.8
Non-violent persuasion	85	20.0
Violent persuasion	261	61.3
Weapon use
No	292	68.5
Yes	134	31.5
Nature of sexual acts
Non-intrusive	110	25.8
Intrusive	316	74.2
Humiliation of victim
No	296	69.5
Yes	130	30.5
Time spent with victim
< 30 minutes	268	62.9
> 30 minutes	158	37.1
Situational variables
Alcohol or drugs
No	167	39.2
Yes	259	60.8
Pornography
No	378	88.7
Yes	48	11.3
Premeditation
None/unstructured	301	70.7
Structured	125	29.3
Time of crime
Day/both day and night	202	47.4
Night only	224	52.6
Offender-victim intimacy
Stranger	94	22.1
Known/friendly	148	34.7
Romantic relationship	184	43.2
Type of resistance
Low level/passive	88	20.7
Verbal	116	27.2
Physical	222	52.1
Dependent variable
Offender reaction to victim resistance
Violent	250	58.7
Non-violent	176	41.3

A sequential logistic regression was chosen as the first multivariate analysis because of the dichotomy of the dependent variable. The main purpose of the logistic regression was to determine the model that best predicted a violent reaction by sex offenders to victim resistance. Sequential logistic regression was used so that variables could be entered into the model according to their logical sequence of occurrence during the criminal event (victim characteristics, precrime, during crime).

Following the sequential logistic regression, Exhaustive Chi-square Automatic Interaction Detection (CHAID) was performed to identify interactions and relationships between the independent variables that affect the prediction of the dependent variable. CHAID is a type of decision-tree technique that automatically computes a series of cross-tabulations for all pairs of independent variables (Kass, 1980). The most significant of these cross-tabulation results are then incorporated into a classification tree. The tree divides the data into mutually exclusive subsets—or nodes—that account for all of the data and that best describe the dependent variable (Kass, 1980).

The ordering of each successive split of the data is an important feature of CHAID. The top node contains all of the data and the most significant variable associated with the dependent variable determines the first split. Each node, or group of cases, is then sequentially split based on the significance of variables and interactions within that node. This splitting process is continued until a stopping rule is invoked or until there are no more variables that significantly split the remaining cases. In this analysis, we used a variation of the CHAID procedure known as Exhaustive CHAID (Biggs, De Ville, & Suen, 1991). It differs slightly in its algorithm from standard CHAID, but optimizes the selection of the appropriate variables. The predictive accuracies of the logistic regression model, as well as the Exhaustive CHAID model, were tested using Receiver Operating Characteristic (ROC) analysis.

Results

The first multivariate analysis was the development of a sequential logistic regression model that best predicted the “offender reaction” outcome. As can be seen in Table 2, the six variables that provided significant (p ≤ .05) additions to the explained variance of the model at their point of entry were retained for the parsimonious model¹²: victim age, type of victim resistance, strategies to commit the crime, weapon use, humiliation of the victim, and time spent with the victim. The area under the curve (AUC) increased with each block addition, from .794 in Block 1 to .850 in Block 2 to .932 in Block 3, indicating a high level of discriminatory accuracy.

Table 2.

Parsimonious Sequential Logistic Regression Model of Offender’s Reaction to Victim Resistance

	Block 1					Block 2					Block 3
Predictor	B	SE	Sig	OR	95% CI	B	SE	Significance	OR	95% CI	B	SE	Significance	OR	95% CI
Victim characteristics
Victim age (≥16 years)	—	—	.00	—	—	—	—	.00	—	—	—	—	.18	—	—
≤12 years	−2.75	.27	.00	0.06	.04-.11	−2.14	.29	.00	0.12	.07-.21	−0.63	.39	.11	0.53	.25-1.15
13-15 years	−2.16	.33	.00	0.12	.06-.22	−1.80	.36	.00	0.17	.08-.33	−0.67	.47	.15	0.51	.20-1.28
Situational variables
Type resist (physical)	—	—	—	—	—	—	—	.00	—	—	—	—	.00	—	—
Low level/passive	—	—	—	—	—	−1.72	.33	.00	0.18	.09-.34	−1.46	.43	.00	0.23	.10-.54
Verbal	—	—	—	—	—	−1.85	.30	.00	0.16	.09-.28	−1.72	.39	.00	0.18	.08-.39
Crime variables
Strategies (violent)	—	—	—	—	—	—	—	—	—	—	—	—	.00	—	—
No strategy	—	—	—	—	—	—	—	—	—	—	−2.19	.47	.00	0.11	.04-.28
Non-violent	—	—	—	—	—	—	—	—	—	—	−1.43	.43	.00	0.24	.10-.55
Weapon use	—	—	—	—	—	—	—	—	—	—	0.80	.40	.05	2.22	1.00-4.90
Humiliation	—	—	—	—	—	—	—	—	—	—	2.01	.43	.00	7.49	3.26-17.25
Time with victim	—	—	—	—	—	—	—	—	—	—	0.91	.36	.01	2.49	1.24-4.99
AUC	.794					.850					.932
Cox and Snell R²	.286					.363					.508
Hosmer and Lemeshow significance	—					.235					.259

Note: B = beta weights; SE = standard error; OR = odds ratio; CI = confidence interval; AUC = area under the curve.

With reference to the type of victim resistance, results indicate that offenders were less likely to become violent when they encountered low level/passive resistance (odds ratio [OR] = .23; p < .001) or verbal resistance (OR = .18; p < .001) as compared to physical resistance. When an offender used a weapon, the reaction of the offender to victim resistance was 2.22 times more likely to be violent. When the offender resorted to humiliation, victims who resisted were 7.49 times more likely to be subjected to physical violence. When the offender spent more than 30 minutes with the victim, the offender reaction was 2.49 times more likely to be violent if the victim resisted.

The second multivariate analysis to be conducted was an Exhaustive CHAID. The variables that had been found to be significant at p ≤ .25 at some point in the sequential logistic regression models were entered into the Exhaustive CHAID model.¹³ Of the 14 variables initially entered into the logistic regression model, only two—victim from poor/dysfunctional background and pornography consumption—did not achieve this significance level and therefore, were left out of the Exhaustive CHAID analysis.

The remaining 12 variables were entered into the Exhaustive CHAID analysis to determine the interaction model that best estimated the violent reactions by sex offenders upon encountering victim resistance. The resulting Exhaustive CHAID classification tree is presented in Figure 1. The percentage of cases that involved violent and nonviolent reactions is presented within each box (node). The level with the highest percentage within each node represents the associated offender response at that point in the classification tree.

Figure 1.

Exhaustive CHAID decision-tree of victim, situation, and crime characteristics on offender’s reaction to victim resistance.

The specifications of the Exhaustive CHAID tree were dependent on the factors that increased the classification accuracy. Various models were run, and the presented model is the best fitting, most predictive model.¹⁴ As Figure 1 indicates, the variable most strongly associated with offender reaction to victim resistance is the offender’s strategy. This variable classifies the sample into two nodes—one in which the offenders report nonviolent strategies or no specific strategy to commit the offense and, alternatively, offenders who utilize a violent persuasive strategy. Within the former, a nonviolent reaction is identified in 82.4% of cases (n = 136) with violent reactions occurring in 17.6% of cases (n = 29). As might be expected, verbal or passive resistance is associated with a nonviolent reaction and physical resistance is associated with a violent reaction. When victim resistance was passive or verbal, violence was encountered in 8.3% of cases (n = 11), whereas nonviolent reactions took place in 91.7% of cases (n = 122). Victims who physically resisted elicited violent reactions in 56.2% of cases (n = 18) and nonviolent reactions in 43.8% of cases (n = 14).

If an offender began, however, with a violent persuasive strategy, violence is the response consistently identified by the model: violence took place in 84.7% of cases (n = 221) and nonviolent reactions in only 15.3% (n = 40). This gives an indication of the importance of the first splitting variable, offender strategy.¹⁵ A violent persuasive strategy interacts with humiliation of the victim, so that humiliation is associated with a greater likelihood of a violent reaction to victim resistance. When the offender subjected the victim to humiliation, violence was present in 93.8% of cases (n = 105) and no violence in only 6.2% of cases (n = 7). When the assault did not include humiliation, violence—although still the predicted response—was present in 77.9% of cases (n = 116) and a nonviolent reaction to resistance occurred in 22.1% of cases (n = 33).

In the absence of humiliation of the victim, possession of a weapon is an important predictor of violence. Violence increased when the offender was in possession of a weapon, with 86.4% of cases (n = 57) resulting in violence and 13.6% of cases (n =6) not resulting in violence. When the offender did not possess a weapon, the likelihood drops: 71.1% of offenders (n = 59) became violent, whereas 28.9% of offenders (n = 24) did not.

Interestingly, when the offender does humiliate the victim during the sexual assault, the type of resistance by the victim is again identified by the CHAID model. Again, passive and verbal resistance were combined by the model and compared with physical resistance, with the results indicating a greater probability of violence in the case of physical resistance. When the victim passively or verbally resisted, violence occurred in 82.4% of cases (n = 28) and a nonviolent reaction in 17.6% of cases (n = 6). When the victim physically resisted, offender violence was present in the highest proportion in the entire model, with 98.7% of offenders (n = 77) reacting violently to victim resistance and only 1.3% of offenders (n = 1) reacting nonviolently.

Discussion

Some victims of sexual assault do not resist the attacks of their assailants, often because they fear for their lives or are wary of the negative social consequences—such as embarrassment or rejection—that may result from misinterpreting or overreacting to a man’s cues (Rozee & Koss, 2001). Women especially are afraid that resistance will only serve to increase their chances of being injured or killed by their attacker (Rozee & Koss, 2001).

The results of the analyses presented here indicate that specific variables and interactions among variables increase the risk of sexual assault victims experiencing a violent reaction to their resistance. Overall, the most influential situational factor in both the original logistic regression and the CHAID models is the strategy used by the offender to commit the crime. If the offender uses a violent persuasive strategy, the most likely scenario is one of violence; even after interaction with other variables within the CHAID tree, violence remains the consistently predicted offender response. This relationship is logical, because it is reasonable to suppose that an offender who begins his crime in a violent manner is more likely to react violently to resistance from his victim. This violence may be due to personality characteristics that predispose the offender to act in a violent manner (Carr & VanDeusen, 2004). However, it is also plausible that situational factors influence violence and increase the offender’s anger or modulate others factors that increase aggression. Tark and Kleck (2004) have found that when victims who resisted were hurt, it was almost always injury that came first, suggesting that the offender had decided to use violence before any interaction with the victim.

The present results also consistently found the type of victim resistance to be significant, and to interact with offender strategy. The importance of this variable, and specifically its interaction with offender strategy, is consistent with previous findings indicating that resistance strategies of the victim tend to match offender strategies (Ullman, 1998, 2007; Ullman & Knight, 1992). Essentially, if the offender begins the assault using physical aggression, the victim is more likely to react with physical resistance. Furthermore, Scott and Beaman (2004) report physical resistance by adult women to be triggered by threats with a weapon or being physically hit or punched. Thus, the modification of victims’ resistance strategy in response to violent persuasion by the offender was expected and is reported in the literature to be a valid, predictive relationship. The converse of this relationship was supported in this study: The victim’s physical resistance, in contrast to passive or verbal resistance, increases the likelihood of a violent reaction by the offender. Although our data are cross-sectional, making it impossible to conclude for sure, our findings seem to suggest that violent persuasion by the offender increases physical resistance by the victim, which further increases the use of violence by the offender.

The fact that humiliation had a significant association with violence in both models is suggestive of the importance of offender motivation for sexual assault. Knight and Prentky (1990) emphasize elements of anger and sadism as pertinent components of underlying motivations to rape. According to Knight and Prentky (1990), offenders motivated by anger are undifferentiated in their expression of this emotion, and those motivated by sadistic tendencies use physically damaging assault techniques, confusing sexual and aggressive drives. It is not unrealistic to predict that offenders motivated by angry or sadistic goals will resort to humiliation as a specific form of abuse, especially because humiliation has been found to be a form of abuse practiced by those diagnosed with sexual sadism (Fedoroff, 2008).

The final variable determined to be a significant addition to both the logistic regression and the CHAID models was the use of a weapon by the offender, with violence more likely when a weapon was present. There are several possible reasons that an offender who utilizes a weapon would be more likely to become violent. First, an offender who brandishes a weapon potentially has a different range of coercive possibilities than does an offender who does not use a weapon. If the attacker is prepared with a weapon, he is likely willing to use it to either injure or threaten the victim if the need arises (Coker et al., 1998), which could very well be the case if the victim tries to resist his assault. An attacker without a weapon is likely not anticipating resistance and, although he could subdue a resistant victim without a weapon—using physical force or verbal threats—he might also view fleeing the scene as the easiest option. Second, if an offender possesses a weapon during the commission of a sexual assault, use of that weapon could be viewed as a much easier solution than physical restraints or threats. If there is no weapon to use, the only options for the offender are to manually restrain or subdue the victim or flee. Thus, if the offender is in possession of a weapon and is motivated to complete the sexual assault, threats with or use of a weapon are more likely to ease the completion of the assault (Mieczkowski & Beauregard, 2010). Finally, victims may be more evenly matched to an attacker without a weapon and are therefore possibly more likely to physically resist. Although male perpetrators are often more physically capable than their victims—who are primarily women or children—the victim has a better chance to escape or to make the offender doubt his chances of assault completion if the offender does not have a weapon.

The findings that victim age and time spent with the victim were significant in the logistic regression model were not unexpected. Victims older than 16 years were found to be more at risk of encountering violent reactions to their resistance than were victims 12 years old and younger or victims between the ages of 13 and 15. This finding reiterates previous evidence about differences in violence and physical injuries sustained by adults and children within a sexual assault, with greater levels of violence and injury reported for adults (Scott & Beaman, 2004; Weaver et al., 2004). Furthermore, charges and convictions of aggravated sexual assault—which involve physical injury of the victim—are less likely in cases of sexual abuse of children than in cases of sexual assault of adults (MacMartin, 2004). Thus, age of the victim has consistently been determined to be an important variable associated with offender violence and resultant victim injury, and its role was confirmed by the current analyses.

The final significant factor determined to influence offender violence was the amount of time the offender spent with the victim during the assault. The likelihood of violence in response to victim resistance was greater when the sexual assailant spent more than 30 minutes with the victim. This connection could be because of a greater amount of time available in which to harm the victim. Alternatively, the assailant may become frustrated and angry when the sexual assault takes longer than he had planned, leading to aggressive and violent behavior (Mieczkowski & Beauregard, 2010).

Although not all of the aforementioned variables were found to be significant within the post hoc sequential logistic regression model, there were enough similarities to support and validate the original findings. Even after deleting all the cases that involved a physically coercive offending strategy to begin the assault, the model still found the victim age, the type of victim resistance, humiliation of the victim, and the amount of time spent with the victim to be important factors in determining the offender’s reaction to victim resistance. These post hoc results are evidence of the fact that although the offender’s strategies at the outset of the crime are important in how he reacts to victim resistance (as was expected), this one variable does not tell the whole story. There are other factors that affect how the crime progresses, not the least of which relating to the victim’s characteristics and behaviors. This serves as solid support for the original findings as well as for the theoretical background in general.

The factors determined to be important to the assaultive interchange are particularly congruent with those from the study by Luckenbill (1977). Although Luckenbill posits that certain factors are related to an increase in victim resistance, an offender’s reaction to resistance may also increase because of similar factors. Features of the abusive scenario that increase victim desperation, leading to increased resistance, would also arguably increase the offender’s belief in his own capabilities. Thus, his increased confidence could make the offender less afraid of consequences and more likely to believe that he can overcome resistance with violence.

The significance of the offender attack strategy and offender weapon use relate to this interpretation of Luckenbill’s (1977) role-related conditions. Both crime elements would potentially increase threat severity as well as its believability and the offender’s perceived capacity to carry out the threat. An increased level of offender violence at the outset of the attack would arguably increase the victim’s fear of threats because they have already been privy to a level of violence that the offender is capable of and willing to use. In most cases, the victim’s fear would be evident to the offender, thus increasing his satisfaction with his performance or at least his belief in his own power and domination. Similarly, offender weapon use could easily demonstrate an obvious coercive advantage to the offender who wields it, giving credence to his threat, increasing victim resistance and fear.

Also in support of Luckenbill’s (1977) role-related conditions are the findings that more time spent with the victim, victim humiliation, and increased victim age are all related to an increase in offender violence. As time passes, threats may become more believable and the victim may begin to lose hope, resulting in desperation and a decrease in their believed capacity to oppose. Whether transmitted through victim actions or his own interpretations, the offender may also recognize that desperation grows and victim capacity decreases as time under his control builds. In the same way, humiliation increases the power—whether real or simply perceived—that the offender holds over the victim, which also acts to increase the offender’s perceived capability to overcome any resistance with physical violence. Younger victims may be viewed by the offender as possessing less oppositional capabilities; furthermore, because threats from the victim (delivered via the use of resistance) are perceived as less severe and believable when the victim is young, the offender’s reaction would be correspondingly less coercive.

The finding that the type of victim resistance has such a momentous impact on how the offender reacts demonstrates the inherent link between victim behavior and subsequent offender behaviors. Not only does this support previous literature, in that the behavior of one actor impacts the behavior of others within the exchange (Luckenbill, 1977; Tedeschi & Felson, 1994) but it also highlights the importance of Luckenbill’s (1977) working agreement that is developed between the victim and offender. When a victim physically resists his or her attacker, although more than justified, the offender perceives this physical resistance as entrance into an agreement in which violence is an acceptable tool that may be used to achieve his goals. Thus, Luckenbill’s (1977) theoretical reasoning predicted increased physical coercion in response to physical resistance.

At this point, it is imperative to address possible limitations of this study. The fact that a group of nonresisting victims was not included as a comparison group could potentially be cause for concern. It is possible that lack of resistance may lead to differing levels of offender violence. However, this study was intentionally designed to examine how an offender reacts to victim resistance, so that this specific step in the offense process could be better elucidated. If there is no resistance from the victim, this step does not exist within that particular criminal event. The variable examining the level of victim resistance was the most logical alternative within the current study—addressing the differences in victim resistance without a nonresister comparison group. Further studies of the phenomenon of victim resistance should attempt to include a true comparison group while maintaining focus on social interactions and direct, microlevel victim–offender dynamics.

A second limitation that must be addressed is the fact that this research was analyzed from the perspective of the offender. Thus, the data gathered, although corroborated through consultation of official data, were derived from the offenders’ perceptions during the course of the assault. For this study, this is actually viewed as a strength, as the focus remains on the offender’s decision-making processes during the crime itself and thus, relevant information may only be obtained from the offender himself. However, future research should endeavor to determine whether the factors determined to be important in how offenders react to victim resistance within this study are paralleled in the victim’s perception as well. For example, is the offender’s perception of what “physical resistance” looks like similar to what a victim views as physical resistance? This is an important factor to examine in order for policy to be capable of properly informing victims.

It appears that the most dangerous situation for a sexual assault victim is the combination of the offender’s use of a violent persuasion strategy, the offender using humiliation tactics as a form of abuse during the assault, and physical resistance by the victim; the presence of a weapon is an additional aggravating factor under particular circumstances. Furthermore, victims older than 16 years are at greater risk of a violent reaction from the offender as are victims of an assault that lasts longer than 30 minutes. The logistic regression and CHAID models are both compelling endorsements of the importance of the individual significant variables, as well as the interactions that take place between variables, within the criminal event.

Conclusion

The social interactionist perspective emphasizes the true importance of behaviors of the various players within a coercive interchange (Luckenbill, 1977; Tedeschi & Felson, 1994). As evidenced by the findings within this study, offender behavior has the potential to drastically alter victim behavior, especially the level of victim resistance. However, the current analyses have shown the reverse to be true as well: Victim behavior also considerably affects offender behavior. Behavior appears to be cyclical in the sense that actors continually affect one another: Offender assaultive behavior leads to victim resistance, which, depending on the presence or degree of certain factors, may then lead to offender coercion in response to that resistance.

These results support the original hypotheses with respect to previous research on the subject (Luckenbill, 1977; Tedeschi & Felson, 1994). Not only did many of the victim-, situational-, and offender-controlled (modus operandi) variables prove to be important in the level of coercion in response to victim resistance but, also, the most influential variables were the victim and offender behavior variables. The victim behavior variable (type of victim resistance) and one of the most relevant offender behavior variables (strategies to commit the crime) were shown to be among the strongest predictors of the offender’s reaction to victim resistance. Overall, these results reemphasize the significance of the victim and offender roles and the impact that victim and offender have on one another within a coercive interchange.

The information discovered herein could have potential utility in the context of programming and policy initiatives. More significantly, it could be employed to help educate potential sexual assault victims in harm reduction and situational prevention strategies. Previously, rape prevention programs have focused on increasing general knowledge about rape as well as attempting to reduce rape myth acceptance (Ullman, 2007). However, these programs are generally ineffective in changing rape attitudes or reducing rape (e.g., Brecklin & Forde, 2001). Specifically, with respect to the current results, potential adult victims could instead be taught when it is in their best interests to resist their attacker and at what point it becomes in their best interests—in terms of harm reduction—to stop resisting. For example, if an offender is using violent means to attempt to persuade the victim, it appears to be in the victim’s best interests to verbally resist (such as calling for help and ensuring the offender hears an undeniable and unambiguous “No”), on the off-chance that this is enough to stop the assault or get help without inducing greater physical coercion, or attempt to flee as soon as possible. However, perhaps more research should first be conducted on the psychological effects of not fully resisting a sexual assault, as surrendering to an assailant could possibly take a greater toll on the victim than physical injury (Rozee & Koss, 2001). Furthermore, a completed rape is itself an injury and must not be disregarded as such. Although reducing physical coercion and offender violence levels is imperative to increase victim safety, recommendations should not be made to victims without accounting for the injurious aspect of the rape itself.

Regardless of these possibilities in victim programming, as the overall results unfolded, it began to become increasingly clear that, at this point, the issue is far too complex and unexplored to make any blanket statements that apply to all potential victims. This study examined only one small portion of the entire event that results in a completed sexual assault: How does victim resistance impact offender reactions? Further studies are needed to evaluate the criminal event so that the entire process can be better understood with an aim of prevention and situational intervention. Only after uncovering the complex offender–victim interactions that lead to varying levels of offender violence and coercion throughout the offending sequence will researchers be able to begin to advise potential victims as to the most protective course of action. Educating victims prematurely could lead to a greater proportion of sexual assaults resulting in victim injury, hospitalization, and death (Mieczkowski & Beauregard, 2010).

Although the emphasis of this study was clearly on event characteristics, the social interactionist perspective suggests the importance of offender characteristics as well (e.g., offending history, psychiatric diagnoses, deviant arousal). For instance, it can be hypothesized that previous violence and offense history are likely to affect the level of violence in subsequent offenses, as a result of the development of a script for victim resistance.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interests with respect to the research, authorship, and/or publication of this article.

Funding

This research was supported by the Social Sciences and Humanities Research Council.

Notes

Bios

Samantha Balemba is currently a PhD student in criminology at Simon Fraser University. Her research interests include sex offender treatment and etiology, situational crime prevention and harm reduction, victimology, and offender violence. She has published in Violence and Victims and is a member of the American Society of Criminology, the Association for the Treatment of Sexual Abusers, and the Centre for Research on Sexual Violence.

Eric Beauregard received his doctorate from the University of Montreal. He has specific research interests in understanding the offending process of sex offenders, verifying how psychological, environmental, and situational factors influence the modus operandi of sex offenders. He is the codirector of the Centre for Research on Sexual Violence and has published in a number of prominent journals, including Violence and Victims, Aggression and Violent Behavior, Criminal Justice and Behavior, Justice Quarterly, Journal of Research in Crime and Delinquency, Sexual Abuse: A Journal of Research and Treatment, and Journal of Interpersonal Violence. He is a member of the American Society of Criminology and the Association for the Treatment of Sexual Abusers.

Tom Mieczkowski is a researcher and academic whose interests have included drug smuggling, theories of syndicated crime organizations, street gangs, drug distribution organizations and methods, drug epidemiology, the validation of various drug detection technologies, and the estimation of drug prevalence and incidence using bioassays and survey methods. He has published more than 100 scholarly articles, book chapters, and three books. Since receiving his PhD from Detroit’s Wayne State University in 1985, he has received more than one million dollars in research funding. He is a member of the International Association of Forensic Toxicology, The British Academy of Forensic Sciences, The European Hair Research Society, and The American Society of Criminology.

References

Beauregard

Stone

M. R.

Proulx

Michaud

(2008). Sexual murderers of children: Developmental, precrime, crime, and postcrime factors. International Journal of Offender Therapy and Comparative Criminology, 52, 253-269.

Biggs

de Ville

Suen

(1991). A method of choosing multiway partitions for classification and decision trees. Journal of Applied Statistics, 18, 49-62.

Brecklin

L. R.

Forde

D. R.

(2001). A meta-analysis of rape education programs. Violence and Victims, 16, 303-321.

Brecklin

L. R.

Ullman

S. E.

(2005). Self-defense or assertiveness training and women’s responses to sexual attacks. Journal of Interpersonal Violence, 20, 738-762.

Carr

J. L.

VanDeusen

K. M.

(2004). Risk factors for male sexual aggression on college campuses. Journal of Family Violence, 19, 279-289.

Clay-Warner

(2003). Context of sexual violence: Situational predictors of self-protective actions. Violence and Victims, 18, 543-556.

Coker

A. L.

Walls

L. G.

Johnson

J. E.

(1998). Risk factors for traumatic physical injury during sexual assaults for male and female victims. Journal of Interpersonal Violence, 13, 605-620.

Fedoroff

J. P.

(2008). Sadism, sadomasochism, sex, and violence. Canadian Journal of Psychiatry, 53, 637-646.

Hartwick

Desmarais

Hennig

(2007). Characteristics of male and female victims of sexual coercion. Canadian Journal of Human Sexuality, 16, 31-44.

10.

Hosmer

D. W.

Lemeshow

(2000). Applied logistic regression (2nd ed.). Chichester, UK: Wiley.

11.

Kass

G. V.

(1980). An exploratory technique for investigating large quantities of categorical data. Applied Statistics, 29, 119-127.

12.

Knight

R. A.

Prentky

R. A.

(1990). Classifying sexual offenders: The development and corroboration of taxonomic models. In Marshall

W. L.

Laws

D. R.

Barbaree

H. E.

(Eds.), Handbook of sexual assault (pp. 23-54). New York, NY: Guilford.

13.

Leclerc

Proulx

Lussier

Allaire

J. F.

(2009). Offender-victim interaction and crime event outcomes: Modus operandi and victim effects on the risk of intrusive sexual offenses against children. Criminology, 47, 595-618.

14.

Luckenbill

D. F.

(1977). Criminal homicide as a situated transaction. Social Problems, 25, 176-186.

15.

MacMartin

(2004). Judicial constructions of the seriousness of child sexual abuse. Canadian Journal of Behavioural Science, 36, 66-80.

16.

Mieczkowski

Beauregard

(2010). Lethal outcome in sexual assault events: A conjunctive analysis. Justice Quarterly, 27, 332-361.

17.

Quinn

J. F.

Forsyth

C. J.

Mullen-Quinn

(2004). Societal reaction to sex offenders: A review of the origins and results of the myths surrounding their crimes and treatment amenability. Deviant Behavior, 25, 215-232.

18.

Rozee

P. D.

Koss

M. P.

(2001). Rape: A century of resistance. Psychology of Women Quarterly, 25, 295-311.

19.

Scott

H. S.

Beaman

(2004). Demographic and situational factors affecting injury, resistance, completion, and charges brought in sexual assault cases: What is best for arrest? Violence and Victims, 19, 479-494.

20.

Tark

Kleck

(2004). Resisting crime: The effects of victim action on the outcomes of crime. Criminology, 42, 861-909.

21.

Tedeschi

J. T.

Felson

R. B.

(1994). Violence, aggression, and coercive actions. Washington, DC: American Psychological Association.

22.

Ullman

S. E.

(1998). Does offender violence escalate when rape victims fight back? Journal of Interpersonal Violence, 13, 179-192.

23.

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.

24.

Ullman

S. E.

Knight

R. A.

(1992). Fighting back: Women’s resistance to rape. Journal of Interpersonal Violence, 7, 31-43.

25.

Weaver

G. S.

Wittekind

J. E.

Huff-Corzine

Corzine

Petee

T. A.

Jarvis

J. P.

(2004). Violent encounters: A criminal event analysis of lethal and nonlethal outcomes. Journal of Contemporary Criminal Justice, 20, 348-368.