EFTA00156736.pdf

Article Violence Atonu Woolen 16(1) 262-2)9 Women's Reporting of °The Amt.:y(02010 Wens and permason hopinnwr Sexual and Physical sagopciacomflorntheonusoonsroo Da 10 1177/1077001209360061 Assaults to Police in the taplivnes3gepscom OSAGE National Violence Against Women Survey Yingyu Chen I and Sarah E. Ullmani Abstract Previous research has assumed that rape reportingis unique.but no study has systematically tested this assumption. The present study used a probability sample of female victims from the National Violence Against Women Survey to compare factors affecting rape and physical assault reporting using multinomial logit regression. Overall. results suggested that there was similarity in reporting decisions between rapes and physical assaults in terms of main effects. However.inceractions suggested chat age. marital status.and physical force each influenced reporting differently by assault type.Implications of these results are discussed and directions for future research are offered. Keywords police reporting. physical assault, rape Of all personal crimes, rape/sexual assault has been considered the most serious and trau- matic. short of homicide (Koss & Harvey, 1991; Resick & Nishith, 1997). Women may be attacked by various types of perpetrators, most often someone they know (e.g.. acquain- tances. partners). The National Women's Study (NWS) indicated that only 22% of rape victims were assaulted by strangers or someone they did not know well, 9% by husbands or ex-husbands, 11% by fathers or stepfathers. 10% by boyfriends or ex-boyfriends. 16% by other relatives, and 29% by other nonrclatives, such as friends or neighbors (Kilpatrick, Edmunds. & Seymour. 1992). Victims of sexual violence may experiences rangeof behav- ioral and psychological problems, which can last from a few months to as long as several ITaipes.Taivean 'University of Minas at Chicago Corresponding Author: Yingyu Chen. 4F. No.24. Lane 342. Lontliang Road.Talpe. 10474.Tarman Email yywchenagmail.com 3502-012 Page 1 of 18 EFTA_00001395 EFTA00156736 Chen and Ullman 263 years (Frazier. 2003: Rothbaum, Foa. Riggs, Murdock, & Walsh, 1992). Despite these seri- ous consequences. victims of sexual offenses am less likely to report incidents to the police than victims ofother violent crimes. According to the National Crime Victimization Survey (NCVS). rape/sexual assault was least likely to be reported of all violent crimes between 1992 and 1999 (Bureau ofJustice Statistics. 2003). and this trend persisted in 2006 (Bureau of Justice Statistics. 2007). Sexual assault victims who do not report incidents to the police may adhere to socially constructed perceptions of rape such as rape myths (Bun 1991; Lonsway & Fitzgerald. 1994) and other beliefs that lead people to regard rape as justifiable (Nluchlenhard. Friedman. & Thomas. 1985). Fear of receiving negative responses from legal authorities may also dis- courage sexual assault victims from reporting their experiences to police (Campbell. 2005). Researchers have found that police were less likely to make an arrest if the victim's credibil- ity was questionable (Frazier & Haney. 1996; Martin & Powell. 1994). if the suspect was acquainted with the victim (fistrich. 1987; Frazier & Haney. 1996: LaFree. 1989). or if there was no substantive evidence (Frazier & Haney. 1996: Kerstetter. 1990). The decision by pros- ecutors to charge is also affected by similar legal and extralegal factors that affect police decisions to arrest (Frazier & Haney. 1996: Roll:mann. 1991: Kerstetter, 1990). Given that socially constructed perceptions of rape and negative responses from legal authorities are common, such as disbelief and blame of rape victims. it is reasonable to assume that factors affecting their decisions to report are different from those affecting victims ofother violent crimes. Myers and LaFree (1982) found few significant differences in prosecutorial decisions between sexual assaults and other crimes but suggested that future research should examine whether this conclusion holds for stages before prosecu- tion, such as victims' reporting decisions. Although research exists on correlates of rape reporting. fewer studies have looked at physical assault reporting. Research on correlates of Tenoning has been atheorctical and has often examined victim demographics and assault characteristics. Past research in both of these areas is now reviewed. Correlates of Rape Reporting Several demographic characteristics ofrape victims relate to their reponing decisions. Victim age is unrelated to police reporting (Bachman. 1998: Clay-Warner & Bun. 2005; Fisher, Daigle. Cullen. & Turner. 2003; Golding. Siegel. Sorenson. Bumam. & Stein. 1989: Lizotte, 1985), education is negatively related to rape reporting (Lizotte. 1985; Pino & Meier. 1999), and victim's marital status (Bachman, 1998: Lizotte. 1985: Russell & Bolen. 2000). mcd ethnicity (Bachman. 1998; Fisher ct al.. 2003: Wyatt. 1992). and socioeconomic status (Bachman. 1993; Pino & Meier, 1999) show inconsistent relationships with rape reporting. Studies of both probability and nonprobability samples show that rapes/sexual assaults committed by strangers are more frequently reported to police than those perpetrated by nonstrangers (Edson & Pare, 2005: Fisher et al.. 2003: Golding et al.. 1989: Greenberg & Ruback, 1992; Lizotte, 1985; Pino & Meier. 1999; Ruback & Menard. 2001; Russell & Bolen, 2000), with the exception of three studies using NCVS data (Bachman, 1993. 1998; Banner. Felson. & Messner. 2003), perhaps due to known measurement limitations of this survey (Koss. 1992). Women whose experiences meet the legal definition of rape/sexual 3502-012 Page 2 of 18 EFTA_00001396 EFTA00156737 264 Vioknce Against Women 16(3) assault may not acknowledge themselves (i.e.. self-label) as rape victims if they believe that physical force or weapons are essential elements of rape/sexual assault (Fisher et al.. 2003: Littleton. Axsom, Breitkopf, & Berenson. 2006). Physical force by assailants (Bachman. 1993; Du Mont. Miller, & Myhr. 2003: Russell & Bolen. 2000). presence of weapons (Fisher et al., 2003; Russell & Bolen. 2000). completed rape (Bachman. 1993; Golding ct al., 1989: Russell & Bolen, 2000). and victim injuries (Bachman. 1993. 1998: Du Mont ct al., 2003: Lamont, 1985; Pino & Meier. 1999) am all related to increased rape reporting. Women arc most likely to report incidents happening at home or in their cars and least likely to report attacks in social situations such as dates (Williams. 1984). The effect ofdnig or alcohol consumption on rape reporting has been less studied, but women who are intoxi- cated may be more likely to be viewed as sexually available and to be blamed for being raped (Norris. Nurius. & Dirndl: 1996), resulting in less police reporting. However, research on rape reporting shows that neither offender nor victim use of drugs/alcohol prior to inci- dents affects victims' reporting decisions (Clay-Warner & Bun. 2005; Fisher et al., 2003). Correlates of Physical Assault Reporting Only a handful of existing studies have examined correlates of physical assault reporting to police. Two studies found that police reporting did not differ by victim race/ethnicity (Baumcr, 2002; Felson. Messner. & Hoskin. 1999), but one study found that Black victims were more likely to notify police than White victims (Folsom. Messner. Hoskin. & Deane, 2002). Older victims were more likely to report assaults to police than younger victims (Balmer, 2002; Felson et al., 2002). Victim income and educational level were negatively related to simple assault reporting but had no effect on aggravated assault reporting, whereas married victims were more likely than unmarried victims to report assaults (Batumi. 2002). Unlike rape reporting studies, which have consistently shown that victims arc more likely to report rapes/sexual assaults committed by strangers. research on physical assault reporting based on NCVS data has shown mixed findings regarding the effect of victim— offender relationship. One study showed that victims were less likely to call police if the attacker was a nonstzanger outside the family than if the attacker was a stranger (Felson et al.. 2002). and another study showed more reposing when the offender was an ex-spouse rather than a stranger (Felson et al...1999).A third study showed that simple assault, but not aggravated assault, was more likely to be reported for family member than for stranger assailants (Baumer. 2002). Studies have consistently shown a positive association between seriousness ofassaults and reporting rates. Felson et al. (1999) applied the Sellin-Wolfgang Scale (Sellin & Wolfgang, 1964) and found the more serious a criminal event was, the more likely the victim was to call police. Researchers also found that assaults were more likely to be reported to police if the victim suffered injury or if the offender was armed (Baumer, 2002; Felson et al., 2002). Incident location and offender use of drugs/alcohol arc also related to physical assault reporting. Police were more likely to be notified when incidents happened in victims' homes (Felson et al., 2002) or in their neighborhoods (Baumer, 2002). Jasinski (2003) analyzed NCVS (1992-1994) data and found that female assault victims were more likely to report to police if offenders used either drugs or akohol at the time of the incident. but the 3502-012 Page 3 of 18 EFTA_00001397 EFTA00156738 Chen and Unman 265 association disappeared after other individual and incident-related correlates were included in the analysis. Comparing Sexual and Physical Assault Reporting Some studies have combined sexual and physical assault with other crimes (e.g., robbery. theft) to examine correlates of victims' reporting behavior (e.g., Conaway & Lehr. 1994: Singer. 1988; Webb & Marshall, 1989). without examining each type of crime separately or comparing correlates of reporting decisions. Only Lizottc's (1985) study of National Crime Survey (NCS) (1972-1975) data showed that factors that made strong cases for prosecution. such as stranger assailants and serious injuries, were better predictors ofrape reporting than of physical assault reporting, leading him to conclude that rape reporting may be unique. Although important. Lizottc's (1985) study was limited by reliance on the NCS. which has been criticized for methodological problems (Eisenberg. 1990; Koss, 1992): statistical analyses comparing two assault types that were not stringent; combining males and females: and not separating victim reporting from third-party reporting. The latter point is important because third parties notified police ofnearly halfof violent crimes (46%) between 1992 and 2000 (Bureau ofJustice Statistics.. 2003), yet determinants of third-party reporting are rarely studied. In a few studies. third parties were less likely to report rapes/sexual assaults if vic- tims were older or weapons were involved (Greenberg & Ruback. 1992), and less likely to report intimate partner sexual or physical assaults (Sauna et al.. 2003; Pelson et al., 1999). Present Study Compansons between rape reporting and physical assault reporting arc important because if distinct correlates are found, special methods for encouraging rape victims to report will need to be developed. However, if they are similar, possibly public awareness of sexual violence against women has increased, and women arc equally likely to report sexual and nonsexual incidents. Due to the lack of research on the uniqueness of rape reporting. the present study used the National Violence Against Women Survey (NVAWS) to evaluate whether factors identified in the past research arc unique to rape reporting compared with physical assault reporting in women. It was hypothesized that assault characteristics (c.g., physical force, injury, weapons. stranger assailants. perceived life threat) would play a stronger role in sexual assault reporting than physical assault reporting, whereas hypothe- ses were not made about victim demographics given past inconsistent results. Unlike past research, exploratory analyses were done to see if correlates of reporting varied if reported by victims or third parties. Method Sample The National Institute of Justice (NU) and the National Center for Injury Prevention and Control (NC1PC), Centers for Disease Control and Prevention (CDC) jointly sponsored the 3502-012 Page 4 of 18 EFIA_00001398 EFTA00156739 266 Violence Against Women 16(3) NVAWS. conducted by a survey research firm, to funkier the understanding of violence against women (Tjaden & Thoennes. 1999). A national sample of 8.000 women, aged 18 years and older and living in the United States, was drawn by random-digit dialing (RDD) from households with a telephone and was interviewed by female interviewers using a computer-assisted interviewing system between November 1995 and May 1996. In households with more than one eligible woman. the woman with the most recent birthday was selected (Tjaden & Thoennes. 1998a). The participation rate for female respondents was 72.1% (Tjaden & Thacnnes. 19986). The NVAWS defined rape as "an event that occurred without the victim's consent, that involved the use or threat of force to penetrate the victim's vagina or anus by penis. tongue, fingers, or object, or the victim's mouth by penis" (Tjaden & Thoennes 1998a, p. 13). The NVAWS used a set of five screening questions to identify rape victims. Questions were phrased using specific behavioral terms to ask whether respondents had ever been forced or threatened by a man or woman to have vaginal or anal intercourse by penis, tongue. fin- gers, or objects or to have oral intercourse by penis. Both completed and attempted forms of sexual assault were assessed. Physical assault was defined as "behaviors that threaten, attempt, or actually inflict physical harm" (Tjaden & Thocnnes. 1998a. p. 13). A modified version of the Conflict Tactics Scale (Straus, 1990) was used to screen respondents for physical assault experienced as an adult by any type ofperpetrator. The present study selected incidents involving a male offender and a female victim who was at least 14 years of age at the time of assault. If a woman was victimized by more than one type of perpetrator at different periods, the most recent incident was selected. This resulted in 531 sexual assault and 1,278 physical assault cases. There were 441 incidents in which victims experienced both sexual and physical assault in different years. Previous research shows that victims are more likely to report future victimization to the police if their previous reporting experiences have positive results (Conaway & Lohr, 1994: Shapland. Willmore, & Duff. 1985) or if they have a positive view of their prior interac- tions with police (Waller. 1990). Therefore. for women who had been victimized both sexually and physically, their first victimization incident was selected. If sexual and physi- cal victimization occurred in the same year (it = 88), neither incident was included in the analysis because the order of incidents could not be determined. This selection process resulted in a final sample of 874 sexual assault and 1,376 physical assault cases. Measures Dependent variable. Respondents who disclosed their experiences to interviewers were asked whether incidents were reported to police, and if so, who notified police. The depen- dent variable was coded with three outcomes: 0 = no. I =yes. reportedby a thirdparty, and 2 = yes. reported by the respondent. In the multinomial logit model. "reported by the respondent" was used as a reference category. Victim demographic characteristic. To be consistent with the past studies, women who were assaulted at the age of 14 or older were selected. Although age at assault less than 16 may be considered child sexual or physical abuse, the results did not differ when the 3502-012 Page 5 of 18 EFTA_00001399 EFTA00156740 Chen and Ullman 267 analyses were conducted excluding those under 16. Respondent age at the time of assault was coded in years, and employment status was coded as 0 = unemployed or I = employed. Highest level of education was coded as 0 = no college or 1 = some college, and marital status was coded 0 = unmarried or I = married (including common-law relationships). Respondent ethnic background was dichotomized as 0 = non-White or I = White. Respon- dent income was measured by ordinal categories in the NVAWS and was coded in the present study using the midpoint of each income category to treat income as continu- ous: USS0, USS2,500, USS7.500, USS12.500, USSI 7,500, USS22,500. USS30.000. USS42,500, USS65,000, USS90.000. and USSI10,000. Assault characteristic. As the present study focused on male violence against women, respondents victimized by same-sex perpetrators were excluded from analyses. The victim—offender relationship was examined in a more elaborate manner for bivariate analyses: 0 = stranger, 1 = current or former spouse or partner. 2 = relative (father, stepfather, brother, stepbrother. brother-in-law, uncle, grandfather. step-grandfather. male cousin, son/stepson, son-in-law, nephew, nephew-in-law, another male relative). 3 = boy- friend or date, and 4 = acquaintance. To avoid losing degrees of freedom and find the best-fit model for the multivariate analysis. the victim-offender relationship was coded as 0 = non-stranger or I = stranger. Three measures were used to assess assault seriousness: use of physical force (e.g.. slap, kick, beat), presence of weapons (e.g.. gun, knife), and victim physical injuries (each coded 0 = no. I =)ts). Incident location was coded 0=place other than the respondent's home or I = the respondent's home. Respondents were asked about their own and offender use of substances (e.g., drugs. alcohol, or both) at the time of the assault. Victim substance use was coded 0 = no or I =yes and offender substance use was coded 0 = no, I = yes, and 2 = don know. Offender substance use was recoded into two dummy variables (i.e., I =yes vs. 0= all other cases, I = don know vs. 0 = all other cases) to enter the multinomial logit model. Alcohol and drug use were also examined separately, but results did not differ based on the coding schemes. Victim perception of life threat during assault (0 = no or I =yes) was scented by asking whether they believed they or someone close to them would be seriously harmed or killed during the incident. Finally, assault type was coded as 0 = sexual assault and I = physical assault. Results Sample Characteristics Sample characteristics are shown in Table I. Nearly two thirds of the sample (61.1%) con- sisted of physical assault incidents. Three quarters of incidents (74.9%) went unreported. 17.7% were reported by victims, and 7.4% were reported by third panics. Victims' age ranged from 14 to 72 years at the time of the assault (31= 26, SD = 9.45), and personal annual income ranged from USS0 to USS110.0D0 = USS21,000, SD = 18.95). Two thirds of women were employed.. 58.3% had some college education. 79.4% were White. and more than half (53.8%) were married. 3502-012 Page 6 of 18 EFTA_00001400 EFTA00156741 268 ViolenceAgainst Women 16(3) Table I. Description of Variables Variable M (SD) Victim age at assault (from 14 to 72) 26.0 (9.45) Victim nad•polnc of Income (In units of US$1,000) 21.5 (18.95) Assault type Sexual 38.9 Physkal 61.1 Victim age at assault <16 6.9 216 911 Victim employed 65.7 Victim college education 58.3 Victim White 79.4 Victim married 53.8 Stranger offender 11.8 Use of physical force 69.5 Use of weapon 13.2 Victim injured 33.4 Assault happened at home 57.7 Victim substance use 13.2 Perceived life threat 42.4 Incident reported No 74.9 Yes. by thrd party 74 Yes. by victim 17.7 Perpetrator type Stranger 11.8 Former or current spouse or partner 51.6 Relative 5.8 Boyfriend or date 17.8 Acquaintance 13.0 Offender substance use No 36.0 Yes 50.0 Donk know 14.0 Data on assault characteristics showed that 11.8% of incidents involved stranger perpe- trators. Of known perpetrators, more than half (58.5%) were former or current spouses or live-in partners, followed by boyfriends or dates (20.2%), acquaintances (14.7%). and rela- tives (6.6%). Seventy percent of incidents involved physical force. 13.rA involved weapons. and 57.7% happened at the victim's home. Victims perceived that 50% of inci- dents involved offender substance use. 36% did not, and in 14%, victims were unsure. 3502-012 Page 7 of 18 EFTA_00001401 EFTA00156742 Chen and Ullman 269 Table 2. Comparisons of Victim Characteristics and Assault Characteristics Between Sexual Assaults and Physical Assaults Sexual Assault Physical Assault Variable Ms' n Ms' p Victim demographics Age at assault 22.10 874 28.47 1.376 Midpoint of income 20.73 775 21.95 1.160 .165 Employed 0.63 873 0.67 1.375 .076 College education 0.62 874 0.56 1.371 .005° White 0.79 866 0.79 1.344 .941 Married 0.52 871 0.55 1.371 .301 Assault characteristics Use of physical forte 0.33 864 0.92 1.368 .000° Use of weapon 0.09 868 0.16 1.370 .000° Victim injured 0.27 868 0.38 1.366 .000° Happen at home 0.40 869 0.69 1.372 .000° Victim substance use 0.18 857 0.10 1.361 .000° Perceived life threat 0.41 858 0.43 1.353 .289 avariables from "Employed- to-Perceived life threat" mob.e proportion dfference tests. b.Setnificant at the Holm-adjusted .05 level Only 13.2% of victims used substances at the time of the incident. One third of victims were physically injured, and 42.4% feared serious harm or death during the incident. Bivariate Comparisons of Sexual and Physical Assault Independent-samples 1 tests for continuous and dichotomous variables (see Table 2) and chi-square tests for categorical variables (sec Table 3) were conducted to compare sexual and physical assaults on victim demographics and assault characteristics. The Holm pro- cedure was applied to adjust the Type 1 error rate at the .05 level (Aickin & Gensler, 1996). The mean age of women at the time of sexual assault was significantly younger than that of physical assault (22.1 vs. 28.5), 42004) = —16.90, p < .001. Sexual assault victims were significantly more likely to have a college education (62%) than physical assault victims (56%). $1889) = 2.81, p < .01. There were no significant differences in income, employment status, ethnicity, or marital status between victims of sexual assault and physical assault. All assault characteristics significantly differed between sexual and physical assaults except for perceived life threat. Physical assaults were more likely to involve physical force (92% vs. 33%), 41214)= 33.54,p < .001, and weapons (16% vs. 9%). gm!). 5.24, < .001, than sexual assaults. Physical assaults were also more likely to happen at home than sexual assaults (69% vs. 40%), 41768) = 14.15. p < .001. Although sexual assault vic- tims were less likely than physical assault victims to suffer physical injuries (27% vs. 38%), 3502-012 Page 8 of 18 EFTA_00001402 EFTA00156743 270 WofenceAgoinst Wbmen 16(3) Table 3. Comparisons of Whether Reported. Perpetrator Type. and Offender Substance Use Between Sexual Assaults and Physical Assaults Sexual Assault Physical Assault Variable p Whether reported Not reported 713 82.3 956 70.2 Reports by victim 92 10.6 302 22.2 Reported by third party 61 7.0 103 7.6 .000' Perpetrator type Stranger 145 16.6 120 8.7 Spouse/parmer 205 23.5 955 69.5 Relative 92 10.5 39 2.8 Boyfriend/date 222 25.4 178 13.0 Acquaintance 210 24.0 82 6.0 .000' Offender substance use No 276 31.6 532 38.8 Yes 442 50.6 680 49.6 Don't know 155 17.8 160 11.7 .000' a. Sign/cant at the Holm.adiusted .05 level. 41966) = —5.47.p < .001. they were more likely than physical assault victims to use sub- stances at the time of the incident (18% vs. 10%), $1518) = 4.89, p <.001. Women were most likely to be sexually assaulted by boyfriends or dates (25.4%), acquaintances (24%), spouses or parmcrs (23.5%), and strangers (16.6%). whereas physical assaults were committed by spouses/partners (69.5%). boyfriends or dates (13%), strang- ers (8.7%). and acquaintances (6%), e(4. = 2.248)=482.32,p <.001. Both assault types were least likely to be perpetrated by relatives. Sexual assaults were more likely to involve perpetrator substance use and victims unsure about offender substance use compared with physical assaults, N= 2,245) = 21.84, p <.001. Finally, incident reporting varied by assault type, with sexual assaults less often reported to police than physical assaults, x2(2, N= 2,227) = 50.54,p <.001. Multivoriate Analyses of Police Reporting First, a multinomial logic model (sec Table 4) was estimated using the entire sample to examine the effect of victim demographics, assault characteristics, and assault type on reporting both sexual and physical assaults. Second. interaction terms were added to the multinomial logit model (see Table 5) to examine whether correlates differed by assault type. "Reported by victims" was chosen as the reference group for the model because the present study mainly focused on two comparisons: (a) nonsponing versus victim report- ing and (b) third-party versus victim reporting. In addition, victim annual income was omitted due to missing data and the fact that it was nonsignificant in preliminary analyses. 3502-012 Page 9 of 18 EFIA_00001403 EFTA00156744 Chen and Unman 271 Table 4. Multinomial Logic Model Predicting Polka Reporting (N = 2.072) No vs Vktim Report' Third Party vsVicUm Reports Variable ISE) 9(SE) Intercept 3.649 (0.325) 0.377 (0.488) Victim demographics Age at assault -0.022 (0.007)" -0.031 (0.01l)" Employed -0.014 (0.140) -0.518 (0.206)" College education 0.281 (0.133)" -0.333 (0.203) White 0.530 (0.155)" 0.030 (0.231) Married 0.213 (0.132) 0.173 (0.204) ASSIUIL characteristics Stranger perpetrator -1.388(0.220)" 0.361 (0.290) Use of physical force -0.360 (0.208)* -0.478 (0.288)* Use of weapon -0.476 (0.176)" 0.215 (0.240) Victim injured -0.788 (0.140" 0.184 (0.227) Happen at home —0.252 (0.156) 0.024 (0.240) Offender substance use lb —0.280 (0.154)* 0.402 (0.269) Offender substance use 2° —0.843 (0.213)" 0.484 (0.315) Victim substance use 0.271 (0.214) —0.426 (0.354) Perceived life threat —1.048 (0.140)" 0.019 (0.229) Assault type Physical assault —0.703 (0.175)" —0.304 (0.242) —2 log likelihood 2418657 Likelihood ratio '(, I 525.823 df 30 p value .000 McFadden's R2 .179 a"Reported by the mum" is the reference category:0 = warn report vs. I = not reported: 0 = mann report vs. I = thrdixtrty report b.Offender substance use 1:0 = other cases. I = offender used substances: Offender substance use 2: 0 = ether cases. I = morns clef not know whether Wenders used substances. < lap < OS. No multicollinearity was detected among the independent variables. A total of 2,072 cases (804 sexual assaults and 1.268 physical assaults) were included in the analyses: 1.559 unre- ported. 360 reported by victims, and 153 reported by third parties. Age at assault, college education, and ethnic background had significant effects on the probability of reporting by victims. with more reporting by older, less-educated, and non- White victims. Moreover. five assault characteristics significantly affected victims' reporting decisions. The odds of victim reporting compared with nonrcporting were greater for attacks by strangers. with perceived life threat, weapons, victim injury. and when victims were unsure whether offenden used substances. Also, victim reporting was twice as likely for physical assault as for rape. With respect to third-party reporting, two victim demographics. 3502-012 Page 10 of 18 EFIA_00001404 EFTA00156745 272 Violence AgainstWomen 16(3) Table S. Multinomial Logic Model With Interaction Terms Predicting Police Reporting (N = 2.072) No vsVlcdrri Report' Third Party vs Victim Report' Variable 13 (SE) a (SE) Intercept 3.945 (0.587r 1.331 (0.836) Victim demographic Age at assault -0.023(0.015) -0.074 (0.025)', Employed 0.148 (0.275) -0.607 (0.363)• College education 0.011 (0.273) -0.224 (0.368) White 0.480 (0.312) 0.209 (0.419) Married 0.228 (0.274) -0.367 (0.380) Assault chancterisocs Stranger perpetrator -1.088 (0.359y" 0.238 (0.453) Use of physical force -0.790 (0.324y" -0.702 (0.452) Use of weapon -0.821 (0.368)•' 0.023 (0.449) Victim injured -0.809 (0.310) 0.123 (0.436) Happen at home -0.413 (0.294) -0.001 (0.397) Offender substance use 1° -0.125 (0.348) 1.110 (0.569)• Offender substance use 2° -0.627 (0.401) 1.001 (0.603)• Victim substance use 0.604 (0.444) -0.328 (0.622) Perceived life threat -1.069 (0.311y" -0.259 (0.429) Assault type Physical assault -1.799 (0.754y" -2.344 (1.114)" Interaction terms Assault type x Age at assault 0.003 (0.017) 0.057 (0.028)" Assault type x Employed -0.255 (0320) 0.156 (0.445) Assault type x Education 0.378 (0.314) -0.254 (0.446) Assault type x White 0.111 (0.361) -0.227 (0.506) Assault type x Married -0.076 (0.314) 0.832 (0.456)• Assault type x Stranger -0.460 (0.475) 0.502 (0.612) Assault type x Physical force 1.120 (0.452) 0.609 (0.620) Assault type x Weapon 0598 (0.423) 0.272 (0.539) Assault type x Injury 0.060 (0.349) 0.060 (0.515) Assault type x Home 0.208 (0.348) 0.197 (0311) Assault type x Offender use I -0.147 (0.389) —0.941 (0.648) Assault type x Offender use 2 —0.192 (0.480) —0.709 (0.719) Assault type x Victim use —0589 (0.510) —0.060 (0.771) Assault type x Threat 0.046 (0.351) 0.496 (0515) -2 log likelihood —1191.641 Likelihood ratio z2 561.199 Of 58 p value .000 McFadden's R2 .191 a.lteported by the victim is the reference =terry:0u 'Scorn report sale not reported:0= victim report vs. I = dirdparty report. b. Offender substance use 1:0 = other cases. I = offender used substance:Offender substance use 2: 0 = other cases. I = vicems did not know whether offenders used substances. <.05. 3502-012 Page 11 of 18 EFIA_00001405 EFTA00156746 Chen and Unman 273 but no assault characteristics, had significant effects. Older age and being employed were related to greater reporting by victims than by third parties. To test whether influences on reporting differed by assault type, interaction terms were created by multiplying assault type (i.e.. 0=sexual wadi, 1 =physical assault) with each independent variable. The multinomial logit model was then reestimated by including all predictors and interaction terms. Results arc presented in Table 5. This composite model significantly predicted police reponing. LRX2(58,N= 2.072)=561.20,p <.001. However. addition of interaction terms did not improve prediction of reporting for either rapes or physical assaults, LRX2(28, N = 2,072) = 35.38. p = .159.1n other words, there was a high degree of similarity in correlates ofpolice reporting between rapes and physical assaults. Only one interaction term, assault type by physical force, was significant (left panel of Table 5). showing the effect of physical force on victim reporting decisions differed by assault type. Follow-up multinomial logit analyses for each assault type (results not shown) indicated that use of physical force had a significant effect on rape reporting but not on physical assault reporting. Specifically, physical force increased rapt reporting but actu- ally reduced physical assault reporting. With respect to third-party reporting. interactions of assault type by age at assault and assault type by marital status were significant at the .05 and .10 levels, respectively (right panel of Table 5). Follow-up analyses for each assault type comparing victim and third- party reporting suggested that age at assault was a significant predictor for rape reporting but not for physical assault reporting. whereas victim marital status had a marginal effect on physical assault reporting but not on rape reporting. As victim age at assault increased. the predicted probability of third-party reporting of physical assault remained nearly the same. whereas that of third-party reporting of sexual assault decreased. The assault type by marital status interaction showed that victims who were married had lower predicted prob- ability of reporting both sexual and physical assaults than victims who were not married. Married victims decreased the predicted probability ofthird-party reporting ofsexual assault but increased third-party reporting ofphysical assault. Discussion The present study used a national sample of women (NVAWS; Tjaden & Thoennes. 1999) to examine whether factors affecting rape reporting arc different from those affecting phys- ical assault reporting. an understudied issue in the literature on victim reporting to police. Initial analyses of correlates of reporting including assault type as a predictor showed that women who were older at the time of the assault were more likely to report. consistent with past research on physical assault reporting (Baumer. 2002; Felson et al.. 2002). Also con- sistent with prior rape (Bachman, 1998; Fisher et al.. 2003: Lizotte. 1985: Pino & Meier. 1999) and physical assault reporting findings (Baumer. 2002: Felson et al.. 2002). women without a college education or who were non-White were more likely to report to police. Women were more likely to report to police incidents perpetrated by strangers. with weap- ons, or victim injury, as in previous research on reporting of rape (Bachman. 1993. 1998; Felson & Pare. 2005; Fisher et al., 2003; Golding a al., 1989; Greenberg & Ruback. 1992; 3502-012 Page 12 of 18 EFIA_00001406 EFTA00156747 274 Violence Against Women 16(3) Lizette. 1985: Pino & Meier. 1999: Ruback & Nlenard. 2001: Russell & Bolen, 2000) and physical assault (Balmer, 2002; Pelson et al.. 1999; Pelson et al.. 2002). Women were also more likely to report when they were unsure of offender substance use, but more research is needed to interpret this result. Perceiving life threat during the assault also increased the likelihood of reporting, possibly because these victims were more distressed following the attack and/or viewed these assaults as more serious. This is a new finding and shows that both objective and subjective measures of assault severity predict police reporting. With respect to third-party reponing, exploratory analyses showed that victims were more likely than third parties to report to police if they were older or employed. More research is needed to explain these results, but it is possible that these women were more independent and likely to make their own decisions. Results showed that sexual assaults were less likely to be reported than physical assaults. consistent with government statistics (Bureau of Justice Statistics. 2003). However, con- trary to the hypothesis that assault characteristics would influence rape reporting more than physical assault reporting. multinomial logit regression analyses adding interactions of assault type indicated that both types of victims were equally likely to base their reporting decisions on personal demographics and assault characteristics. Only physical force dis- tinguished rape reporting from physical assault reporting: specifically, it increased rape reponing but reduced physical assault reporting. This may be due to rapes with force being more likely to be acknowledged (Littleton et al.. 2006) and thus reported. Conversely, phys- ical assaults may be likely to be domestic violence-related and thus less reponed, even with greater physical violence. These data suggest physical assault victims with more offender violence may be less likely to receive help from police and other needed services such as medical attention for their injuries. In terms of third-party reporting, the predictor of victim age at assault and marital status distinguished rape reporting from physical assault reporting. Whereas victim age was a significant predictor for rape reporting but not for physical assault reporting. marital status had a marginal effect on physical assault reporting but not on rape report- ing. As victim age at assault increased, the probability of reporting rapes by third panics decreased, which may be due to differences in lifestyle/routine activities of women of varying ages. Younger women may be more likely to be out with friends at parties where rapes occur. so third parties may be more likely to know about and to report these rapes. Younger women may also be more likely to tell informal support sources such as parents. who then report to police (Konradi, 1996). Women were less likely than third panics to report physical assaults when they were married. Most of these physical assaults may be domestic violence incidents that third panics may witness or hear about and/or women may view these assaults as private matters, less serious crimes, or be more fearful of perpetrator retaliation. The present study's findings contrast with the popular assumption that rape reporting is unique. There arc three possible explanations for similarities in correlates of victim report- ing of sexual and physical assaults. The first possibility is that rape reponing has never been different from physical assault reporting. However, except for Lixotte's (1985) study, no prior research compared reponingof sexual and physical assaults, so this is unknown.