When a Date Changes From Fun to Dangerous

Abstract

Women’s ability to perceive risk of acquaintance sexual assault is key to decreasing its likelihood. This study examined the relationship between women’s recognition of known situational risk factors and global perceived risk, prior victimization, and drinking habits. Women consistently distinguished between clear and ambiguous situational risk factors although these were not related to perceived global risk. They also manifested positivity bias, the tendency to underestimate their own risk of being assaulted relative to other women. Both prior victimization and drinking habits were related to risk judgments. This study raises important issues concerning developing effective defensive strategies for combating sexual assault.

Early recognition that a social situation with a male acquaintance or intimate partner has become threatening can aid a woman in preventing a serious incident of sexual aggression. Despite substantial evidence that women are more likely to be assaulted by someone they know than by a stranger (Koss, Gidycz, & Wisniewski, 1987), women typically fear stranger sexual aggression more (Hickman & Muehlenhard, 1995) and prepare themselves better to fend off an aggressive incident by a stranger (Fischoff, Furby, & Morgan, 1987). This study was conducted to explore some of the cognitive processes that underlie women’s perceptions of sexual aggression risk by a male acquaintance.

The first goal of this study was to assess the extent to which women recognize known risk factors associated with acquaintance sexual aggression as presenting a potential threat for them personally. Judging that a social situation is dangerous is a more complex psychological task than making the same judgment about a stranger attack (Amick & Calhoun, 1987). Many risk factors for sexual assault such as alcohol consumption, a man paying for dating expenses, and being alone with a man (Harney & Muehlenhard, 1991; Mandoki & Burkhart, 1989; Muehlenhard & Linton, 1987) are also common elements of socializing. Thus, women may not view these risk factors as danger cues, but rather see them as positive or normal aspects of socializing with men.

The second goal of this study was to examine the relationships between situationally specific and global risk perceptions. Are women’s judgments of specific situational risk factors related to a global sense of being at risk? As noted by Nurius (in press), the two phenomena involve different types of cognitive assessments. Evaluating risk for a specific outcome requires focusing on the situation as well as on one’s relationships, goals, and history. Within this framework, one considers what might pose a threat, how large the threat is, and how it might be countered (Smith, Haynes, Lazarus, & Pope, 1993; Weinstein, 1993). In contrast, global risk perception involves focusing on broad environmental conditions that pertain to whole groups rather than to an individual herself. One might understand in general terms that women are at high risk for sexual assault, but not relate that global risk to one’s specific circumstances. Because these two types of risk perceptions have different cognitive foci, we hypothesized that specific situational and global risk perceptions would at best be only moderately correlated.

A third goal of this study was to examine whether women judge the risk of sexual aggression for other women the same as for themselves. People often maintain an unrealistic level of optimism, or positive illusions, about the likelihood of “bad things” happening to oneself relative to others (Taylor & Brown, 1988; Weinstein, 1984). Research in a number of health-related areas indicates that individuals generally regard their own risk for injury, illness, or other health hazards as significantly lower than for others (see, for instance, McKenna, 1993). This optimistic bias also inclines women to underestimate their own likelihood of encountering sexual aggression. Surveys by Hoecker (1992) and Nason (1995) have shown that women hold optimistic bias about their likelihood of experiencing sexual aggression to a greater extent than they do about other health problems such as breast cancer or sexually transmitted diseases, and that women also maintain more of an optimistic bias toward acquaintance rape than toward stranger rape (Hoecker & White, 1995). Furthermore, experimental evidence has demonstrated that college women who read a vignette depicting a social encounter between a man and a woman tended to judge a higher likelihood of nonconsensual sex occurring if the woman was presented as someone else rather than themselves (Cue, George, & Norris, 1996). Thus, we hypothesized that women’s global perceptions of risk would be lower for themselves than for other women.

The final goal of this study was to investigate the relationship between two individual difference factors—prior victimization experience and drinking habits—to women’s perceptions of situational and global risk for acquaintance sexual aggression. There is evidence suggesting that women with a history of prior adult sexual victimization have a higher level of global risk perception than those who have not been victimized (Hoecker & White, 1995; Norris, Nurius, & Dimeff, 1996). Prior victimization is also related to a number of factors that could affect situational risk perception such as alcohol consumption, embarrassment, fear of rejection, and the belief that alcohol clouds one’s abilities (Norris et al., 1996). History of victimization is also related to a higher number of sex partners (Mandoki & Burkhart, 1989; Norris et al., 1996). Because a higher number of partners is linked to an increased risk for sexual aggression (Koss & Dinero, 1989; Mandoki & Burkhart, 1989), and because previously victimized women seem to recognize at a global level that they are at increased risk, they may have heightened awareness of risk in specific situations. Yet, the findings of Norris et al. (1996) indicate that specific situational impediments such as embarrassment and fear of rejection or alcohol consumption, may impede recognition of situational threat. Therefore, we hypothesized that prior victimization would be more strongly related to global risk perception than to a recognition of specific situational risk factors.

Alcohol consumption can impair a woman’s ability to perceive risk both through its pharmacological and its expectancy effects (Abbey, 1991; Abbey, Ross, & McDuffie, 1994; Abbey, Ross, McDuffie, & McAuslan, 1996; Graham, 1996; Nurius & Norris, 1996; Nurius, in press). Alcohol consumption in social situations is normative and may be associated with expectations of positive social outcomes—having fun, easier conversation, better sex (for relevant reviews, see Crowe & George, 1989; Leigh, 1989). A woman holding these positive expectations might not be on the lookout for possible threat (Nurius & Norris, 1996). However, drinking (especially excessively) in social situations may subject women to an increased risk of experiencing sexual aggression (Abbey, 1991; Abbey et al., 1996; Muehlenhard & Linton, 1987), while making them less able to perceive a threat or respond to it (Abbey, 1991; Abbey et al., 1996; Norris et al., 1996). Furthermore, Cue et al. (1996) found that college women did not perceive drinking as a risk factor for encountering sexual aggression, although they did not examine the influence of participants’ drinking habits on this perception. In general, heavier drinkers tend to perceive fewer negative consequences of drinking than lighter drinkers (see Leigh, 1989 for a review); women who drink, especially to excess, may be less likely than nondrinkers to associate drinking with being sexually victimized. Thus, because they hold positive associations between alcohol consumption and socializing, we hypothesized that drinkers would be less likely than nondrinkers to recognize situational risk, especially those risks associated with drinking.

In summary, this study investigated several aspects of women’s perceptions of risk associated with acquaintance sexual aggression. First, to what extent do women actually recognize known risk factors as presenting a danger to themselves? Second, are these specific situational risk judgments associated with global perceptions of risk? Third, do women judge the risk of sexual aggression as less for themselves relative to other women? Finally, are individual differences in prior victimization experiences and drinking habits related to women’s situational and global risk judgments?

METHOD

PARTICIPANTS

Participants were 152 undergraduate women recruited from the Psychology Department Human Subjects Pool at the University of Washington. Twelve questionnaires were discarded because of insufficient or uninterpretable data. The resulting 140 participants had a mean age of 20.3 years (range 18 to 50), and the majority were either freshmen (41.4%) or sophomores (21.4%); the rest were juniors (24.3%) and seniors (10.7%). Of the participants, 58% were White, 32% Asian/Pacific Islander, and the remaining 10% either Latina, African American, Native American, or multiple ethnicity.

PROCEDURE AND DESIGN

The women participated in groups of up to 12. A research assistant introduced the study, obtained written informed consent from each participant, and then instructed participants about the risk rating procedure and completion of the background measures. When everyone had completed the questionnaires, the research assistant debriefed the group, answered questions, and distributed credit slips. All participants were given a resource sheet listing campus and community resources for dealing with sexual assault trauma. Sessions took approximately 1 hour. Participation was voluntary, responses were anonymous, and participants received experimental credit in a psychology class.

Each participant completed risk ratings for one of three dating scenarios—a first, fifth, or “typical” date. The first and fifth date scenarios were presented in the following way: “You are on a ‘typical first [or fifth] date.’ It has gotten toward the end of the evening, and you have started to ‘mash.’ That is, you are enjoying kissing each other on the lips, face, and neck. However, you know that you do not want to have sexual intercourse.” For the typical date, participants were instructed to describe what they would consider a typical next date with their current dating partner or, if not currently involved with someone, to describe a typical first date. This last condition was included to broaden generalizability of the findings. After reading or composing the scenario, participants completed the risk rating task described below with regard to heterosexual dating relationships.

MEASURES

Situational Risk Factors

Nine risk factors associated with acquaintance sexual aggression were chosen from a review of the research literature and preliminary interviews with 20 undergraduate women. These included man paying for dating expenses, man drinking, woman drinking, isolation of physical setting, man’s physical size, man making sexual comments and jokes such as “You got great tits”; verbal persuasion such as “Don’t you like me?”, physical pressure while fondling, and persistence after the woman tightens up or closes down. After reading one of the scenarios, participants were instructed to imagine how they would react if each of the following situations arose. Each risk factor was then presented along with a 7-point rating scale, for example, “Man pays for dating expenses” from 1 (spends a little) to 7 (spends a great deal). Because recognition of risk is not an “all or nothing” phenomenon, participants made three increasing levels of risk judgments, specifically at what point on the continuum they would feel on guard, really uncomfortable, and seriously at risk. These terms were chosen as a result of the preliminary interviews.

To aid in data reduction, principal components factor analysis using varimax rotation was conducted on each of the three sets of ratings separately. Each factor analysis consistently yielded two factors with eigenvalues greater than 1.0, accounting for 50%—70% of the variance. One variable, man paying for dating expenses, did not load on either factor and was dropped. In general, the first factor consisted of variables that indicated relatively discernible risk (henceforth referred to as clear risk) for sexual aggression, including sexual comments or jokes, verbal persuasion, physical pressure while fondling, and male persistence (refusing to take no for an answer). The Cronbach’s alphas were: on guard =.64, uncomfortable =.69, and seriously at risk = .71. The second factor consisted of variables that were somewhat ambiguous in their indications of risk, including male and female drinking, male size, and degree of isolation. Each of these variables could be seen as having both positive and negative elements; for example, drinking together is frequently viewed as a normal part of socializing. The Cronbach’s alphas were: on guard =.73, uncomfortable = .62, and seriously at risk = .74. Based on these factor analyses and alphas, these items were combined to form the clear risk and ambiguous risk scales.

Global Perception of Risk

Respondents were asked to assess their own and other college women’s likelihoods of encountering six types of unwanted sexual activities including fondling, verbal coercion to engage in activities both inclusive and exclusive of sexual intercourse, being physically restrained to engage in activities both inclusive and exclusive of sexual intercourse, and rape. Although the rape item overlapped with two of the others, it was included to replicate earlier work by Koss (1985) that has shown that some women do not identify forced intercourse by an acquaintance as rape. Respondents rated the likelihood that each unwanted activity would occur to her and to other women on a 7-point (not at all to extremely likely) scale. The ratings were averaged to form two scales: perception of risk for self (alpha = .90) and perception of risk for other women (alpha = .86).

Prior Victimization

The occurrence of both child sexual abuse and adult sexual victimization were assessed. Child sexual abuse was measured with a dichotomous item (yes/no) that asked whether the subject had experienced child sexual abuse by an adult or someone more than 2 years older than her before the age of 14.

Ten items derived from the Sexual Experiences Survey (Koss & Oros, 1982) were used to measure the frequency with which respondents had encountered sexual aggression since they had started dating.

The items included the following:

1. consensual intercourse;

2. man becoming so sexually aroused that he could not stop;

3. unwanted kissing or petting;

4. unwanted breast fondling;

5. unwanted genital touching;

6. unwanted intercourse after man said he would end the relationship;

7. intercourse after being given drugs or alcohol;

8. attempted, but not completed, intercourse with physical restraint;

9. intercourse with physical restraint; and

10. rape.

Frequency categories of 0; 1 to 2 times, 3 to 5 times, 6 to 9 times, and greater than 10 times were employed.

Principal components factor analysis on these items with varimax rotation yielded two factors with eigenvalues greater than 1.0, accounting for 61% of the variance. Scores were averaged to form two scales. The first (alpha = .81), labeled low severity, included four items: man could not stop because of his arousal, unwanted kissing/petting, unwanted breast fondling, and unwanted genital touching. The second (alpha = .66), labeled high severity, also included four items: man would end relationship, physical restraint when sexual intercourse either did or did not occur, and rape. The items concerning consensual sexual intercourse and intercourse after being given drugs or alcohol loaded on separate factors and were not included in these scales.

Drinking Habits

Participants were classified as drinkers or nondrinkers based on their self-reported drinking habits. These included the average number of drinks consumed on each day of the week over the previous month and over how many hours drinking had occurred. A drink was defined as 12 ounces of beer, 4 ounces of wine, or 1 ounce of hard liquor. Mean blood alcohol level (BAL) was computed using the participant’s self-reported weight in conjunction with these variables (Mathews & Miller, 1979). Those whose computed BAL for the previous month was no higher than 0.01 were classified as nondrinkers (N = 82); those with BALs above this level (N = 54) were classified as drinkers. Four respondents provided insufficient data and had to be dropped from this classification. Although the two groups did not differ in age, drinkers were more likely to be White than Asian. They were also more likely to be members of sororities and to have had consensual sex. Drinkers’ mean BAL was 0.04 (approximately half the legal limit for drunkenness).

Background Characteristics

Several items assessing participants’ demographic and other characteristics were asked including age, weight, year in school, ethnic identity, living arrangement, social activities, age of first date, and age of first intercourse.

RESULTS

ANALYTIC APPROACH

Data analyses were conducted in four stages. First, multivariate analysis of variance was used to examine whether perceptions of situational risk differed across type of dating situation. Second, participants’ ratings for each level of the two types of situational risk factors were compared by means of paired t tests to determine whether participants distinguished between the two types of situational risk. Third, hypotheses regarding global risk perceptions were tested. Their relationship to each level of the types of situational risk factors were examined by means of Pearson correlations. A paired t test was conducted to compare global risk perception ratings for self versus others. Finally, hypotheses related to the individual difference variables were tested. Pearson correlations were calculated between the prior victimization scales and each situational and global risk perception scale. Following are findings based on these analyses.

SITUATIONAL RISK FACTORS

Multivariate analysis of variance (MANOVA) was conducted on the three levels of clear and ambiguous risk factor scales using type of dating situation as the independent variable. This analysis was not significant (all ps > .05); therefore, all further analyses collapsed across type of dating situation.

Paired t tests were conducted on the means of the clear and ambiguous risk factor scales at each level of perceived risk to determine whether participants distinguished between the two types of risk factors. Results showed that only at the initial level of risk perception (on guard) did this occur, t(1,126) = 5.52, p < .0001. Specifically, women making judgments about being on guard needed a higher level of ambiguous than clear risk factors to detect any risk (see Table 1). Judgments of feeling uncomfortable or seriously at risk did not differ between the two types of risk factors. In general, women seemed to need the lowest presence of clear risk factors to feel on guard and a somewhat higher level of either clear or ambiguous risk factors to feel either uncomfortable or seriously at risk, as indicated by their relative means.

TABLE 1.

Mean Levels of Perceived Risk for Ambiguous Versus Clear Risk Factors

Risk Level	n	Risk Factor Type
		Ambiguous (SD)	Clear (SD)
On guard	127	3.25 (1.48)	2.62* (1.14)
Uncomfortable	124	4.07 (1.58)	4.01 (1.23)
Seriously at risk	124	4.53 (2.24)	4.82 (1.76)

^*p < .0001.

GLOBAL PERCEPTIONS OF RISK

Pearson correlations between the two scales for global perception of risk and for both the clear and ambiguous risk factors showed that the two types of risk perceptions were not, for the most part, significantly related (−.09 < rs < .11, ps > .10) across the three levels of risk perception. Of the 12 correlations computed, only three were significant: global risk for self with situational-ambiguous-on guard (r = .22, p < .05) and ambiguous-uncomfortable (r = .27, p < .01), and global risk for others and situational-ambiguous-on guard (r = −.26, p < .01). A paired t test on the means for perceived risk of sexual aggression for self and for others was conducted. The participants viewed themselves at considerably less risk than they saw other women, t(1,138) = 15.07, p < .0001, Ms = 2.98, SD = 1.32 versus 4.74, SD = 1.01.

INDIVIDUAL DIFFERENCES

Prior Victimization

Pearson correlations were calculated between the low and high severity victimization scales and the two types of risk perceptions—situational and global (see Table 2). At the initial risk perception level (on guard), having experienced either low or high severity sexual victimization correlated with the ambiguous risk factor scale. In other words, as severity of victimization increased, participants needed a higher level of ambiguous risk factors to feel on guard. The same trend was evident at the higher risk perception levels (uncomfortable and seriously at risk), but differed for type of victimization. That is, as high severity victimization experience increased, participants needed a higher level of ambiguous risk factors to judge themselves as uncomfortable or seriously at risk; however, as low severity victimization experience increased, participants needed a higher level of clear risk factors to make the same risk judgments. Both low and high severity victimization experience correlated with global perception of risk for self but not for other women. No significant relationships were found between child sexual abuse and either situational or global risk perceptions. However, child sexual abuse correlated moderately with both low severity (r = .25, p < .003) and high severity (r =.27, p < .001) adult victimization.

TABLE 2.

Correlations Between Prior Victimization Experience and Risk Ratings

Risk Factor Type	Severity of Victimization
	Low	High
Situational
On guard—Ambiguous	.18*	.24**
On guard—Clear	.15	.11
Uncomfortable—Ambiguous	.16	.22*
Uncomfortable—Clear	.25**	.17
At risk—Ambiguous	.06	.19*
At risk—Clear	.25**	.16
Global
Self at risk	.18*	.25**
Others at risk	−.05	−.01

^*p < .05.

^**p < .01.

Drinking Habits

A MANOVA was conducted on the clear and ambiguous risk factor scales using drinking classification (drinkers vs. nondrinkers) as the independent variable for each of the three levels of perceived risk. These analyses were not significant (ps > .05). A MANOVA was then conducted on the two specific risk factor items related to male and female drinking for each of the three levels of perceived risk. There were no differences between drinkers and nondrinkers in perceptions of male drinking as a risk factor at any level of perceived risk. However, at the on-guard level of perceived risk, drinkers needed a higher amount of female drinking to make a risk judgment than nondrinkers, multivariate F(3,132) = 2.67, p < .05. Although the same trend was evident at the other levels of perceived risk, the mean differences were not statistically significant (see Table 3 for means). A MANOVA conducted on the global risk perception scales for self and others showed that drinkers viewed themselves at a higher risk for sexual aggression than nondrinkers viewed themselves, but that nondrinkers viewed other women at a higher risk for sexual aggression than drinkers judged other women, multivariate F(2,135) = 8.66, p < .0001 (see Table 3 for means).

TABLE 3.

Mean Differences in Perceived Risk for Drinkers (N = 45) Versus Nondrinkers (N = 93) on Female Drinking Risk Factor Item and Global Risk Scales

Female Drinking	Drinkers (SD)	Nondrinkers (SD)
Risk level
On guard	3.33 (1.93)	2.40 (1.92)*
Uncomfortable	4.11 (2.29)	3.54 (2.12)
Seriously at risk	5.09 (2.78)	4.40 (2.86)
Global risk
Self	3.40 (1.31)	2.78 (1.30)*
Others	4.50 (.93)	4.86 (1.02)*

^*p < .001.

DISCUSSION

This study has provided us with important insights about the process through which women assess their risk for acquaintance sexual aggression. Women consistently appear to classify situational risk factors as clear or ambiguous across three levels. Ambiguous risk factors tend to be those that are commonly accepted as being a normal part of socializing with a man, such as being alone and drinking. In and of themselves, it is difficult for a woman to make a judgment that their presence constitutes a clear risk to her safety because, in her experience, most of the time they have not. In fact, over time she may have developed strong positive associations between these factors and her interactions with men and consequently look forward to them rather than avoid them. It is not surprising then that a much higher level of these risk factors is required before a woman makes a judgment that she needs to be on guard compared with those situations that are judged to constitute clearer risk (see Hock, Krohne, & Kaiser, 1996, for a discussion of difficulties in detecting threat-related meaning in ambiguous stimuli). Specific unwanted actions taken by a man proved to be more readily discernible or “clear” risk factors. Although unwanted actions by a man are not uncommon experiences and do not necessarily lead to assaultive behavior, such behaviors appear to carry more threat-related meaning, making it easier for women to encode these as risks for further escalation of unwanted sexual engagement.

Focusing on this initial level of risk perception may be crucial to developing effective interventions that aid women in exiting or proactively managing situations to better protect themselves. As might be expected, a lower level of both clear and ambiguous risk factors evoked feelings of initial guardedness compared with what was necessary for a woman to feel downright uncomfortable or seriously at risk. However, even these latter ratings clustered around the scale midpoint. These data may not speak entirely to whether women’s perceptions of risk in actual situations would be the same as their hypothetical judgments. However, these findings indicate that women recognize cues indicating when men’s behaviors deviate from acceptable social norms. A possible next step in helping women to behave in the interest of their own safety might be to develop interventions that start from this point. For example, programs can build on women’s general awareness of risk cues to develop habits of shifting into a more questioning mode when they encounter such cues. In any given situation the risk may not be great, but focused appraisal can assist women to assert their refusals early and effectively, thereby increasing the likelihood of avoiding a seriously distressing situation.

Other findings from this study are also important in understanding the larger cognitive framework within which judgments of risk are made. First, global perceptions of risk in general were not related to judgments of specific situational risk. This may indicate, as noted by Nurius (in press), that making these two types of judgments involves distinctly different processes: General knowledge about the risk of acquaintance rape does not necessarily translate into the belief that one is just as much at risk as the average college woman and, thus, does not necessarily translate into taking personal precautions. This is consistent with findings reviewed by Jeffery (1989) emphasizing differences in individual and population perspectives and data relative to risk. Although information about population risk (e.g., incidence data about acquaintance sexual violence, variables broadly found associated) is crucially important, most people do not think at this more epidemiological level. Moreover, although cumulative risk for women is high, the probabilistic risk of experiencing sexual aggression on any one given occasion such as a date is low. Add to this the psychological and interpersonal factors that bias individuals toward expecting and seeking out information consistent with a presumed context or preferred conclusion (Ditto, Jemmott, & Darley, 1988; Kunda 1987, 1990) and reasons for this gap between global and situational risk assessment become more evident.

It is notable that in thinking about one’s global risk, women appear to have a strong optimistic bias. That is, they view other women at substantially higher risk than they see themselves, rating themselves generally below the scale midpoint and other women above it. This may have implications for the types of preventive actions women take, or whether they take any at all. Women who see their chances of experiencing acquaintance sexual aggression as only slight would probably not be motivated either to plan ahead about precautions to take such as deciding in advance a reasonable drinking level, or to look for threat cues once in a situation. Instead of maintaining a vigilant skepticism about a man’s motivations, a woman who perceives herself at little risk of being sexually victimized may erroneously assume that the man will behave in an acceptable manner.

Norris et al. (1996) found that although women were generally reluctant to admit any personal vulnerability, they were able to express clearly protective measures they could and would take toward other women. Perhaps one means of addressing how to overcome the barriers to safety that optimistic bias presents without forcing women to relinquish its positive psychological effects, would be to appeal to women’s tendency to look out for others. Acquaintance rape prevention efforts might incorporate discussions encouraging women to develop ways of sending signals or performing other interventions that would promote a feeling of group empowerment and camaraderie rather than of individual isolation in coping with an ambiguous but potentially threatening situation. In a related vein, recent findings (Stapel, Reicher, & Spears, 1994) indicate that risk information related to a group with whom an individual shares membership or identity significantly increases cognitive availability and judgments of personal risk, further underscoring the potential value of women’s referent and friendship groups in prevention interventions.

The relationship between prior victimization experience and perceptions of situational risk is complex. Women who had previously experienced sexual victimization as an adult needed a higher level of both clear and ambiguous risk factors at the initial risk perception level (being on guard) than nonvictimized women before making a judgment that they would be at risk. The same trend is evident at the higher risk perception levels (uncomfortable and seriously at risk), but differs somewhat for level of previously experienced victimization. Women who had experienced low level victimization needed evidence of a higher level of clear risk factors before rating themselves as uncomfortable or seriously at risk. Having never experienced severe sexual assault, these women may not have felt justified in making a higher level risk judgment than those who had. It is also a matter of concern that even women who had experienced serious victimization in the past needed a higher level of ambiguous risk factors before making a higher level risk, perception judgment. Perhaps the inability to recognize risk and thereby act on it, may be one reason they have experienced more sexual aggression in the first place. Prevention programs could focus on risk perception training as one avenue to help women avoid sexual assault.

This is somewhat disturbing because one might have thought that previously victimized women might be more wary than the nonvictimized; however, having been victimized does not apparently result in a lower threshold for perceiving risk. Nonetheless, previously victimized women saw themselves at increased risk in general for sexual aggression relative to nonvictimized women’s global self-risk perception. This increase in self-risk perception did not generalize to their perceptions of other women. Thus, rather than maintaining a high level of optimistic bias about their global risk, as women in general do (Hoecker & White, 1995; Nason, 1995), previously victimized women seem to have a diminished level of optimism in this regard. Having been victimized in the past, they appear to revise upward their views that this could happen again, although their mean level of expectancy remains relatively low. This increase in perception of global threat does not appear to lead to an increased sense of wariness concerning potentially risky situations.

Both alcohol consumption and an increased number of sex partners have been discussed as possible mediating variables in the relationship between prior and subsequent victimization (Harney & Muehlenhard, 1991; Mandoki & Burkhart, 1991; Nurius, Norris, Dimeff, & Graham, 1996). It is possible that both of these factors may have played a role in increasing the likelihood of initial dating victimization, and both may be related to a history of childhood sexual abuse as well (Browne & Finkelhor, 1986). Whereas an increased number of sex partners may increase a woman’s chances of encountering a sexually aggressive man (Koss & Dinero, 1989), alcohol consumption in particular may reduce her ability to detect threat cues (Nurius, in press; Nurius & Norris, 1996) and impair her ability to react in a decisively self-protective manner (Nurius et al., 1998). Childhood sexual victimization in this study was not related to either specific or global risk perception; however, measuring this variable dichotomously may have limited our ability to detect its relationship to risk perception. Nevertheless, it was correlated with adult victimization and experiencing even low level sexual victimization since the commencement of social dating was in turn associated with a decreased ability to detect potential risk.

Despite much prevention work targeted at emphasizing the role of alcohol as a risk factor for acquaintance rape, the current findings indicate that women in general view alcohol as, at most, an ambiguous risk factor. Furthermore, those women whose current drinking habits identify them as drinkers required a higher level of female drinking compared with nondrinkers before making a risk judgment at any level—on guard, uncomfortable, or seriously at risk, although there were no differences in their judgments of male drinking as a risk factor. This has implications for victimization. If women believe that they can drink at a high level before being at increased risk for sexual aggression, they may engage in behavior that unfortunately makes them particularly vulnerable. Perhaps realistically then, drinkers saw themselves at higher risk for sexual aggression in general than nondrinkers did, indicating that positivity bias had shrunk for drinkers compared with nondrinkers. It is also notable that post hoc chi-square analyses showed that drinkers were significantly more likely to have experienced high severity sexual victimization, and the same trend was evident for low severity victimization. However, it was not possible to address whether previous victimization had been alcohol-related. Future research using multivariate models should address whether alcohol consumption itself decreases the ability to perceive risk or whether deficient risk perception among drinkers may be solely traceable to greater prior victimization experience.

This study was limited in other ways as well. Although college women, being at high risk for acquaintance sexual assault (Koss et al., 1987), are an appropriate group for investigation in this area, it is not known to what extent our findings would generalize beyond the college population. In addition, subjects in this study made judgments about hypothetical situational factors disconnected from one another. The onset of an actual sexual assault would present many cues simultaneously. Thus, we do not know whether the judgments women might make when faced with combinations of risk factors during an actual dating situation would be identical to those made about specific individual situational risk factors.

These findings are both encouraging and discouraging. They are encouraging because inroads have been made in understanding how women make judgments about their risk for acquaintance sexual aggression. On the other hand, the findings concerning both prior victimization and alcohol highlight the complexity of understanding this phenomenon and point to the need for further work to develop effective, well-targeted interventions. First, how can women’s optimistic bias regarding sexual aggression (“It only happens to others”) be decreased without having to experience sexual aggression in the first place? Second, how can the message that alcohol is a serious risk factor for aggression be communicated convincingly? Of course, understanding how women perceive risk and which situational or background factors might influence their judgments is not sufficient for developing effective rape prevention programs. It is also essential to understand how the perception of risk can be transformed into effective defensive action. Are women who perceive risk early in a situation more likely than others to extricate themselves before a serious assault has occurred? How is the perception of risk related to specific responses to a man’s sexually aggressive actions?

Although attention to acquaintance sexual aggression has grown in recent years, many of the difficulties associated with assessing and protectively responding to risk are by no means unique to this arena. A recent issue of the American Psychologist (September, 1996), for example, dedicated several articles to prediction of violence. Although important progress has been made over the years, these articles underscore the enormous complexity in assessing this dimension of human conduct, and the limited range that even our best assessment techniques can achieve. It is little wonder that women struggle with assessing such conduct from men that are part of their everyday world. Implications from our findings primarily guide what Schopp (1996) refers to as inquisitive prescriptions—efforts to engage in information gathering and monitoring that support an internal warning system; that is, detection of risk signs and, as meterologists would say, “getting out of harm’s way” (see Monahan & Steadman, 1996). A parallel set of efforts have to do with preventive prescriptions that trigger social interventions intended to prevent harmful conduct (Schopp, 1996). This could include a broad range of possibilities, including efforts targeting men and the cessation of aggression against women. In contemporary life, both sets of efforts are needed. Further investigation of risks as women see them within the contexts of their individual lives and relationships will continue to provide important insights into how women can better arm themselves against acquaintance sexual aggression.

Biographies

Jeanette Norris is a research scientist at the Alcohol and Drug Abuse Institute, University of Washington. Her current research interests include cognitive mediation of women’s risk perceptions associated with and responses to sexual assault, as well as the relationship between alcohol consumption and sexual violence.

Paula S. Nurius is a professor of social work at the University of Washington. Her current research interests include self-concept functioning and cognitive mediation of stressful events, including sexual violence.

Thomas L. Graham is an assistant professor of social work at Portland State University. His research interests include alcohol and interpersonal violence.