Taasa.org

The community response to rape: victims' experiences with the legal, medical, and
mental health systems
Publisher: American Journal of Community Psychology. Volume: 26. Issue: 3 This research examined how the legal, medical, and mental health systems respond to theneeds of rape victims. A national random sample of rape victim advocates (N = 168)participated in a phone interview that assessed the resources available to victims in theircommunities. as well as the specific experiences of the most recent rape victim withwhich they had completed work. Results from hierarchical and iterative cluster analysisrevealed three patterns in victims' experiences with the legal, medical, and mental healthsystems. One group of victims had relatively positive experiences with all three systems,a second group had beneficial outcomes with only the medical systems, and the finalgroup had difficult encounters with all three systems. Multinominal logistic regressionwas then used to evaluate an ecological model predicting cluster membership.
Community-level factors as well as features of the assault and characteristics of thevictims predicted unique variance in victims' outcomes with the legal, m edical, andmental health systems. These findings provide empirical support for a basic tenet ofecological theory: environmental structures and practices influence individual outcomes.
Implications for ecological theory and interventions to improve the community responseto rape victims' needs are discussed. KEY WORDS: rape victims; community response; rape victim advocates. When women go public with their stories of rape, they place a great deal of trust in oursocial systems as they risk disbelief, scorn, shame, punishment, and refusals of help(Madigan & Gamble, 1991). How these interactions with system personnel unfold canhave profound implications for victims' recovery (Kerstetter, 1990). If women receive theservices they need, and are treated in an empathic and supportive manner, then our socialsystems can work as effective catalysts for healing (Estrich, 1987: Fairstein, 1993;Golding, Siegel, Sorenson, Burnman, & Stein, 1989; Madigan & Gamble, 1991; Parrot,1991; Russell, 1990; Warshaw, 1988; Wyatt, Notgrass & Newcomb, 1990). Conversely,if victims do not receive the services they want and are treated in an insensitive manner,then interactions with community personnel can magnify feelings of powerlessness,shame, and guilt for rape victims (Feldman-Summers & Palmer, 1980; Madigan &Gamble, 1991). These negative experiences have been termed "the second rape" or"second ary victimization" (Madigan & Gamble, 1991). Analysis of these interactionsbetween victims and social systems may uncover ways to promote a community responseto rape that is psychologically beneficial to victimized women. Community resources for rape victims are often piecemeal and uncoordinated as differentsystems perform different functions. For example, victims go to the hospital for the rapeexam and evidence collection; to the police station to meet with a detective; to the state'sattorney to discuss prosecution; to the rape crisis center for information and crisiscounseling; and to many other agencies as well. Not surprisingly, research in this area islikewise diffused across disciplines and methodologies. Typically, each system-legal,medical, mental health--is studied in isolation. From the perspective of the victims,however, these lines of demarcation may not be as distinct, meaningful, or useful. Thisflurry of activity is about one event in their lives, one trauma that is then parceled outamong many for attention. Focusing on how the legal, medical, and mental healthsystems respond to victims' needs increases our understanding of victims' experienceswith community systems. By taking this more holistic view o f how communitiesrespond, we can begin to see what victims experience and evaluate how well our socialservices are responding to their needs. A primary obstacle in the development of this holistic view of victims' experiences hasbeen the recruitment of survivors who have sought community services. Rape victims areoften difficult to identify and the emotional trauma of sexual assault often leaves themreluctant to discuss their experiences with researchers. An alternative sample to consideris rape victim advocates. Most communities in the United State have a rape crisis centerwith staff members who work as community-based advocates, helping victims negotiatethe process of interacting with each community system (Webster, 1989). Advocatesexplain the services that are available to victims, determine what victims want from eachsystem, and then work to bring about outcomes consistent with their needs. Through thisprocess, rape victim advocates become privy to a great deal of information about bothrape victims' needs and how service systems respond to victims. In this research, a national random sample of rape victim advocates was recruited toaddress two issues. First, this study sought to identify patterns of victims' experiencesacross multiple community systems. Advocates were asked to describe the most recentsexual assault case they had completed, and what actions were taken in that case by thelegal, medical, and mental health systems. In describing these interactions, threedimensions were considered: (a) What services were provided to victims by eachsystem?; (b) Did those outcomes fit with victims' needs (i.e., did the system respond in amanner that was consistent with victims' wishes)?; (c) How readily available were thoseservices? This information was used to develop clusters profiling different patterns ofcommunity response to sexual assault. The second goal is to determine what factorspredict different experiences: Which victims receive which services? A multi-level modelpredicting cluster membership was evaluated. Presented first is an overview of existingliterature on rape victims experiences' with social systems that examines research onservice delivery in each major community system. Then, extending this work, anecological model predicting victims' experiences with the legal, medical, and mentalhealth systems is described and evaluated. RAPE VICTIMS' EXPERIENCES WITH COMMUNITY SYSTEMS Prosecuting a rape is a complicated process, which starts with reporting the assault to thepolice. This initial report may be given to a detective for a more detailed investigation. Insome jurisdictions, this report/investigation is automatically forwarded to the prosecutor,but in others, the police decide whether to forward the report. The prosecutor thenchooses whether to authorize an arrest and press charges-either for the original charge ofsexual assault or a lesser offense (e.g., simple assault, reckless endangerment). Thesecharges may be dropped later, but if not, the accused rapist has the choice of pleadingguilty to the charged offense, or, if a bargain has been struck, to a lesser offense, or goingto trial. If he is convicted at the trial, the judge may choose either probation or jail aspunishment. With a system this complex, it is to be expected that some cases will slip through thecracks, and indeed over half of reported rapes are filtered out of the criminal justicesystem (Galvin & Polk, 1983; LaFree, 1980). Which cases proceed, and which arefiltered out, is influenced by multiple factors. At the community level, the resourcesallocated to address sexual assault and the coordination of those services increaseprosecution efforts (Fairstein, 1993; U.S. Department of Justice, 1994). The type of rapeis also significant, as several studies have found that stranger rapes are investigated morethoroughly and are less likely to be filtered out of the system than nonstranger-rape cases(Fairstein, 1993; Finkelhor & Yllo, 1985; Kerstetter, 1990; Madigan & Gamble, 1991;McCahill, Meyer, & Fischman, 1979: Russell, 1990). Assaults that involve the use of aweapon and result in physical injuries to the victim are more likely to be pursued (LaFree,1981: Kerstetter, 1990; Rose & Randall, 1980). Characteristics of the victim alsoinfluence case disposition. Victims who are perceived as less credible are more likely tohave their cases rejected for prosecution (Rose & Randall 1989). In a similar vein,Madigan and Gamble (1991) suggested that system workers distinguish between "goodvictims" and "bad victims." "Good victims" show visible, expressive signs of trauma(e.g., crying) and are receptive to help from system personnel. Consequently, they mayreceive more help than victims who do not show as much visible distress. There are four reasons why victims may need medical attention following a rape. First,forensic evidence can be collected (e.g., semen, blood, and/or hair/fiber/skin samples).
Second, a medical examination is helpful to detect and treat physical injuries from theassault. Third, victims often want information and testing for sexually transmitteddiseases (STDs), and some hospitals administer of a preventative dose of antibiotics totreat any STDs that might have been contracted in the assault. Finally, althoughpregnancy from rape is rare (5% of the time, Beebe, 1991: Koss, Woodruff, & Koss,1991), it is a concern for many victims, and some hospitals administer the morning-afterpill to prevent pregnancy (i.e., ethinyl estradiolnorgestrel [Ovral]). There has been very little research examining whether hospitals offer these services to victims. The National Victim Center's (1992) survey of female survivors of sexual assaultindicated that 60% of victims were not advised about pregnancy testing or how to preventpregnancy. Although 43% of the women were concerned about contacting HIV from theassault, 73% were not given information about testing for exposure to HIV Another 40%were not given information about the risk of contracting other STDs. Campbell andBybee (1997) found that emergency room personnel rarely provide rape victims with themorning-after pill to prevent pregnancy (38% of the time). These findings suggest thatthere is some inconsistency in what information and services are offered to victims duringthe medical exam. Mental health workers may be called upon to help both victims and those close to themwho are also traumatized by the rape, including her family, friends, and/or husband(significant other). This assistance could be limited to providing information about rapeand its effects, or could extend to short-term or long-term counseling or support groups.
Most research on psychological services for rape victims has sought to identify effectiveforms of therapy (e.g, Foa, Rothbaum, Riggs, & Murdock, 1991; Frank & Stewart, 1984;Frank, Stewart, Dancu, Hughes, & West, 1988; Koss & Harvey, 1991; Resick, Jordan,Girelli, Hutter, & Marhoefer-Dvorak, 1989; Resick & Schnicke, 1992) and assess theprevalence of posttraumatic stress disorder (PTSD) in rape victim populations (e.g., Foa,Steketee, & Olasov. 1989; Norris, 1992; Rothbaum, Foa, Riggs, Murdock, & Walsh,1992), rather than documenting whether these services are provided to victims and theirfamilies. Some light was shed on this issue by Forman and Wadsworth's (1983) study ofcommunity mental health centers (CMHC), which found that over 75% of CMHCs intheir sample offer these services to victims and their families. Other work by Campbell,Baker, and Mazurek (1998), Gornick, Burt, and Pittman (1985), and Harvey (1982a,1982b, 1985) revealed that rape crisis centers have begun to offer these mental healthservices. King and Webb's (1981) survey of 24 rape crisis centers found that all victimswho requested counseling received it, and 62% of the victims had at least one follow-upcontact with center staff. AN ECOLOGICAL MODEL PREDICTING RAPE VICTIMS' EXPERIENCES WITHCOMMUNITY SYSTEMS The existing literature on service delivery for sexual assault victims suggests thatmultiple factors influence the response of the legal system, but less is know about theresponse of the medical and mental health systems. In effort to understand multiplecommunity systems and the multiple influences impacting system response, a multilevelmodel was evaluated in this research. Ecological theory with its attention toperson-environment fit and ecological settings can provide one framework through whichto theorize and research victim-system interaction that spans individual and systemicfactors (Kelly, 1966, 1968, 1971; Trickett, Kelly, & Vincent, 1985). Thus, the focus ofthis research was on the fit between victims' needs and system response, and howenvironmental factors impact this fit. The proposed ecological model expands previous rape research in two areas. First, asnoted previously, much of the existing research has focused on the legal response tosexual assault, and in order to further a more holistic picture of victims' experiences, theaim of this study was to predict outcomes in all three systems simultaneously (legal,medical, and mental health). Whereas there are multiple ways to define victims'outcomes, this model focuses on the ecological conception of person-environment fit:Did the system respond in a manner consistent with victims' needs? A "good" outcomefor one victim may be different from another as their needs vary. Focusing on the conceptof fit defines good outcomes as those consistent with individual victims' needs. Second,several community-level, rape-related, and individual-level factors have beendemonstrated to affect the legal system response. This study expands this literature byconsidering the impact of such variables across all three community systems (lega l,medical, and mental health), as well as examining the effects of untested variables, suchas victims' use of alcohol and demeanor when interacting with social system personnel. Two community-level factors are explored in this model. First, Kelly (1966, 1968, 1971)and other ecological theorists have suggested that the quantity and quality of resourcesavailable to address a social problem (e.g., staffing resources, financial support) willaffect service delivery. Second, interorganizational coordination has been suggested toimprove service delivery (Agranoff & Pattakos, 1979; Baker & O'Brien, 1971; Tausig,1987: Turner & TenHoor, 1978). For instance, some communities have a coordinatedresponse to sexual assault that brings multiple service providers together to assist victims(e.g., Sexual Assault Response Teams). How a system responds to victims may be afunction not only of what resources are available, but also how embedded that system iswithin a network of social agencies. Several characteristics of the assault itself may influence how social systems respond tovictims. Four features of the rape itself are considered in this model: (a) type of rape(stranger or nonstranger); (b) use of a weapon; (c) presence of physical injuries; and (d)whether the rape occurred while the victim was under the influence of alcohol. There hasbeen growing interest in the role of alcohol in rape (e.g., Abbey, Ross, McDuffie, &McAuslan, 1996: Richardson & Hammock, 1991), and as such, this variable is includedin the model along with previously established predictors. Whereas there are numerous characteristics of victims that may influence systemresponse, three broad-based characteristics are considered: race/ethnicity, social class,and victims' demeanor when interacting with system personnel. Several studies in thevictimology literature (as well as research on other social problems) suggest thatmembers of disadvantaged or stigmatized groups (i.e., non-White, lower socioeconomicstatus) receive differentially worse treatment by social systems (Davis & Proctor, 1989: Gordon-Bradshaw, 1988: Mama, 1989: Pinderhughes, 1989: Wyatt, et al., 1990).
Furthermore, Madigan and Gamble (1991) suggested that some women are perceived bysystem personnel as "good victims" (i.e., they are visibly distressed and receptive tohelp), and hypothesized that women who behaved in this manner may receive more helpthan those who do not. This assumption is explicitly tested in this model. A two-step process was used to select the national random sample of rape victimadvocates. First, using a national directory of services for sexual assault victims(Webster, 1989), 759 agencies that provide advocacy services to rape victims wereidentified: 390 free-standing rape crisis centers and 369 combined rape crisis-domesticviolence programs. An a priori power analysis indicated that 165 participants would beneeded to have power = .80 ([alpha] = .05), assuming a medium effect size ([f.sup.2][standardized effect size] = .15, equivalent to a multivariate [R.sup.2] = .40 andmultiple-[R.sup.2] = .13) (Cohen, 1988). A random sample was selected, stratifying foragency type. In the second step, the directors of the randomly selected agencies were contacted bytelephone. They were asked if they had paid or volunteer staff who providecommunity-based advocacy services to adult rape victims, and if so, to provide the nameof the advocate who provides the most direct-advocacy work at that agency. If thedirector stated that no one provided these services, the agency was removed from the listof target agencies and a randomly selected replacement from the same type of agency wasdrawn. A total of 213 agencies were screened for participation in this study, and 177 wereeligible. (3) From this pool of eligible centers, 168 advocates participated in the interview (95%response rate). The advocates were on average 37.25 years old (range 22-65). The majority of the samplewas White (88%) (5% African American, 4% Latina, 2% Native American Indian, 1%Asian American, and 1% Arabic American). Most of the advocates had at least abachelors degree (74%). The advocates had been working in their current position at therape crisis centers for an average of 5.28 years (range 5 months-20 years). As mentioned previously, the advocates were asked to describe the most recent adultsexual assault case they had completed. The victims were, on average, almost 10 yearsyounger than the advocates (M age 28.35 years, range 17-78). Most of the victims wereWhite (71%) (14% African American, 9% Latina, 4% Native American Indian, and 2%Asian American). Most of the victims had at least a high school degree (60%). Over onethird of the victims were described by the advocates as working class (35%), with 25%described as lower class and 26% as middle class. Only 14% were described as abovemiddle class. Consistent with previous research, most of the rape survivors in this sample wereassaulted by someone they knew (acquaintance, date, partner) (76%), and most wereraped by a single assailant (90%). Whereas almost all of the women were subjected toforced vaginal penetration (95%), some experienced anal rape (20%), oral rape (23%).
and/or rape by a physical object (14%). Some women, therefore, experienced multipleforms of rape. In this sample of victims, women were raped on average 1.89 times by theassailant in the assault with a range of 1-20 times. Forty-one percent experienced nophysical injuries from the attack, but 48% experienced some bruising, 23% were cutduring the assault, and 19% experienced some type of head injury, such as a blow to thehead and/or broken blood vessels in the eyes and face from being choked. Most of thewomen did not have a weapon used against them in the assault (71%), and most were notunder the influence of alcohol (66%). Interviews with advocates were conducted by phone with a mean duration of 1.34 hours(SD = 27 minutes, range 40 minutes-3.25 hours). The interviews were conducted by theauthor and 10 undergraduate research assistants, who received course credit for theirparticipation in the project. To assess test-retest reliability, 25% of the completedinterviews were randomly selected and the advocates were recontacted by a differentinterviewer to repeat only the questions assessing rape victims' experiences with themedical system. Test-retest reliability was .95. To assess interrater reliability, a randomselection of 25% of the phone interviews were listened to by a second interviewer to codethe entire interview. Interrater agreement for the entire questionnaire was 95%, whichwas corrected for chance agreement ([kappa] = .86). The measures to assess the constructs in the proposed model were created for this studyby the author with input from two focus groups of rape victim advocates (10 advocates ineach group). For the community-level factors, two scales were created to measureresources available for rape victims and the coordination of those services. Therape-related and individual-level factors were assessed with single items in the interviewprotocol. Community-Level Variables. To assess the resources within each community, theadvocates were asked what services were available in their communities (15 total): policesex crimes unit, police rape protocol, prosecutor sex crimes unit, prosecutorvictim-witness program, prosecutor court accompaniment program. information providedat the hospitals on pregnancy, information provided on STDs, morning-after pill. STDpreventive treatment, scheduling follow-up medical visit, crisis intervention therapy,short-term rape-related therapy, long-term rape-related therapy, rape-related counselingservices for women with special needs (e.g., women with disabilities), and support groups. (4) The Community Resources scale was created by summing number of servicesavailable ([alpha] = .63). The Community Coordination of services scale summed thenumber of programs in a community that brought together the various agencies servingrape victims to streamline their services (e.g., Sexual Assault Response Teams). Theinternal con sistency for this 10-item scale was .64. (The Community Resources scale andthe Community Coordination scale were based on items that were dichotomously coded,so it is to be expected that their alphas will be somewhat low.) Rape-Related Variables. The rape-related variables in the model were assessed withindividual items from the interview protocol. For the analyses, Type of Rape was codedto distinguish stranger rape and nonstranger rape (acquaintance, date, and marital). Use ofa Weapon and Physical Injuries were coded to reflect presence/absence of a weapon andthe presence/absence of physical injuries to the victim. Finally, Victims' Use of Alcoholat the time of assault was coded to reflect use of alcohol/no use of alcohol. Individual-Level Variables. The individual-level variables in the model were alsoassessed with single items in the interview protocol. For the analyses, Victims' Race andAssailants' Race were coded to reflect persons who were described as White/Caucasianand those that were described as people of Color. Match in Race was coded to distinguishinter- and intraracial rapes. Victims' SES was similarly coded to distinguish betweenthose described as lower class from those described as working, middle, or upper class.
Assessment of Victims' Demeanor was operationalized using a definition of "goodvictim" behavior suggested by Madigan and Gamble (1991). The advocates were asked ifthe victim was crying or showing some other obvious, expressive signs of distress, andwas receptive to help and suggestions from system personnel. This variable wasdichotomously coded for the analyses. Dependent Variable in the Model (Victims' Experiences) To describe the responses of the legal, medical, and mental health systems to victims'needs, three dimensions of information were collected: (a) what actions were taken byeach system; (b) did those actions taken (or those actions not taken) fit with what thevictim wanted from the system (i.e., did the system respond in a manner consistent withvictims' needs); and (c) how readily available services were. Thus, for each action thatcould have been taken by each community system, three pieces of information weregathered: Was the action taken? (scale: 1 = yes; 0 = no); Did that outcome fit with whatthe victim wanted? (scale: 1 = yes; 0 = no); How much, if any, advocacy was needed tobring about that outcome (scale: 0 = none; 1 = a little; 2 = some; 3 = quite a bit; 4 = agreat deal). For each system (legal, medical, and mental health), three scales wereconstructed: System Action, System Fit, and System Advocacy. To create the scalescores, the number of "Actions," "Fits," and "Advocacies" were summed withi n eachsystem. Higher scale scores reflect more actions taken, higher consistency with victims'wishes, and higher levels of intervention employed to bring about each outcome. Internalconsistencies (a) for these scales were Legal Action =.77, Legal Fit = .76. LegalAdvocacy = .73, Medical Action =.68, Medical Fit =.57. Medical Advocacy =.65. MentalHealth Action = .75, Mental Health Fit = .38. Mental Health Advocacy = .61. The intercorrelations among these nine scales was generally quite low except between theMental Health Action and Mental Health Advocacy scales (r =.72, p < .01). To addressthis potential problem of multicollinearity, the Mental Health Advocacy Scale wasdropped for the analyses and only the Mental Health Action and Mental Health Fit scaleswere used. This high correlation suggested that to receive mental health services, someadvocacy was necessary. Complete psychometric information for all scales can beobtained from the author. Two issues were examined in this research. First, rape victims' experiences with the legal,medical, and mental health systems were explored, focusing not only on services offeredbut also on whether those outcomes fit their needs and the intervention required by theadvocates. A combination of hierarchical and iterative cluster analyses was used toidentify patterns of system responses that summarize these experiences. The secondquestion was whether these patterns of experiences could be successfully predicted by theproposed ecological model. Rape Victims' Experiences With Community Systems Cluster analysis was performed in two stages on the eight remaining scales assessingvictims' experiences with community systems: Legal Action, Legal Fit, Legal Advocacy,Medical Action, Medical Fit, Medical Advocacy, Mental Health Action, and MentalHealth Fit. The scale scores were standardized prior to clustering to account for theirdiffering variances (Aldenderfer & Blashfield, 1984; Rapkin & Luke, 1993). First, initialgroupings were derived through hierarchical clustering using Ward's Method and squaredEuclidean distances as the measure of proximity. Second, the centroids of these initialclusters were submitted to an iterative clustering procedure (K-means) to refine finalcluster membership and reduce the incidence of cluster misassigment that is commonwith agglomerative methods (Blashfield & Aldenderfer. 1988; Mowbray, Bybee, &Cohen, 1993). The iterative procedure used the initial centroids to maximize within-groupsimilarity and between-group difference. From this two-step procedure, a three-clu stersolution emerged. The reliability/stability of this solution was examined using a split-halftest (Luke, Rappaport, & Seidman, 1991). Half of the victims' data were randomlyselected and the cluster procedures described above were employed. The samethree-cluster solution emerged from this split-half test, suggesting that the clustersrepresented a stable organization of the data. Figure 1 presents the profile of scores found in the first cluster. To aid in interpretation,each cluster was given a short name to summarize its pattern of victims' experiences.
Cluster 1 ("Approaching Justice") contained 53 (32%) women, and was characterized bya high number of actions, high fit, but with low advocacy across all three systems. Thesevictims received a relatively large number of services, which was what they wanted, andthe advocates did not need to intervene to bring about those outcomes. For example, mostof the cases in this cluster progressed all the way through the criminal justice system. In89% of the cases in this cluster, the prosecutor issued some type of charge against the assailant, and 39% of the cases ended in a plea bargain and 28% were convicted at trial.
The outcomes of the trials and plea bargains were not always desirable (only 66% of thetime was the outcome consistent with the victims' wishes), but the assailants in thiscluster were most likely to receive some jail tim e for their crime (53%). In the medicalsystem, these women also received many of the services they wanted. Most victimsreceived information about pregnancy and STDs (86 and 92%, respectively), and 54%were able to obtain the morning-after pill. Almost all of them received a preventive doseof antibiotics to treat any STDs that might have been contracted in the assault (94%). Thewomen in this cluster also received many of the counseling services they wanted forthemselves and those close to them. Over half received either short-term or long-termcounseling (89 and 65%, respectively), and almost half of the victims' family or friendsreceived counseling (40%). It is important to note that many women in this cluster stillhad difficult interactions with system personnel, and had outcomes that were inconsistentwith their needs. But this first cluster described the best outcomes victims received in thisstudy. The second cluster contained the largest portion of the sample with n = 65 (39% of thesample). Figure 2 shows the profiles in this cluster, which was named "One SavingGrace." The women in this group did not have uniform experiences across all threesystems. First, their legal involvement was characterized by low actions, low fit, andfairly low advocacy. In other words, these cases did not progress very far in the legalsystem, but the victim did want to pursue prosecution. There was a marked misfitbetween what the system did and what the victim wanted. Most often, these cases werefiltered out in the early stages of processing, usually by the police deciding not to forwardtheir reports to the prosecutors. For example, only 43% of the police reports wereforwarded to the prosecutor (with only 43% of the women reporting that this outcomewas consistent with their wishes) and only 8% of the cases in this cluster were charged bythe prosecutors. None of the rapists in this cluster received jail time. The adv ocatesusually did not intervene to try to push these cases forward. One advocate summarized anexplanation for this lack of intervention that was similar to one voiced by manyadvocates: You could tell it (the case) was going to go no where and nothing I could do wouldchange that. The police were adamant. I would have had better luck banging my headagainst a brick wall. Besides, you have to think of the next victim you'll be working with.
You don't want to anger the police so badly on one case that it may hurt the next victimwho comes through. By contrast, the victims in this second cluster had better experiences with the medicalsystem. Although they received fewer services than the women in the first cluster, whatthey received was what they wanted. The advocates did not feel they had to intervene tosee that the victims' needs were being met. For example, most of the women receivedinformation about pregnancy and STDs (73 and 73%, respectively), but only 39% wereable to obtain the morning-after pill. Most of them received a preventive dose ofantibiotics to treat any STDs that might have been contracted in the assault (77%). The response of the mental health system was mixed in this second group. Most womendid not receive short-term or long-term counseling (which was somewhat consistent withwhat they wanted), but their family or friends often did not receive information aboutrape and its effects (which was inconsistent with what victims wanted; 32% receivedinformation, which only 25% of the victims stated was consistent with what theywanted). Finally, the third cluster was characterized by very high yet ultimately unsuccessful levelsof advocacy (see Fig. 3 for profiles of this cluster), and was named "Exercises in Futility"(n = 50, 29%). For these women, their involvement with the legal system wascharacterized by fairly high action, fairly high fit, and very high advocacy. These casesprogressed quite far in the criminal justice system (i.e., they were not dismissedimmediately as in the second cluster). But, these cases did not go as far, or as well, as thecases in the first cluster. At every step along the way, the advocates had to intervene tokeep these cases progressing. But, in the final stages of court processing, many casesunexpectedly fell apart--as one advocate described it, it was an "11th-hour catastrophe."For example, 70% of these cases were charged by the prosecutor, but only 39% of thewomen reported that the outcome of that charge was consistent with what they wanted.
The cases were proceeding (with the encouragement of the a dvocates), but in the finalstages, something went wrong: A case was dismissed a few days or hours before trial, aplea was struck that reduced the charges to misdemeanors such as simple assault, recklessendangerment, or "terroristic tendencies." Thirty-four percent of the assailants pled guilty(28% were to a reduced charge), 12% were convicted of the original charge, and 34%received jail time. The advocates were unable to prevent or reverse these negativeoutcomes. A somewhat similar pattern of results was found in the medical system: very low numberof services, even lower fit, and very high advocacy. These women wanted far moreservices than they received, and the advocates were largely unsuccessful in obtainingthese services for the victims. Only 48% of the women received information aboutpregnancy, and 34% received information about STDs. Most of the women were not ableto obtain the, morning after pill (11%), and only 34% of the women stated that this wasconsistent with what they wanted. Over half of the women, however, did receivepreventive antibiotic treatment for STDs (66%). In the mental health system, many of these women received some counseling (84%short-term; 37% long-term), but only about half of the time did their family or friendsreceive desired information about rape and its effects (51%). It appeared that for thewomen in this cluster, their families and friends wanted to learn more about rape, but thisassistance was largely unavailable. An Ecological Model Predicting Victims' Experiences The second question explored in this research was whether rape victims' experiences withcommunity systems could be successfully predicted by the proposed ecological model.
Given that the dependent variable was cluster membership, a categorical variable with three levels, multinominal logistic regression was used to evaluate the model (Hosmer &Lemshow, 1989). Table I presents the intercorrelations among the predictor variables. (5) To evaluate this model, four sets of analyses are needed. First, the overall fit of the modelmust be considered with the likelihood ratio (LR) statistic. Second, the significance of theindividual predictors to differentiate cluster membership must be considered using oddsratios and Wald tests. Third, the prediction success index and the percentage correctlyclassified address the degree to which the model successfully classified the cases intotheir correct cluster. Finally, because this is a multilevel model (community-level,rape-related, and individual-level predictors), the utility of each level must be examined(i.e., are all three levels necessary to predict cluster membership). A series of LR statisticstesting for the significant effect of each level to predict unique variance in the outcomevariable must be computed. The LR test for goodness-of-fit for the overall model was significant, indicating that themodel provided a good fit of the data: LR [chi square] (18, N = 168) = 71.03, p < .001.
Table II presents the results of the odds ratio and Wald tests. For an outcome variablewith three levels, two sets of contrasts are performed, as well an overall test for eachpredictor (Wald). In the first contrast, the women who had the best possible outcomes(Cluster 1: Approaching Justice) were compared to the women who had a positiveoutcome with only the medical system (Cluster 2: One Saving Grace). The results of theodds ratio tests indicated that women who were raped by a stranger without the use of aweapon were 15 times more likely than those raped by strangers with a weapon to be inthe One Saving Grace cluster, which was the group whose cases were dropped out earlyin the stages of legal processing. Similarly, women who were raped by someone theyknew without the use of a weapon were approximately 6 times more likely to be in theOne Saving Grace cluster. Taken together, these odds ratios suggest that being raped by anonstranger and/or the absence of a weapon placed victims in the cluster where caseswere dropped by the legal system. Furthermore, victims who were drinking at the time ofthe assault were also 4 times more likely to be in the One Saving Grace cluster wheretheir involvement with the legal system was cut short. This contrast between Cluster 1 and 2 (Approaching Justice and One Saving Grace) alsoindicated that women who lived in communities with more resources for addressingsexual assault, and those who lived in communities where there was more coordination ofsuch services had relatively positive experiences across all three systems (ApproachingJustice). Women who were injured in the assault were also somewhat more likely to be inthis positive experiences cluster. Finally, women who exhibited "good victim" behavior(i.e., showed visible distress) were also more likely to be in the Approaching Justicecluster. In the second contrast, Cluster 1 (Approaching Justice) and Cluster 3 (Exercises inFutility) were compared. Fewer variables differentiated these two clusters. Women in theExercises in Futility cluster were 4 times more likely to have been raped by a someoneknown to them without the use of a weapon. Again, nonstranger rape without the use of aweapon was associated with a negative response from the legal system. A trend emerged for match in race: Intraracial rapes were somewhat more likely to be in the clusterApproaching Justice, and interracial rapes in the cluster Exercises in Futility. Specifically,women of Color who were raped by White men were somewhat more likely to have beenin the Exercises in Futility cluster. The Wald test provides an overall test of the predictive value of each variable in theequation by averaging across these contrasts. So, as expected, variables significant in theindividual contrasts remained significant, but the trends in the contrasts were not. Overalleffects were found for community resources, community coordination, stranger rapewithout a weapon, nonstranger rape without a weapon, alcohol use by victim, andwhether the victim exhibited "good victim" behavior. The prediction success indices revealed that this model could successfully classify 54%of the cases, which was significantly better than chance (34%): 53% for Cluster 1(Approaching Justice), 62% for Cluster 2 (One Saving Grace), and 46% for Cluster 3(Exercises in Futility). The prediction success index, which measures the gain the modelexhibits in the number it correctly predicts versus a purely random model. was .21 forCluster 1, .23 for Cluster 2, and .16 for Cluster 3. The larger the success index, the betterthe model did in successfully classifying cases. This index can be negative if theclassification was worse than chance. Both the percentage correctly classified and theprediction success index for Cluster 3 (Exercises in Futility) were somewhat low, butgiven that fewer variables distinguished the third cluster, it is not surprising that thesevalues pulled down the overall rate for the model. In the final set of analyses, the utility of each level in the proposed model was considered.
It has been argued that the advantage of an ecological perspective is that it can help usunderstand phenomena from a multiple-level perspective. At issue for this final set ofanalyses is whether all three levels of ecological theory predict unique variance in clustermembership. Do we really need information about the community, about the rape, andabout the victim to predict outcome? A series of LR tests were performed to examine theunique variance explained by each level. In this test, the LR for the full model iscompared to the LR for a model with a block of variables removed. The full model iscompared to this nested model. The LR from the nested model is subtracted from the LRof the full model. This difference is a chi-square statistic that is then evaluated forsignificance (with the difference degrees of freedom). This chi-square statistic should besignificant, indicating that the model with this block i s significantly different from amodel without this block of variables. (The logic of these procedures was suggested byDarlington, 1968, for ordinary least squares regression, and is adapted here for logisticregression.) With the variance accounted for by the rape-related and individual-level factors alreadyestablished, the community-level factors explained additional unique variance (differenceLR [chi square] = 6.17, p < .05). Moreover, after taking into account the effects of thecommunity and individual level factors, the rape-related factors could still explain uniquevariance (difference LR [chi square] = 38.70, p < .001). Finally, the variance accountedfor by the community level and rape-related variables still left unexplained variance that could be successfully predicted by the individual level factors (difference LR [chi square]= 10.84, p < .05). Thus, each level specified by ecological theory explained uniquevariance in victims' experiences with community systems. These results suggest that allthree levels of variables specified by ecological theory were necessary to predict victims'experiences with the legal, medical, and mental health systems. This study provides national-level data that describe rape victims' experiences with thelegal, medical, and mental health systems, and explains how multilevel factors impactservice delivery. Cluster analysis was used to describe victims' experiences with thesesystems. A three-cluster solution was supported. The first cluster (Approaching Justice)was characterized by relatively high services, high fit, and low advocacy across all threesystems. These women were able to obtain most of the services they wanted with littleintervention. They also lived in communities where there were more resources forvictimized women, and there were more programs in place to coordinate those resourcesto streamline service delivery. Stranger assaults with the use of a weapon were morecommon in this cluster, and these victims were slightly more like to have been injured inthe assault. These findings are consistent with previous research by Estrich (1987),Kerstetter (1990), LaFree (1981), Madigan and Gamble (1991), McCahil l et al. (1979),and Rose and Randall (1982), which indicated that the legal system respondsdifferentially when weapons and injuries are involved. These are perceived as "real"crimes and as such are more likely to be prosecuted. The medical and mental healthsystems were also quite responsive under these circumstances. Furthermore, Madigan andGamble (1991) described from their clinical work with rape victims the "good victim"phenomenon: Women who are clearly distressed and receptive to help may be treatedpreferentially. The results of this quantitative, larger scale study indicate that suchdemeanor was associated with the most favorable outcomes. For many of the cases in thiscluster, several of the key ingredients that appear to prompt social systems to respondwere present. In the second cluster (One Saving Grace), women did not have uniform experiencesacross these three systems. Most of these cases were not forwarded for criminalprosecution, which was not consistent with what the victims wanted. In the mental healthsystem, victims in this cluster received only some of the services they wanted. Bycontrast, the medical system was more responsive to these women's needs--their "savinggrace." The victims second cluster were more likely to have been raped by someone theyknew without the use of a weapon and were more likely to have been drinking at the timeof the assault. These findings provide some quantitative support to qualitative narrativescollected by Finkelhor and Yllo (1985), Russell (1990), and Warshaw (1988), whichindicated that rapes between known parties are often met with skepticism Very littleresearch has examined how alcohol affects service delivery, and these findings supportattitudinal research that suggested that alcohol use "negates" the rape (Richardson &Hammock, 1991). This profile of findings is consistent with what is often thought of asthe typical date rape--rape between known parties, under the influence of alcohol. Theseresults imply that although the medical system may not respond differentially to these cases, the legal and mental health systems may have implicit rules for service delivery indate rapes. The third cluster (Exercises in Futility) was characterized by negative experiences acrossall three systems. Due to the advocates' efforts, many cases did proceed through the initialsteps of criminal prosecution, but were irrevocably stalled in the final stages. In themedical and mental health systems, this pattern of frustration repeated itself. The victimswanted far more services than they were able to receive, despite the efforts of theadvocates. Only two of the variables in the ecological model differentiated this thirdcluster rape by a nonstranger without the use of a weapon, and a mismatch between thevictims' and assailants' races (trend). In some respects, it appears that this cluster wasdefined as much by what it lacked as what it included. These cases did not involvestrangers; they did not involve weapons; they did not involve injuries; they did notinvolve alcohol use; they did not have victims who exhibited "good victim" behavior. Inother words, these cases lacked many of the factors that our social systems may use todecide how to respond. What these cases did involve was rape between known parties,and for some of the women of Color in this cluster, a rape committed by a White man. The findings from this research suggest that for rape victims to receive desired servicestheir cases may need to fit a rather constricted mold. When certain characteristics of thevictim, the assault, and the community are in careful alignment, the likelihood of anoutcome that is consistent with victims' needs is most probable. As these factors start todeviate from this narrow path, the number of services may drop off, the fit with victims'wishes may be compromised, and the advocacy needed to bring about beneficialoutcomes may rise. Furthermore, it appears that the legal system may be the leastforgiving of such "deviations." Cases that do not conform were often filtered out of thesystem The medical system may not work under such stringent implicit rules. In cases ofdate rape, for example the doctors and nurses in this study often responded in a mannerconsistent with victims' needs. The implicit rules of mental health system, however, didnot emerge as clearly. Many women received the short-term help they desired, butlonger-term help was not as readily available, creating relatively negative experiences forwomen in two of three the clusters. This research also provides an empirical examination of ecological theory. In this study,community-level, rape-related, and individual-level factors explained unique variance inthe outcome variable (cluster membership/victims' experiences). Whereas the individualpredictors in this model may or may not have been successful predictors, the levelsappear to be instrumental. In other words, drawing on information from the environmentthe rape itself, and some characteristics of the victims were necessary to predict victims'outcomes. For instance. race and victims' demeanor explain only part of the pattern. Withadditional information about the type of rape, the use of a weapon, injuries sustained, andalcohol use at the time of the assault, more variance can be accounted for. Furthermore,an even more distant variable, such as the coordination of the services available tovictims in these communities, had an effect. The programs and policies in a communitythat work to streamline services to victims may trick le down to affect the specificexperiences of individual victims. These findings provide some empirical support for Kelly's (1966, 1968, 1971) theoretical arguments that environmental structures andpractices may impact individual outcomes. The individual level is a necessary but notsufficient approach to understand victim-system interactions such as these. Two methodological limitations temper the conclusions of this study First, collecting datafrom rape victim advocates raises issues of the reliability and validity of these reports.
The reliability of the data may have been influenced by the advocates' ability to recall thedetails of the assault and system response. To address this issue, the advocates were askedto review their case notes prior to conducting the interview, and we assessed test-retestreliability (r = .95). To provide some insight into the validity of the advocate reports,complete validating information from the advocates, victims, hospital staff, police, andprosecutors was obtained for only ten cases (randomly selected from the sample). In theseten cases, there was consistent agreement from all parties as to what services wereprovided to the victim. But due to the fact that validity was not assessed for the entiresample, it is possible that the advocates may have had a different perspective than that ofthe victims and social system pe rsonnel. Second, because this study did not include a comparison group of victims who did notwork with advocates, we do not know what effect the mere presence of the advocate mayhave had on service delivery. If this presence was beneficial, then these data mayoverreport the frequency with which victims are receiving help. Given that many womenstill did not receive all of the help they desired, and had an advocate, then we may havereason to be quite concerned for victims who do not have such assistance. In conclusion, this study provides a view into rape victims' experiences with the legal,medical, and mental health systems. When women go public with their stories of rape,they do not all have negative experiences, nor do they all have positive outcomes. Whenwe consider victims involvement with multiple community systems, we can begin to seethe uniqueness of each system, as well as the totality of victims' experiences. Theresponse from these social systems appears to be function of several variables.
Characteristics of the victim affect these interactions as do features of the rape itself.
Even more distant variables, such as the community coordination of resources forvictims, can have tangible effects on victims' experiences. These findings present someinitial ideas as to how our social systems can be reorganized to create settings that aremore receptive to victims' needs. Improving victims' experiences with the legal, medical,and mental health systems may require both individual and structural chan ges. (1.) The author thanks Ana Man Cauce, Bill Davidson, Chris Keys, Deborah Salem, andSarah Ullman for their helpful comments on this paper; the members of the CommunityResponse to Rape Project for their assistance in data collection; and the rape victimadvocates who participated in this study for their time, expertise, and feedback on thismanuscript. This research was the first-place recipient of the Society for CommunityResearch and Action (Division 27 of the American Psychological Association) 1997Dissertation Award, William S. Davidson, II, Chair. (2.) A11 correspondence should be addressed to Rebecca Campbell, Department of Psychology. University of Illinois at Chicago, 1007 West Harrison, Chicago, Illinois60607-7137. (3.) Of the 36 centers that were not eligible for the study, 8 were no longer in existence,19 have changed their services since the publication of the directory and now providedonly therapy for victims, 2 had not had a sexual assault case in the past 3 years, 5 nowworked only with domestic violence victims, and 1 now worked only with victims ofchild sexual abuse. (4.) Another item, rape exam and evidence collection, was asked, but dropped from thescale due to lack of variance as all communities offered this service. (5.) In the first set of analyses where all predictor variables were entered into the model,two problems became apparent. First, several variables were not significant: victims' race,assailants race, and victims' SFS. Second, a suppressor effect emerged because of amoderate correlation between type of rape and use of a weapon. As a result, Hosmer andLemshow's (1989) recommendations were followed: Drop variables that have nopredictive value and rerun a smaller, better fitting model. Retaining the variables thatmake no significant contribution to the outcome variable can artificially inflate thegoodness-of-fit indices. To address the suppressor effect, three dummy-coded variableswere created to capture four relationships that could exist between type of rape (strangeror nonstranger) and weapon use (weapon or no weapon). Abbey, A., Ross, L. T, McDuffie, D., & McAuslan, P. (1996). Alcohol and dating riskfactors for sexual assault among college women. Psychology of Women Quarterly, 20,147-169. Agranoff, R., & Pattakos, A. (1979). Dimensions of services integration: Servicedelivery, program linkages, policy management, and organizational structure. HumanService Monograph Series 13. Washington, DC: U.S. Government Printing Office. Aldenderfer, M. S., & Blashfield, R. K. (1984). Cluster analysis. Newbury Park. CA:Sage. Baker, P., & O'Brien, G. (1971). Intersystems relations and coordination of humanservice organizations. American Journal of Public Health, 61, 130-137. Beebe. D. K. (1991). Emergency management of the adult female rape victim. AmericanFamily Physician, 43, 2041-2046. Blashfield, R. K., & Aldenderfer, M. S. (1988). The methods and problems of clusteranalysis. In J. R. Nesselroade & R. B. Cattell (Eds.), Handbook of multivariateexperimental psychology (pp. 447-474). New York: Plenum. Campbell, R., Baker, C. K., & Mazurek, T. (1998). Remaining radical? Organizationalpredictors of rape crisis centers' social change initiatives. American Journal ofCommunity Psychology, 26, 457-483. Campbell, R., & Bybee, D. (1997). Emergency medical services for rape victim:Detecting the cracks in service delivery. Women's Health: Research on Gender, Behavior,and Policy, 3, 75-101. Cohen. J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.).
Hillsdale, NJ: Erlbaum. Darlington, R. B. (1968). Multiple regression in psychological research and practice.
Psychological Bulletin, 69, 161-182. Davis, L. E., & Proctor, E. K. (1989). Race, gender, and class: Guidelines for practicewith individuals, families, and groups. Englewood Cliffs, NJ: Prentice-Hall. Estrich, S. (1987). Real rape: How the legal system victimizes women who say no.
Cambridge. MA: Harvard University Press. Fairstein, L. A. (1993). Sexual violence: Our war against rape. New York: WilliamMorrow. Feldman-Summers, S., & Palmer, G. P. (1980). Rape as viewed by judges, prosecutors,and police officers. Criminal Justice and Behavior, 7, 19-40. Finkelhor, D., & Yllo, K. (1985). License to rape: Sexual abuse of wives. New York:Free Press. Foa, E. B., Rothbaum, B. O., Riggs, D. S., & Murdock, T. B. (1991). Treatment ofposttraumatic stress disorder in rape victims: A comparison between cognitive-behavioralapproaches and counseling. Journal of Consulting and Clinical Psychology, 59, 715-723. Foal E. B., Steketee, G., & Olasov, B. (1989). Behavioral/cognitive conceptualization ofposttraumatic stress order. Behavior Therapy, 20, 155-176. Forman, B. D., & Wadsworth, J. C. (1983). Delivery of rape-related services in CMHCs:An initial study. Journal of Community Psychology, 11, 236-240. Frank, E., & Stewart, B. D. (1984). Depressive symptoms in rape victims: A revisit.
Journal of Affective Disorders, 7, 77-85. Frank, E., Stewart, B. D., Dancu, C., Hughes, C., & West. D. (1988). Efficacy ofcognitive behavior therapy and systematic desensitization in the treatment of rape trauma.
Behavior Therapy, 19, 403-420. Galvin, J., & Polk, K. (1983). Attrition in case processing: Is rape unique? Journal ofResearch in Crime and Delinquency, 20, 106-154. Golding, J. M., Siegel, J. M., Sorenson, S. B., Burnam, M. A., & Stein, J. A. (1989).
Social support following sexual assault. Journal of Community Psychology, 17, 92-107. Gordon-Bradshaw, R. H. (1988). A social essay on special issues facing poor women ofcolor. Women and Health, 12, 243-259. Gornick, J., Burt, M. R., & Pittman, K. J. (1985). Structure and activities of rape crisiscenters in the early 1980s. Crime and Delinquency, 31, 247-268. Harvey, M. (1982a). Helping victims and preventing rape: A look at three effectiveprograms. Response to the Victimization of Women and Children, 5, 4-6. Harvey, M. (1982b). Helping victims and preventing rape: Underpinnings of programeffectiveness and success. Response to the Victimization of Women and Children, 5, 7-9. Harvey, M. (1985). Exemplary rape crisis program: Cross-site analysis and case studies.
U.S. Department of Health and Human Services, National Institute of Mental Health,Rockville, MD. Hosmer, D. W., & Lemeshow, S. (1989). Applied logistic regression. New York: Wiley. Kelly, J. G. (1966). Ecological constraints on mental health services. AmericanPsychologist, 21, 535-539. Kelly, J. G. (1968). Towards an ecological conception of preventive interventions. In J.
W. Carter, Jr. (Ed.), Research contributions from psychology to community mental health(pp. 75-99). New York: Behavioral Publications. Kelly, J. G. (1971). Qualities for the community psychologist. American Psychologist,26, 897-903. Kerstetter, W. A. (1990). Gateway to justice: Police and prosecutorial response to sexualassaults against women. Journal of Criminal Law and Criminology 81, 267-313. King, E. H., & Webb, C. (1981). Rape crisis centers: Progress and problems. Journal ofSocial Issues, 37, 93-104. Koss, M. P., & Harvey, M. R. (1991). The rape victim: Clinical and communityinterventions. Newbury Park, CA: Sage. Koss, M. P., Woodruff, W. J., & Koss, P. G. (1991). Criminal victimization amongprimary care medical patients: Prevalence, incidence, and physician usage. BehavioralSciences and the Law, 9, 85-96. LaFree, G. D. (1980). Variables affecting guilty pleas and convictions in rape cases:Toward a social theory of rape processing. Social Forces, 58, 833-850. LaFree, G. D. (1981). Official reactions to social problems: Police decisions in sexualassault cases. Social Problems, 28, 582-594. Luke, D. A., Rappaport, J., & Seidman, E. (1991). Setting phenotypes in a mutual helporganization: Expanding behavior setting theory. American Journal of CommunityPsychology 9, 147-167. Madigan, L., & Gamble, N. (1991). The second rape: Society's continued betrayal of thevictim. New York: Lexington Books. Mama, A. (1989). The hidden struggle: Statutory and voluntary sector responses toviolence against black women in the home London: The Runnymede Trust. McCahill, T. W., Meyer, L. C., & Fischman, A. M. (1979). The aftermath of rape.
Lexington, MA: Lexington Books. Mowbray C. T., Bybee, D., & Cohen, E. (1993). Describing the homeless mentally ill:Cluster analysis results. American Journal of Community Psychology, 21, 67-93. National Victim Center. (1992). Rape in American: A report to the nation. Arlington,VA: Author. Norris, F. H. (1992). Epidemiology of trauma: Frequency and impact of differentpotentially traumatic events on different demographic groups. Journal of Consulting andClinical Psychology, 60, 409-418. Parrot, A. (1991). Medical community response to acquaintance rape: Recommendations.
In L. Bechhofer & A. Parrot (Eds.). Acquaintance rape: The hidden victim (pp. 304-316).
New York: Wiley. Pinderhughes, E. (1989). Understanding race, ethnicity, and power The key to efficacy inclinical practice. New York: The Free Press. Rapkin, B. D., & Luke, O. A. (1993). Cluster analysis in community research:Epistemology and practice. American Journal of Community Psychology, 21, 247-277. Resick, P. A., Jordan, C. G., Girelli. S. A., Hutter, C. K., & Marhoefer-Dvorak, S. (1989).
A comparative outcome study of behavioral group therapy for sexual assault victims.
Behavior Therapy, 19, 385-401. Resick, P. A., & Schnicke, M. K. (1992). Cognitive processing therapy for sexual assaultvictims. Journal of Consulting and Clinical Psychology, 60, 748-756. Richardson, D. R., & Hammock, G. S. (1991). Alcohol and acquaintance rape. In L.
Bechhofer & A. Parrot (Eds.), Acquaintance rape: The hidden victim (pp. 83-95). NewYork: Wiley. Rose, V. M., & Randall, S. C. (1982). The impact of investigator perceptions of victimlegitimacy on the processing of rape/sexual assault cases. Symbolic Interaction, 5, 23-36. Rothbaum, B. O., Foa, E. B., Riggs, D. S., Murdock, T, & Walsh, W (1992). Aprospective examination of post-traumatic stress disorder in rape victims. Journal ofTraumatic Stress, 5, 455-475. Russell, D. E. H. (1990). Rape in marriage (2nd ed.). New York: Macmillan. Tausig, M. (1987). Detecting "cracks" in mental health service systems: Application ofnetwork analytic techniques. American Journal of Community Psychology, 15, 337-351. Trickett, E. J., Kelly, J. G., & Vincent, T. A. (1985). The spirit of ecological inquiry incommunity research. In E. C. Susskind & D. C. Klein (Eds.), Community research:Methods. paradigms, and applications (pp. 283-333). New York: Praeger. Turner, J., & TenHoor, W. (1978). The NIMH community support programs: Pilotapproaches to a needed social reform. Schizophrenia Bulletin, 4, 319-408. U.S. Department of Justice. (1994). The criminal justice and community response to rape.
Washington, DC: Author. Warshaw, R. (1988). I never called it rape: The Ms. report on recognizing, fighting, andsurviving date and acquaintance rape. New York: Harper and Row. Webster, L. (1989). Sexual assault and child sexual abuse: A national directory ofvictim/survivor services and prevention programs. Phoenix, AZ: Oryx. Wyatt, G. E., Notgrass, C. M., & Newcomb, M. (1990). Internal and external mediatorsof women's rape experiences. Psychology of Women Quarterly, 14, 153-176.

Source: http://www.taasa.org/library/pdfs/TAASALibrary164.pdf