U.S. Department of Justice Office of Justice Programs Bureau of Justice Statistics Sexual Victimization in Prisons and Jails Reported by Inmates, 2008-09 Allen J. Beck, PhD, and Paige M. Harrison, BJS Statisticians Marcus Berzofsky, Rachel Caspar, and Christopher Krebs, PhD, RTI International August 2010, NCJ 231169 ----------------------------------------------------------- ----------------------------------------------------------- This file is text only without graphics and many of the tables. A Zip archive of the tables in this report in spreadsheet format (.csv) and the full report including tables and graphics in .pdf format are available from: http://bjs.ojp.usdoj.gov/index.cfm?ty=pbdetail&iid=2202 This report is associated with the PREA Data Collection. For a list of related publications go to http://bjs.ojp.usdoj.gov/index.cfm?ty=dcdetail&iid=278 ------------------------------------------------------------ Bureau of Justice Statistics James P. Lynch Director BJS Website: www.bjs.usdoj.gov The Bureau of Justice Statistics is the statistics agency of the U.S. Department of Justice. James P. Lynch is director. This compendium was written by Allen J. Beck, PhD, and Paige M. Harrison, BJS Statisticians, and Marcus Berzofsky, Rachel Caspar, and Christopher Krebs, PhD, RTI International. Paige M. Harrison, under the supervision of Allen J. Beck, was the project manager for the NIS-2. RTI, International staff, under a cooperative agreement and in collaboration with BJS, designed the survey, developed the questionnaires, and monitored the data collection and processing, including Rachel Caspar, Principal Investigator/ Instrumentation Task Leader; Christopher Krebs, Co-principal Investigator; Ellen Stutts, Co-principal Investigator and Data Collection Task Leader; Susan Brumbaugh, Logistics Task Leader; Jamia Bachrach, Human Subjects Task Leader; David Forvendel, Research Computing Task Leader; Ralph Folsom, Senior Statistician; and Marcus Berzofsky, Statistics Task Leader. Jill Duncan edited the report, Barbara Quinn designed and produced the report, and Jayne Robinson prepared the report for final printing under the supervision of Doris J. James. August 2010, NCJ 231169 Sexual Victimization in Prisons and Jails Reported by Inmates, 2008-09 National Inmate Survey, 2008-09 Federal and State Prisons Local Jails Allen J. Beck, PhD, and Paige M. Harrison, BJS Statisticians Marcus Berzofsky, Rachel Caspar, and Christopher Krebs, PhD, RTI International August 2010, NCJ 231169 BJS Contents Highlights - 5 National Inmate Survey-2 - 6 Incidents of Sexual Victimization - 6 Facility Level Rates - 7 Demographic Characteristics - 12 Sexual History and Orientation - 14 Criminal Justice Status - 16 Predicted Victimization Rates - 18 Inmate-on-Inmate Victimization - 21 Staff Sexual Misconduct - 23 Appendix 1. Survey items related to inmate-on-inmate sexual victimization, National Inmate Survey, 2008-09 - 25 Appendix 2. Survey items related to staff sexual misconduct, National Inmate Survey, 2008-09 - 26 Appendix 3. Follow-up questions for inmates reporting no sexual activity, National Inmate Survey, 2008-09 - 26 Methodology - 27 List of Tables Table 1. Inmates reporting sexual victimization, by type of facility and incident, National Inmate Survey, 2008-09 - 7 Table 2. Facilities with high rates of inmate-on-inmate sexual victimization, by type of facility, National Inmate Survey, 2008-09 - 8 Table 3. Facilities with high rates of staff sexual misconduct, by type of facility, National Inmate Survey, 2008-09 - 9 Table 4. Facilities with low rates of any type of sexual victimization, by type of facility, National Inmate Survey, 2008-09 - 10 Table 5. Rates of sexual victimization in special correctional facilities, by type of incident and facility, National Inmate Survey, 2008-09 - 11 Table 6. Prevalence of sexual victimization, by type of incident and demographic inmate characteristic, National Inmate Survey, 2008-09 - 12 Table 7. Multivariate logistic regression models of sexual victimization, by inmate demographic characteristic, National Inmate Survey, 2008-09 - 13 Table 8 Prevalence of sexual victimization, by type of incident and sexual history, National Inmate Survey, 2008-09 - 14 Table 9. Multivariate logistic models of sexual victimization, by sexual history, National Inmate Survey, 2008-09 - 15 Table 10. Prevalence of sexual victimization, by type of incident and criminal justice status, National Inmate Survey, 2008-09 - 16 Table 11. Multivariate logistic models of sexual victimization, by criminal justice status and history, National Inmate Survey, 2008-09 - 17 Table 12. Final multivariate logistic models of sexual victimization in prisons and jails, by type of incident, National Inmate Survey, 2008-09 - 18 Table 13. Estimated and predicted rates of inmate-on-inmate sexual victimization in high-rate prisons and jails, National Inmate Survey, 2008-09 - 20 Table 14. Estimated and predicted rates of staff sexual misconduct in high-rate prisons and jails, National Inmate Survey, 2008-09 - 20 Table 15. Experiences of victims of inmate-on-inmate victims of sexual victimization, by type of facility and sex, National Inmate Survey, 2008-09 - 21 Table 16. Circumstances surrounding inmate-on-inmate sexual victimization, by type of facility and sex, National Inmate Survey, 2008-09 - 22 Table 17. Circumstances surrounding incidents of staff sexual misconduct, by type of facility and sex of victim, National Inmate Survey, 2008-09 - 23 Table 18. Sex of perpetrator of staff sexual misconduct, by facility type and sex of victim, National Inmate Survey, 2008-09 - 24 Table 19. Sexual touching between inmates and staff during and not during strip searches and pat downs, by sex of victim, National Inmate Survey, 2008-09 - 24 List of Appendix Tables Appendix Table 1. Characteristics of state and federal prisons and prevalence of sexual victimization, by facility, National Inmate Survey, 2008-09 - 34 Appendix Table 2. Percent of prison inmates reporting sexual victimization, by type of incident and facility, National Inmate Survey, 2008-09 - 41 Appendix Table 3. Percent of prison inmates reporting sexual victimization by level of coercion, by facility, National Inmate Survey, 2008-09 - 47 Appendix Table 4. Percent of prison inmates reporting nonconsensual sexual acts and abusive sexual contacts, by facility, National Inmate Survey, 2008-09 - 53 Appendix Table 5. Characteristics of jails and prevalence of sexual victimization, by facility, National Inmate Survey, 2008-09 - 59 Appendix Table 6. Percent of jail inmates reporting victimization, by type of incident and facility, National Inmate Survey, 2008-09 - 67 Appendix Table 7. Percent of jail inmates reporting sexual victimization, by level of coercion, and facility, National Inmate Survey, 2008-09 - 75 Appendix Table 8. Percent of jail inmates reporting nonconsensual sexual acts and abusive sexual contacts, by facility, National Inmate Survey, 2008-09 - 83 Appendix Table 9. Characteristics of special correctional facilities and prevalence of sexual victimization, by facility, National Inmate Survey, 2008-09 - 91 Appendix Table 10. Wald F statistics for inmate risk characteristics in the final multivariate logistic models of sexual victimization in prisons and jails, by type of incident, National Inmate Survey, 2008-09 - 92 August 2010 Highlights Prevalence of sexual victimization * An estimated 4.4% of prison inmates and 3.1% of jail inmates reported experiencing one or more incidents of sexual victimization by another inmate or facility staff in the past 12 months or since admission to the facility, if less than 12 months. Nationwide, these percentages suggest that approximately 88,500 adults held in prisons and jails at the time of the survey had been sexually victimized. * About 2.1% of prison inmates and 1.5% of jail inmates reported an incident involving another inmate. An estimated 1.0% of prison inmates and 0.8% of jail inmates said they had nonconsensual sex with another inmate (the most serious type of acts), including unwilling manual stimulation and oral, anal, or vaginal penetration. * About 2.8% of prison inmates and 2.0% of jail inmates reported having had sex or sexual contact with staff. At least half of the inmates who experienced staff sexual misconduct (1.8% in prison and 1.1% in jail) said that they willingly had sex or sexual contact with staff. Facility rankings * Eight male prisons, 2 female prisons, and 6 jails were identified as "high rate" facilities based on the prevalence of inmate-on-inmate sexual victimization; 4 male prisons, 2 female prisons, and 5 jails were identified as "high rate" based on the prevalence of staff sexual misconduct. Each of these facilities had a lower bound of the 95%-confidence interval that was at least 55% higher than the average rate among comparable facilities. * Seven male prisons, 4 female prisons, and 9 jails were identified as "low rate" facilities based on a small percentages of inmates reporting any sexual victimization by another inmate or staff and a low upper bound for the 95%-confidence interval around the rate. Among the 167 prisons and 286 jails in the survey, 6 prisons and 28 jails had no reported incidents of sexual victimization. * Except for a 6.0% rate of sexual victimization in the U.S. Disciplinary Barracks (Leavenworth, KS), rates in the 5 surveyed facilities operated by ICE, 3 operated by the U.S. Military, and 2 facilities in Indian country were lower than average rates in state and federal prisons (4.4%) and jails (3.1%). Variations in victimization rates * Rates of reported sexual victimization varied among inmates: * Female inmates in prison (4.7%) or jail (3.1%) were more than twice as likely as male inmates in prison (1.9%) or jail (1.3%) to report experiencing inmate-on-inmate sexual victimization. * Sexual activity with facility staff was reported by 2.9% of male prisoners and 2.1% of male jail inmates, compared to 2.1% of female prisoners and 1.5% of female jail inmates. * Rates of inmate-on-inmate sexual victimization in prisons and jails were significantly higher among inmates who were white or multi-racial compared to blacks, inmates with a college degree or more (compared to those who had not completed high school), a sexual orientation other than heterosexual compared to heterosexual, and who had experienced a sexual victimization before coming to the facility compared to those who had not. * After controlling for multiple inmate characteristics, rates of reported staff sexual misconduct were lower among white inmates (compared to black inmates), lower among inmates ages 25 or older (compared to inmates ages 20 to 24), higher among inmates with a college degree (compared to those who had not completed high school), and higher among inmates who had experienced sexual victimization before coming to the facility (compared to those who had not). Circumstances surrounding victimization * Among inmates who reported inmate-on-inmate sexual victimization, 13% of male prison inmates and 19% of male jail inmates said they were victimized within the first 24 hours after admission, compared to 4% of female inmates in prison and jail. * Inmate-on-inmate victimization in prisons and jails was most commonly reported to have occurred between 6 pm and midnight: more than 40% of victims reported this time period. * Most victims of staff sexual misconduct were males; most perpetrators were females. Among male victims of staff sexual misconduct, 69% of those in prison and 64% of those in jails reported sexual activity with female staff. An additional 16% of prison inmates and 18% of jail inmates reported sexual activity with both female and male staff. Among inmates who reported staff sexual misconduct, nearly 16% of male victims in prison and 30% of male victims in jail said they were victimized by staff within the first 24 hours, compared to 5% of female victims in prison and 4% of female victims in jail. Sexual Victimization in Prisons and Jails Reported by Inmates, 2008 09 National Inmate Survey-2 Between October 2008 and December 2009, BJS completed the second National Inmate Survey (NIS-2) in 167 state and federal prisons, 286 jails, and 10 special confinement facilities operated by Immigration and Customs Enforcement (ICE), the U.S. Military, and correctional authorities in Indian country. The survey, conducted by RTI International (Research Triangle Park, NC), was administered to 81,566 inmates ages 18 or older, including 32,029 inmates in state and federal prisons, 48,066 in jails, 957 in ICE facilities, 399 in military facilities, and 115 in Indian country jails. The NIS-2 is part of the National Prison Rape Statistics Program, which collects administrative records of reported sexual violence, and allegations of sexual victimization directly from victims, through surveys of adult inmates in prisons and jails and surveys of youth held in juvenile correctional facilities. Administrative records have been collected annually since 2004. Reports by victims of sexual victimization have been collected since 2007. The NIS-2 survey consisted of an audio computer-assisted self-interview (ACASI) in which inmates, using a touch-screen, interacted with a computer-assisted questionnaire and followed audio instructions delivered via headphones. Some inmates (726) completed a short paper form. Most of these inmates were housed in administrative or disciplinary segregation or were considered too violent to be interviewed. The NIS-2 collects only allegations of sexual victimization. Because participation in the survey is anonymous and reports are confidential, the survey does not permit any follow-up investigation or substantiation of reported incidents through review. Some allegations in the NIS-2 may be untrue. At the same time, some inmates may remain silent about sexual victimization experienced in the facility, despite efforts of survey staff to assure inmates that their responses would be kept confidential. Although the effects may be offsetting, the relative extent of under reporting and false reporting in the NIS-2 is unknown. Incidents of Sexual Victimization 4.4% of prison inmates and 3.1% of jail inmates reported one or more incidents of sexual victimization Among the 76,459 inmates participating in the NIS-2 sexual victimization survey, 2,861 reported experiencing one or more incidents of sexual victimization in the past 12 months, or since admission to the facility, if less than 12 months. Because the NIS-2 is a sample survey, weights were applied for sampled facilities and inmates within facilities to produce national-level and facility-level estimates. The estimated number of prison and jail inmates experiencing sexual victimization totaled 88,500 (or 4.4% of all prison inmates and 3.1% of jail inmates, nationwide) (table 1). Among all state and federal inmates, 2.1% (or an estimated 30,100 prisoners) reported an incident involving another inmate, and 2.8% (41,200) reported an incident involving facility staff. Some prisoners (0.5%) reported sexual victimization by both another inmate and facility staff. About 1.5% of jail inmates (11,600) reported an incident with another inmate, and 2.0% (15,800) reported an incident with staff. Approximately 0.4% of jail inmates (3,400) reported being sexually victimized by both other inmates and staff. The NIS-2 screened for specific sexual activities in which inmates may have been involved during the past 12 months or since admission to the facility, if less than 12 months. Inmates were then asked if they were forced or pressured to engage in these activities by another inmate or staff. (See appendices 1 through 3 for specific survey questions.) Reports of inmate-on-inmate sexual victimization were classified as either nonconsensual sexual acts or abusive sexual contacts. Approximately 1.0% of prisoners and 0.8% of jail inmates said they were forced or pressured to have nonconsensual sex with another inmate, including manual stimulation and oral, anal, or vaginal penetration. An additional 1.0% of prison inmates and 0.7% of jail inmates said they had experienced one or more abusive sexual contacts only, or unwanted touching of specific body parts in a sexual way by another inmate. (See page 7 for definition of terms.) An estimated 1.7% of prison inmates and 1.5% of jail inmates reported that they had sex or sexual contact unwillingly with staff as a result of physical force, pressure, or offers of special favors or privileges. An estimated 1.8% of all prison inmates and 1.1% of jail inmates reported they willingly had sex or sexual contact with staff. Regardless of whether an inmate reported being willing or unwilling, any sexual contact between inmates and staff is illegal; however, the difference may be informative when addressing issues of staff training, prevention, and investigation. Facility Level Rates NIS-2 provides a basis for identifying high rate and low rate facilities As required under the Prison Rape Elimination Act, the NIS-2 provides facility-level estimates of inmateon-inmate sexual victimization and staff sexual misconduct. Since these estimates are based on a sample of inmates rather than a complete enumeration, they are subject to sampling error. (See Methodology for description of sampling procedures.) The precision of each of the facility-level estimates can be calculated based on the estimated standard error. Typically, a 95%-confidence interval around each survey estimate is calculated by multiplying the standard error by 1.96 and then adding and subtracting the result from the sample estimate to create an upper and lower bound. This interval expresses the range of values that could result among 95% of the different samples that could be drawn. For small samples and estimates close to 0%, as is the case with facility-level estimates of sexual victimization by type of incident, the use of the standard error to construct the 95%-confidence interval may not be reliable. An alternative method developed by Wilson has been shown to perform better than the traditional method. ***Footnote 1 Brown, L.D., Cai, T., and DasGupta, A. (2001). Interval Estimation for a Binomial Proportion. Statistical Science, 16(2), pp. 101-138. Wilson, E.B. (1927). Probable Inference, the Law of Succession, and Statistical Inference. Journal of the American Statistical Association, 22, pp. 209-212***. This method provides asymmetrical confidence intervals for facilities in which the lower bound is constrained to be no less than 0%. It also provides confidence intervals for facilities in which the survey estimates are 0% (but other similarly conducted samples could yield non-zero estimates). Definition of terms Sexual victimization--all types of sexual activity, e.g., oral, anal, or vaginal penetration; hand jobs; touching of the inmate's buttocks, thighs, penis, breasts, or vagina in a sexual way; abusive sexual contacts; and both willing and unwilling sexual activity with staff. Nonconsensual sexual acts--unwanted contacts with another inmate or any contacts with staff that involved oral, anal, vaginal penetration, hand jobs, and other sexual acts. Abusive sexual contacts only--unwanted contacts with another inmate or any contacts with staff that involved touching of the inmate's buttocks, thigh, penis, breasts, or vagina in a sexual way. Unwilling activity--incidents of unwanted sexual contacts with another inmate or staff. Willing activity--incidents of willing sexual contacts with staff. These contacts are characterized by the reporting inmates as willing; however, all sexual contacts between inmates and staff are legally nonconsensual. Staff sexual misconduct--includes all incidents of willing and unwilling sexual contact with facility staff and all incidents of sexual activity that involved oral, anal, vaginal penetration, hand jobs, blow jobs, and other sexual acts with facility staff. Although the NIS-2 provides facility-level estimates and measures of precision, it cannot provide an exact ranking for all facilities as required under PREA. Rates of inmate-oninmate sexual victimization and staff sexual misconduct differ across facilities, but the observed differences are not always statistically significant. To address PREA requirements, facilities have been categorized as having high rates or low rates based on criteria applied to the lower and upper bounds of the 95%-confidence interval for each facility. The criterion that the lower bound of the confidence interval be at least 55% higher than the average rate for comparable facilities was used to identify high rate male prisons, female prisons, and jails. The criterion that the upper bound of the confidence interval be lower than 65% of the average rate for comparable facilities was used to identify low rate facilities. To better identify variations among correctional facilities in rates of sexual victimization, prisons and jails are compared separately by type of sexual victimization. Though informative, an analysis of a single, overall prevalence rate of sexual victimization for each sampled facility would confound differing risk factors, circumstances, and underlying causes of victimization. For the same reasons, prisons are compared separately by the sex of inmates housed. The NIS-2 sample was designed to ensure a sufficient number of female-only prison facilities (35 facilities participated) and a sufficient number of female respondents (6,279 completed the survey) to allow for valid comparisons among female prisons. Only 2 of the 286 participating jails in NIS-2 housed only females; as a result, rates of sexual victimization in jails could not be compared separately by sex of inmate housed. Eight male prisons, 2 female prisons, and 6 jails were identified as having high rates of inmate-on-inmate sexual victimization Among the 167 prisons and 286 jails surveyed in NIS-2, 8 male prisons, 2 female prisons, and 6 jails were designated as high rate facilities based on reports of inmate-on-inmate sexual victimization (table 2). Each of these facilities had a rate of inmate-on-inmate sexual victimization that was at least twice the national rate of 1.9% for male prisons, 4.8% for female prisons, and 1.5% for jails. Each had a 95%-confidence interval with a lower bound that was at least 55% higher than the average rate among comparable facilities. Selection of slightly lower criteria would have had only a minor impact on the list of facilities with high rates of inmate-on-inmate sexual victimization. Lowering the criteria of the lower bound to at least 50% higher than the average comparable rate would not have increased the number of high rate facilities (16); lowering the criteria to 35% would have increased the number to 22 (including 10 male prisons, 2 female prisons, and 10 jails). Among male prisons, Hughes Unit (Texas) recorded an inmate-on-inmate sexual victimization rate of 8.6%, and Allred Unit (Texas) recorded a rate of 7.6%. Among female prisons, Taycheedah Correctional Institution (Wisconsin) had a rate of 11.9%, and Fluvanna Correctional Center (Virginia) had a rate of 11.4%. Orleans Parish -- South White Street Jail (Louisiana), a female-only facility, recorded an inmate-on-inmate sexual victimization rate of 7.5%, which was 5 times the average rate among jails nationwide. Madison County Detention Facility (Alabama) reported a rate of 5.5%. Four male prisons, 2 female prisons, and 5 jails were identified as having high rates of staff sexual misconduct Eleven facilities were identified as high rate facilities based on reports of staff sexual misconduct -- 4 male prisons, 2 female prisons, and 5 jails (table 3). Each had a confidence interval with a lower bound that was at least 55% higher than the national rate of male prisons (2.9%), female prisons (2.2%), and jails (2.0%). Selection of slightly lower criteria would have had only a minor impact on the list of facilities with high rates of staff sexual misconduct. Lowering the criteria of the lower bound to at least 50% higher than the average comparable rate would have increased the number of high rate facilities from 11 to 14 (including 5 male prisons, 2 female prisons, and 7 jails); lowering the criteria to 35% would have increased the number to 20 (including 6 male prisons, 2 female prisons, and 12 jails). In 3 state prisons, at least 8% of surveyed inmates reported incidents of staff sexual misconduct, including 8.2% of males in Crossroads Correctional Facility (Missouri), 8.1% of males in Attica Correctional Facility (New York), and 11.5% of females in Bayview Correctional Facility (New York). Two jails, Caroline County Jail (Maryland) with 10.0% and Eastern Shore Regional Jail (Virginia) with 9.9%, had rates of reported staff sexual misconduct that exceeded 8%. The reported use or threat of physical force to engage in sexual activity with staff was generally low among all prison and jail inmates (1.0%); however, at least 5% of the inmates in 2 state prisons and 2 jails said that they had been physically forced or threatened with force. Caroline County Jail (Maryland), with 10%, had the highest percentage of inmates reporting physical force or threat force by staff, followed by Bayview Correctional Facility (New York) and Attica Correctional Facility (New York), with 6.5% and 6.4%, respectively. An estimated 1.6% of prison inmates and 1.3% of jail inmates also reported being coerced by facility staff without any use or threat of force, including being pressured or made to feel they had to have sex or sexual contact. In 6 of the 11 facilities with high rates of staff sexual misconduct, at least 5% of the inmates reported such pressure by staff. Among state prisoners, the highest rates were reported by female inmates (10.8%) in the Bayview Correctional Facility (New York) and by male inmates (7.1%) in the Elmira Correctional Facility (New York). Among jail inmates, the highest rates were reported by inmates in the Caroline County Jail (7.9%). Seven male prisons, 4 female prisons, and 9 jails were identified as "low rate" for sexual victimization overall Six prisons and 28 jails had no reported incidents of sexual victimization of any kind. (See appendix tables 1 and 5.) However, estimates of the number of inmates who experienced a sexual victimization in each of these facilities are also subject to sampling error and could vary if a different group of inmates had been interviewed. Although the lower bound of the 95%-confidence interval in each of these facilities is 0%, the upper bound varies depending on the number of completed interviews in each facility. Combining reports of inmate-on-inmate sexual victimization and staff sexual misconduct, 7 male prisons and 4 female prisons were designated as low rate facilities. These designations were based on their low rate of sexual victimization overall and the upper bound of their 95%-confidence interval that was less than 65% of the average rate among male and female prisons (table 4). Four of these facilities had no reported incidents of sexual victimization; 7 had at least one inmate who reported a sexual victimization. C. Moore Transfer Facility (Texas), with a reported sexual victimization rate of 0.4%, had a confidence interval with the lowest upper bound (1.9%) among male prisons. Halbert Substance Abuse Felony Punishment Facility (Texas), with reported sexual rate of 0.9%, had a confidence interval with the lowest upper bound (2.5%) among female prisons. Nine jails were designated as low rate facilities based on the upper bound of the 95%-confidence interval that was less than 65% of the average for jails nation-wide. Hinds County Penal Farm (Mississippi) and Lake County - Adult Maximum Security Detention Center (Ohio), both with a 0.5% overall sexual victimization rate, had confidence intervals with the lowest upper bounds (1.6%). Low rates of sexual victimization were reported in military, Indian country, and ICE facilities The NIS-2 also surveyed 10 special confinement facilities including 5 ICE facilities, 3 military facilities, and 2 Indian country jails. (See Methodology for sample description.) Except for a 6.0% overall rate of sexual victimization in the U.S. Disciplinary Barracks (Leavenworth, KS), rates in these facilities were lower than the average rates in state and federal prisons (4.4%) and jails (3.1%). (Not shown. See appendix table 9.) Reports of inmate-on-inmate sexual victimization were highest in the El Paso Processing Center (Texas), operated by ICE; however, its rate of 2.1% equaled the average rate among prisoners nationwide (table 5). The U.S. Disciplinary Barracks, which is operated by the U.S. Army and holds the most serious offenders under military jurisdiction, had a rate of staff sexual misconduct (5.6%) that was double the average of prisons nationwide (2.8%). Gila River Department of Corrections and Rehabilitation (Arizona), the largest jail in Indian country, had no reports of inmate-on-inmate sexual victimization and one report of staff sexual misconduct (1%). Demographic Characteristics Rates of reported sexual victimization varied across demographic categories of prison and jail inmates * Rates of inmate-on-inmate sexual victimization among prison inmates were higher among females (4.7%) than males (1.9%), higher among whites (3.0%) or multi-racial (4.4%) than among blacks (1.3%), higher among inmates with a college degree (3.4%) than among inmates who had not completed high school (2.0%), and lower among currently married inmates (1.3%) than among inmates who never married or who were widowed, divorced, or separated (2.2%) (table 6). * Similar patterns of inmate-on-inmate sexual victimization were reported by jail inmates. Females (3.1%), whites (1.5%), and inmates with a college degree reported higher rates of victimization (2.9%) than males (1.3%), blacks (1.2%), and inmates who had not completed high school (1.3%). * Rates of inmate-on-inmate sexual victimization were unrelated to age among state and federal prisoners, but were lower among jail inmates in older age categories (ages 35 to 44, 45 to 54, and 55 or older) than among inmates ages 20 to 24. * Patterns of staff sexual misconduct were different--with reports of staff sexual misconduct being higher among males in prisons (2.9%) and jails (2.1%) than among females in prisons (2.1%) and jails (1.5%), and higher among black inmates in prisons (3.2%) and jails (2.4%) than among white inmates in prisons (2.3%) and jails (1.5%). * In both prisons and jails, rates of reported staff sexual misconduct were lower among inmates in the oldest age categories (ages 45 to 54 and ages 55 or older) compared to inmates ages 20 to 24. These variations in rates of sexual victimization among demographic groups are statistically independent and largely unexplained by covariation with other demographic characteristics. Multivariate logistic regression was used to determine which demographic characteristics are statistically significant for predicting a sexual victimization, while simultaneously controlling for the effects of other inmate demographic characteristics. (See Methodology for discussion of logistic regression.) Results are displayed in terms of their conditional predicted probability, which represents the probability that an inmate with a particular characteristic has experienced a given sexual victimization outcome conditional on the inmate having the mean value for all other predictors in the model (table 7). For example, based on models with demographic characteristics only, a female prison inmate has a 3.8% chance of being sexually victimized by another inmate, while a male inmate has a 1.6% chance (given that the inmates are at the mean of the joint distribution of race or Hispanic origin, education level, and marital status).***Footnote 2 These estimates represent the expected risk of victimization for an inmate, conditional on the inmate belonging to a particular group (defined by each characteristic in the final model) and having the mean value on all of the other characteristics in the model. For characteristics that are categorical (which is the case for every variable in the NIS-2 logistic regression models), the mean value is a weighted value of the joint distribution of all other characteristics in the respective model. See Research Triangle Institute (2008). SUDAAN Language Manual Release 10.0. Research Triangle Park, NC, Section 4.8.3, pp. 209-211 *** Based on four separate models, each representing the type of sexual victimization in prison and jail, variations by sex, race, Hispanic origin, and education remain statistically significant. Except for reports of inmate-on-inmate sexual victimization in prison, older inmates (ages 45 and older in prison and ages 35 and older in jails) still have lower rates of sexual victimization, after controlling for the effects of the other demographic characteristics. Among prison inmates, never married inmates remain somewhat more likely than married inmates to report sexual victimization; however, among jail inmates, there are no differences by marital status. Sexual History and Orientation Large differences in sexual victimization were found among inmates based on their sexual orientation and past sexual experiences Inmates with a sexual orientation other than heterosexual reported significantly higher rates of inmate-on-inmate sexual victimization and staff sexual misconduct: * Among heterosexual state and federal prisoners, an estimated 1.3% reported being sexually victimized by another inmate, and 2.5% reported being victimized by staff (table 8). In contrast, among prison inmates with a sexual orientation other than heterosexual (including bisexual, homosexual, gay or lesbian, or other), 11.2% reported being sexually victimized by another inmate, and 6.6% reported being sexually victimized by staff. * Similar differences were reported among jail inmates, with heterosexual inmates reporting lower rates of inmate-on-inmate victimization (1.1%) and staff sexual misconduct (1.9%) than non-heterosexual inmates (7.2% and 3.5%, respectively). * Inmates who had experienced sexual victimization before coming to the facility were also more likely than inmates with no sexual victimization history to report incidents of sexual victimization involving other inmates and staff. Among inmates who had experienced sexual victimization before coming to the facility, 11.0% of prisoners and 7.4% of jail inmates reported having been sexually assaulted by another inmate at the current facility. An estimated 8.7% of prisoners and 6.1% of jail inmates who had experienced sexual victimization before coming to the facility reported sexual activity with staff. * Prison and jail inmates with 21 or more sexual partners prior to coming to the current facility reported higher rates of staff sexual misconduct than inmates with 1 or no prior sexual partners. Variations in rates of sexual victimization among groups of inmates based on their sexual orientation and past sexual experiences overlapped somewhat. After simultaneously controlling for the effects of these characteristics, the regression models reveal that variations in inmate-on-inmate sexual victimization remain associated with sexual orientation and prior sexual victimization, but are not associated with the number of past sexual partners (table 9). Except for reports of staff sexual misconduct in jails, an inmate's sexual orientation remained an important predictor of victimization. In all models, inmates who had experienced sexual victimization before coming to the facility were more likely than inmates with no sexual victimization history to report incidents of sexual victimization. Criminal Justice Status Inmates held for a violent sexual offense reported higher rates of inmate-on-inmate sexual victimization than inmates held for other offenses An estimated 4.6% of violent sex offenders in prison and 3.9% of violentsex offenders in jail reported being sexually victimized by another inmate in the last 12 months or since admission to the facility, if less than 12 months (table 10). These rates were higher than those reported by inmates held for other offenses. Among state and federal prisoners, rates of inmate-on-inmate sexual victimization were-- * higher among prison inmates serving a long sentence (2.9% with a sentence of 20 or more years and 3.8% with a sentence to life or death) than among inmates serving a sentence of 1 to 5 years (1.5%). * higher among prison inmates who had served 5 years or more in prison prior to coming to the current facility (2.6%) than among inmates who had not served any prior time (1.6%). * higher among prison inmates who had been at their current facility for 5 years or more (3.3%) than among inmates who had been admitted in the last month (1.4%). Among jail inmates, rates of inmateon-inmate sexual victimization-- * were higher among first time offenders (2.0%) than among those who had been arrested 2 to 3 times in the past (1.3%) or 4 to 10 times in the past (1.4%). * increased with the length of time served in the current facility, rising from 0.9% among inmates who had been at the facility for less than a month to 1.7% among inmates in jail for 1 to 5 months, to 2.0% among inmates in jail for 6 to 11 months, and to 2.3% among those in jail for 1 to 5 years. Rates of staff sexual misconduct also varied among inmates based on their criminal justice status and history * Among state and federal prisoners, inmates with a long sentence, inmates who had served 5 years or more in prison prior to coming to the current facility, and inmates who had served 5 years or more at the current facility were more likely to report experiencing staff sexual misconduct than inmates with a sentence of 1 to 5 years, inmates who had not served any prior time, and inmates who had been admitted in the last month. * Among jail inmates, the rates of reported staff misconduct increased with time served in the current facility and were higher among inmates who had previously served time in a correctional facility for 1 year or more. Based on controls for the criminal justice status variables and past history variables, the regression models reveal that rates of inmate-on-inmate sexual victimization remain higher for violent sex offenders in prison and jail. A violent sex offender has a 4.4% chance of experiencing sexual victimization by another inmate in prison and 3.1% chance of being sexually victimized by another inmate in jail. Inmate-on-inmate sexual victimization rates for inmates held for other offenses are significantly lower (table 11). In each of the regression models for jails, rates of sexual victimization remain associated with an inmate's time since admission and prior time served in a correctional facility. Jail inmates who had served longer (i.e., 6 months to 5 years in the current facility and 6 months or more in other facilities) had statistically significant higher rates of staff sexual misconduct than jail inmates who had served less time (i.e., less than one month in the current facility and no time in other facilities in the past). Predicted Victimization Rates Sexual victimization among high rate facilities only partially explained by variations in risk among inmates held The variations in rates of sexual victimization among inmates grouped by demographic characteristics, sexual orientation and history, and criminal justice status and history provide a basis for predicting the probability of victimization for each inmate. The final multivariate logistic regression models summarize the net contribution of each characteristic to the predicted probability of victimization, by type of victimization in prison and jail (table 12). Because the effects of each characteristic overlap, the estimated conditional probabilities are smaller than those in previous regression models. However, the net effects of sexual orientation and prior sexual victimization remain the largest among all characteristics in predicting inmate-on-inmate victimization in prison and jail. An inmate's race (black), age (ages 20 through 24), and education (college degree or more) are found to increase the probability of experiencing staff sexual misconduct, controlling for other factors. Taking into account the effects of sexual history and orientation and criminal justice status and history, female inmates have lower rates of sexual victimization than male inmates in both prison and jail and for both types of victimization. Predicted rates of sexual victimization at the facility level have been calculated by combining all of the inmate-level characteristics into a final logistic regression model for each type of sexual victimization. Based on the predicted conditional probabilities of victimization of each inmate (derived from the multivariate logistic regression models) and the distribution of inmates in each facility, predicted rates of sexual victimization have been calculated for each of the high rate facilities. These predicted rates take into account the underlying inmate risk factors and the variations in the distribution of inmates within each of the facilities. At the facility level, the difference between the predicted rate and observed rate represents the percent not accounted for by variations in inmate characteristics. (See Methodology for facility-level calculations.) ----------------------------------------------------------- Logistic regression models Multivariate logistic regression estimation is a modeling technique used to determine what characteristics are statistically significant for predicting a dichotomous outcome (e.g., victimized or not victimized) while controlling for all the other characteristics in the model. NIS-2 used this technique to determine what inmate-level characteristics were significant predictors of inmate-on-inmate sexual victimization and staff sexual misconduct. In each of the logistic regression models, the conditional predicted probability represents the probability that an inmate with a particular characteristic has experienced sexual victimization (by type) conditional on the inmate having the mean value for all other predictors in the model. For example, based on demographic characteristics only, a female prison inmate has a 3.8% chance of being victimized by another inmate given that she was at the mean of the joint distribution of race or Hispanic origin, education level, and marital status. (See table 7 and note 2 on page 13.) (See Methodology for full discussion on logistic regression models.) ----------------------------------------------------------- Among high rate male prisons, the observed rate of inmate-on-inmate sexual victimization was 6.7%, while the predicted rate was 3.6% (table 13). Among high rate female prisons, observed rate was 11.6%, while the the predicted rate was 8.0%. Among high rate jails, the observed rate was 4.9%, while the predicted rate was 1.9%. Similar patterns are found for rates of staff sexual misconduct among high rate facilities. In the 4 high rate male prisons, the observed rate of staff sexual misconduct was 7.8%, while the the predicted rate was 4.1% (table 14).In the 2 high rate female prisons, the observed rate was 8.1%, while the predicted rate was 2.6%. In the 5 high rate jails, the observed rates of 7.4%, while predicted rate was 2.6%. These data suggest that rates among the 16 facilities with high rates of inmate-on-inmate sexual victimization and in 11 facilities with high rates staff sexual misconduct are only partially explained by variation in inmate demographic characteristics, sexual orientation and past sexual experience,and criminal justice status and history. Differences between the observed and predicted rates were statistically significant at the 95% level of confidence in only 3 of 16 facilities with high rates of inmates-on-inmate sexual victimization and in 1 of the 11 facilities with high rates of staff sexual misconduct. However the for tests for statistical significance within each facility were limited by the small number of surveyed inmates for whom the predicted and observed rates of victimization differed. When the number of inmates in all high rate male prisons, female prisons, and jails were combined, the differences were found to be statistically significant. With the exception of 2 facilities--the Michael Unit (in Texas) and Maine State Prison - Warren--the differences between the observed and predicted rates were large relative to the average rates of sexual victimization among male prisons, female prisons, and jails nationwide. Inmate on-Inmate Victimization Reports of inmate-on-inmate sexual victimization differed among males and females by type of facility Among inmates who reported inmate-on-inmate sexual victimization in state and federal prisons-- * Males (16%) were more likely than females (6%) to have been victimized 11 or more times in the last 12 months, or since admission if less than 12 months (table 15). * Males were more likely than females to report having been bribed or blackmailed to take part in the sexual activity (42% compared to 26%), offered protection (39% compared to 19%), or threatened with harm or a weapon (48% compared to 30%). * Males were more likely than females to report more than one perpetrator (25% compared to 11%), that the perpetrator was of Hispanic or Latino origin (24% compared to 16%), and that one or more incidents were initiated by a gang (20% compared to 4%). Among victims of inmate-oninmate sexual violence in jails-- * Females were more likely than males to have been victimized only once (56% compared to 36%) and less likely to have been victimized 11 or more times (9% compared to 20%). * Males were more likely than females to report all forms of pressure or force (except for being persuaded or talked into it). * Males were more likely than females to report more than one perpetrator (43% compared to 16%), that the perpetrator was of Hispanic or Latino origin (37% compared to 17%), and that one or more incidents were initiated by a gang (36% compared to 6%). Inmate-on-inmate victimization occurred most often in the victim's cell between 6 pm and midnight * Among inmates who reported inmate-on-inmate sexual victimization, males were much more likely than females to experience sexual victimization by another inmate within the first 24 hours after admission. Among inmates reporting victimization by another inmate, 13% of male prison inmates and 19% of male jail inmates said they were victimized within the first 24 hours, compared to 4% of female inmates in prison and jail (table 16). * In both prisons and jails inmate-on-inmate victimization was most commonly reported to have occurred between 6 pm and midnight. More than 40% of male and female victims reported this time period. * In both prisons and jails inmate-on-inmate victimization was most commonly reported to have occurred in the victim's cell or sleeping area. Among victims in prison, more than half of the male and female victims reported an incident in their cell. Among victims in jail, 63% of male victims and 43% of female victims reported at least one incident occurred in their cell or sleeping area. * Among inmates who reported inmate-on-inmate sexual victimization in jail, 37% of males reported being injured, compared to 8% of females. In prison, males and females were almost equally as likely to report being injured (21% and 17%, respectively) during the sexual victimization. * Male and female inmates who experienced inmate-on-inmate sexual victimization in prisons and jails most commonly reported sustaining bruises, scratches, cuts, and other minor injuries. Among victims in prison, anal or vaginal tearing was reported by nearly 10% of males and 6% of females. Among victims in jails, anal or vaginal tearing was reported by 21% of males and 3% of females. Staff Sexual Misconduct Most victims of staff sexual misconduct in prisons and jails reported at least one incident that involved pressure, force, or threat of force * Among victims in prison, male victims of staff sexual misconduct (64%) were more likely than female victims (30%) to report incidents that involved no pressure or force (table 17). A similar pattern was reported by victims in jail, with an estimated 56% of male victims and 31% of female victims reporting one or more incidents that involved no pressure or force by staff. * Nearly 82% of the female victims in prison said they were pressured by staff to engage in sexual activity, compared to 55% of male victims in prison. * In both prisons and jails, male victims were more likely than female victims to report that the first incident of staff sexual misconduct occurred within the first 24 hours following admission to the facility. Nearly 16% of male victims in prison and 30% of the male victims in jail said they were victimized within the first 24 hours, compared to 5% of the female victims in prison and 4% of female victims in jail. * In prisons, both male and female victims reported that the staff sexual misconduct was most likely to have occurred in a closet or locked office, the victim's cell or sleeping area, or the shower or bathroom area. * In jails, 45% of male victims and 24% of female victims identified a closet or locked office as the most common area in which the staff sexual misconduct occurred. * Among victims of staff sexual misconduct in jail, 17% of male victims and 8% of female victims reported they had been injured during the incident. * Among victims of staff sexual been injured by staff; however, misconduct in prison, 9% of males the difference was not statistically and 19% of females said they had significant. Most victims and perpetrators of staff sexual misconduct were of the opposite sex Among the 39,121 male prison inmates who had been victims of staff sexual misconduct, 69% reported sexual activity with female staff; an additional 16% reported sexual activity with both female and male staff (table 18). In comparison, among the 2,123 female prison inmates who had been victimized, 72% reported that the staff perpetrator was male; an additional 19% reported both male and female staff. Similar patterns of staff sexual misconduct were reported by jail inmates. Nearly two-thirds of the male jail inmates who had been victimized said the staff perpetrator was female (64%). About the same percentage (63%) of female victims said the perpetrator was a male staff member at the jail. Reports of staff sexual misconduct were linked to strip searches and pat downs Victims of staff sexual misconduct were asked if they touched a facility staff person's body or had their body touched in a sexual way. Regardless of whether they had wanted it to occur or not, nearly two-thirds of all victims of staff sexual misconduct in prison and jail reported at least one incident of sexual touching. An estimated 38,270 inmates reported such touching in the last 12 months or since admission to the facility, if less than 12 months (table 19). At least 4 of every 10 of these victims said that this had happened at least once as part of a strip search or a pat down. Among victims of sexual touching, male inmates (43%) were as likely as female inmates (40%) to report staff sexual misconduct as part of a strip search or pat down. The majority of victims said they had also touched staff or been touched by staff in a sexual way outside of a strip search or pat down. An estimated 86% of male victims and 91% of female victims in prisons and jails said that this had happened at least once outside of a strip search or pat down. Appendix 1. Survey items related to inmate on-inmate sexual victimization, National Inmate Survey, 2008-09 Males E16. During the last 12 months, did another inmate use physical force to touch your butt, thighs, or penis in a sexual way? E17. During the last 12 months, did another inmate, without using physical force, pressure you or make you feel that you had to let them touch your butt, thighs, or penis in a sexual way? E22. During the last 12 months, did another inmate use physical force to make you give or receive a hand job? E23. During the last 12 months, did another inmate, without using physical force, pressure you or make you feel that you had to give or receive a hand job? E26. During the last 12 months, did another inmate use physical force to make you give or receive oral sex or a blow job? E27. During the last 12 months, did another inmate, without using physical force, pressure you or make you feel that you had to give or receive oral sex or a blow job? E32. During the last 12 months, did another inmate use physical force to make you have anal sex? E33. During the last 12 months, did another inmate, without using physical force, pressure you or make you feel that you had to have anal sex? E34. During the last 12 months, did another inmate use physical force to make you have any type of sex or sexual contact other than sexual touching, hand jobs, oral sex or blow jobs, or anal sex? E35. During the last 12 months, did another inmate, without using physical force, pressure you or make you feel that you had to have any type of sex or sexual contact other than sexual touching, hand jobs, oral sex or blow jobs, or anal sex? Females E18. During the last 12 months, did another inmate use physical force to touch your butt, thighs, breasts, or vagina in a sexual way? E19. During the last 12 months, did another inmate, without using physical force, pressure you or make you feel that you had to let them touch your butt, thighs, breasts, or vagina in a sexual way? E24. During the last 12 months, did another inmate use physical force to make you give or receive oral sex? E25. During the last 12 months, did another inmate, without using physical force, pressure you or make you feel that you had to give or receive oral sex? E28. During the last 12 months, did another inmate use physical force to make you have vaginal sex? E29. During the last 12 months, did another inmate, without using physical force, pressure you or make you feel that you had to have vaginal sex? E32. During the last 12 months, did another inmate use physical force to make you have anal sex? E33. During the last 12 months, did another inmate, without using physical force, pressure you or make you feel that you had to have anal sex? E34. During the last 12 months, did another inmate use physical force to make you have any type of sex or sexual contact other than sexual touching, oral sex, vaginal sex, or anal sex? E35. During the last 12 months, did another inmate, without using physical force, pressure you or make you feel that you had to have any type of sex or sexual contact other than sexual touching, oral sex, vaginal sex, or anal sex? Appendix 2. Survey items related to staff sexual misconduct, National Inmate Survey, 2008-09 These next questions are about the behavior of staff at this facility during the last 12 months. By staff we mean the employees of this facility and anybody who works as a volunteer in this facility. G4. During the last 12 months, have any facility staff pressured you or made you feel that you had to let them have sex or sexual contact with you? G5. During the last 12 months, have you been physically forced by any facility staff to have sex or sexual contact? G7. During the last 12 months, have any facility staff offered you favors or special privileges in exchange for sex or sexual contact? G2. During the last 12 months, have you willingly had sex or sexual contact with any facility staff? G11. [IF G2 OR G4 OR G5 OR G7 = Yes] During the last 12 months, which of the following types of sex or sexual contact did you have with a facility staff person? G11a. You touched a facility staff person's body or had your body touched in a sexual way. G11b. You gave or received a hand job. G11c. You gave or received oral sex or a blow job. G11d. You had vaginal sex. G11e. You had anal sex. Appendix 3. Follow-up questions for inmates reporting no sexual activity, National Inmate Survey, 2008-09 Follow-up questions for inmates reporting no sexual activity in the screener questions for sexual activity with inmates: LCM1. During the last 12 months, did another inmate use physical force, pressure you, or make you feel that you had to have any type of sex or sexual contact? LCM2. How long has it been since another inmate in this facility used physical force, pressured you, or made you feel that you had to have any type of sex or sexual contact? 1. Within the past 7 days 2. More than 7 days ago but within the past 30 days 3. More than 30 days ago but within the past 12 months 4. More than 12 months ago 5. This has not happened to me at this facility LCM3. [If Male] During the last 12 months, did another inmate use physical force, pressure you, or make you feel that you had to have oral or anal sex? [If Female] During the last 12 months, did another inmate use physical force, pressure you, or make you feel that you had to have oral, vaginal, or anal sex? LCM4. [If Male] How long has it been since another inmate in this facility used physical force, pressured you, or made you feel that you had to have oral or anal sex? [If Female] How long has it been since another inmate in this facility used physical force, pressured you, or made you feel that you had to have oral, vaginal, or anal sex? Follow-up questions for inmates reporting no sexual activity in the screener questions for sexual activity with staff: LCM5. During the last 12 months, have you had any sex or sexual contact with staff in this facility whether you wanted to have it or not? LCM6. How long has it been since you had any sex or sexual contact with staff in this facility whether you wanted to or not? 1. Within the past 7 days 2. More than 7 days ago but within the past 30 days 3. More than 30 days ago but within the past 12 months 4. More than 12 months ago 5. This has not happened to me at this facility LCM7. [If Male]In the last 12 months, did you have oral, vaginal, or anal sex with any staff at this facility whether you wanted to or not? [If Female] In the last 12 months, did you have oral, vaginal, or anal sex with any staff at this facility whether you wanted to or not? LCM8. [If Male] How long has it been since you had oral, vaginal, or anal sex with any staff at this facility whether you wanted to or not? [If Female] How long has it been since you had oral, vaginal, or anal sex with any staff at this facility whether you wanted to or not? Methodology The National Inmate Survey, 2008-09 (NIS-2) was conducted in 167 state and federal prisons between October 13, 2008, and March 11, 2009; 286 jails between January 20, 2009, and August 13, 2009; and 10 special (military, Indian country, and Immigration and Customs Enforcement (ICE)) facilities between May 11, 2009, and December 17, 2009. The data were collected by RTI International under a cooperative agreement with the Bureau of Justice Statistics (BJS). The NIS-2 comprised two questionnaires--a survey of sexual victimization and a survey of past drug and alcohol use and treatment. Inmates were randomly assigned to receive one of the questionnaires so that at the time of the interview the content of the survey remained unknown to facility staff and the interviewers. A total of 81,566 inmates participated in the survey, including 32,029 inmates in state and federal prisons, 48,066 inmates in jails, 399 inmates in military facilities, 115 inmates in Indian country jails, and 957 inmates in facilities operated by ICE. The interviews, which averaged 25 minutes in length, used computer-assisted personal interviewing (CAPI) and audio computer-assisted self-interviewing (ACASI) data collection methods. For approximately the first two minutes, survey interviewers conducted a personal interview using CAPI to obtain background information and date of admission to the facility. For the remainder of the interview, respondents interacted with a computer-administered questionnaire using a touch-screen and synchronized audio instructions delivered via headphones. Respondents completed the ACASI portion of the interview in private, with the interviewer either leaving the room or moving away from the computer. A shorter paper questionnaire was made available for inmates who were unable to come to the private interviewing room or interact with the computer. The paper form was completed by 496 prison inmates (or 1.5% of all prison interviews), 226 jail inmates (0.5%), and 4 military inmates (1%). Most of these inmates were housed in administrative or disciplinary segregation or were considered too violent to be interviewed. No inmates in Indian country or ICE facilities completed the paper questionnaire. Before the interview, inmates were informed verbally and in writing that participation was voluntary and that all information provided would be held in confidence. Interviews were conducted in either English (98% in prisons, 95% in jails, 40% in ICE facilities, and 100% in military and Indian country facilities) or Spanish (2% in prisons, 5% in jails, and 60% in ICE facilities). Selection of state and federal prisons A sample of 171 state and federal prisons was drawn to produce a sample representing approximately 10% of the 1,260 state and 192 federal adult confinement facilities identified in the 2005 Census of State and Federal Adult Correctional Facilities. The 2005 census was a complete enumeration of state prisons, including all publicly operated and privately operated facilities under contract to state correctional authorities. The NIS-2 was restricted to confinement facilities--institutions in which fewer than 50% of the inmates were regularly permitted to leave, unaccompanied by staff, for work, study, or treatment. Such facilities included prisons, penitentiaries, prison hospitals, prison farms, boot camps, and centers for reception, classification, or alcohol and drug treatment. The NIS-2 excluded community-based facilities, such as halfway houses, group homes, and work release centers. Based on estimates from 2008 National Prisoner Statistics, the prisons in the study universe held an estimated 1,267,400 state and 190,300 federal inmates age 18 or older on June 30, 2008. State and federal confinement facilities were sequentially sampled with probabilities of selection proportionate to size (as measured by the number of inmates held in state prisons on December 30, 2005, and in federal prisons on September 28, 2006). Facilities on the sampling frame were stratified by gender of inmates housed. The measures of size for facilities that participated in NIS-1 in 2007 were reduced to lower their probability of selection in NIS-2. (See page 32 for a listing of NIS-1 reports.) Within each stratum, facilities on the sampling frame were first sorted by region, state, and public or private operation. The sample size for facilities housing only female inmates was set to 36 facilities to ensure a sufficient number of women and allow for meaningful analyses of sexual victimization by gender. Facilities were sampled ensuring that at least one facility in every state was selected. Federal facilities were grouped together and treated like a state for sampling purposes. The remaining facilities were selected from each region with probabilities proportionate to size. Of the 171 selected prison facilities, 4 were deemed ineligible and excluded from the survey for the following reasons: * Albion Correctional Facility (NY)--Ongoing litigation. * Robert Scott Correctional Facility (MI)--Ongoing litigation. * Dinwiddie Correctional Unit (VA)--Closed prior to the start of data collection. * Waseca FCI (BOP)-Transitioned from holding males to females during the data collection period (treated as a closed facility). All other selected prison facilities participated fully in the survey. Selection of inmates within prisons The number of inmates sampled in each facility varied based on 6 criteria: * an expected sexual victimization prevalence rate of 4%. * a desired level of precision based on a standard error of 1.75%. * a projected 70% response rate among selected inmates. * a 5% chance among participating inmates of not receiving the sexual victimization questionnaire. * an adjustment factor of 1.75 to account for the complex survey design. * the size of the facility. A roster of inmates was obtained just prior to the start of interviewing at each facility. Inmates under age 18 and inmates expected to be released prior to data collection were deleted from the roster. Each eligible inmate was assigned a random number and sorted in ascending order. Inmates were selected from the list up to the expected number of inmates determined by the sampling criteria. A total of 46,189 prison inmates were selected. After selection, 1,302 ineligible inmates were excluded--936 (2%) were released or transferred to another facility before interviewing began, 246 (0.5%) were mentally or physically unable to be interviewed, 13 (0.02%) were under age 18 or their age could not be obtained during the interview process, 11 (0.02%) were selected in error (i.e., an inmate was incorrectly listed on the facility roster), and 96 (0.2%) were on unsupervised work release or only served time on weekends. Of all selected eligible prison inmates, 23% refused to participate in the survey, 2% were not available to be interviewed (e.g., in court, in medical segregation, determined by the facility to be too violent to be interviewed, or restricted from participation by another legal jurisdiction), and 2% were not interviewed due to survey logistics (e.g., language barriers, releases, or transfers to another facility after interviewing began). Overall, 32,029 prison inmates participated in the survey, yielding a response rate of 71%. Approximately 95% of the participating inmates (29,954) received the sexual assault survey. (See appendix table 1 for the number of participating inmates in each prison facility.) Selection of jail facilities A sample of 320 jails was drawn to represent approximately 10% of the 3,007 jail facilities identified in the Census of Jail Inmates, 2005. The 2005 census was a complete enumeration of all jail jurisdictions, including all publicly operated and privately operated facilities under contract to jail authorities. The NIS-2 was restricted to jails that had five or more inmates on June 30, 2005. Based on estimates from the Annual Survey of Jails, 2008, these jails held an estimated 777,200 inmates age 18 or older on June 30, 2008. Jail facilities were sequentially sampled with probabilities of selection proportionate to size (as measured by the number of inmates held on June 30, 2005). Eight facilities that were unable to participate in NIS-1 were selected with certainty, while the measures of size of facilities that participated in NIS-1 were reduced to give them a lower probability of selection. The remaining facilities were stratified such that facilities in each of the 10 largest jail jurisdictions were placed into strata; all other facilities were placed in a single stratum. Within the large jurisdiction stratum, 3 facilities were selected from the 5 largest jurisdictions with probability proportionate to size, and 2 facilities were selected from the next 5 largest jurisdictions with probability proportion to size. Facilities in the second stratum were first sorted by region, state, and public or private operation. Facilities were sampled to ensure that at least one jail facility in every state was selected. The remaining jail facilities were selected from each region with probabilities proportionate to size. Of the 320 selected jails in NIS-2, 10 facilities refused to participate: * Baldwin Co. Corrections Center (AL) * Marengo Co. Detention Center (AL) * Merced Co. Jail (CA) * Columbia Co. Detention Center (FL) * Pike Co. Law Enforcement Center (IN) * Flathead Co. Detention Center (MT) * Rutherford Co. Jail (NC) * Monmouth Co. Correctional Institution (NJ) * Hildalgo Co. Detention Center (TX) * Kenosha Co. Jail (WI). Nine facilities were unable to participate due to lack of space, staffing shortages, or construction, but expect to be included in NIS-3 (to be conducted in 2011). Fifteen facilities were determined to be ineligible: 7 had closed, 4 were community-based facilities, 2 had fewer than 5 inmates, and 2 were prisons. All other selected jail facilities participated fully in the survey. Selection of inmates within jails The number of inmates sampled in each facility varied based on 6 criteria: * an expected prevalence rate of sexual victimization of 3%. * a desired level of precision based on a standard error of 1.4%. * a projected 65% response rate among selected inmates. * a 5% chance among participating inmates of not receiving the sexual victimization questionnaire. * an adjustment factor of 1.75 to account for the complex survey design. * a pre-arraignment adjustment factor equal to 1 in facilities where the status was known for all inmates and less than 1 in facilities where only the overall proportion of inmates who were pre-arraigned was known. A roster of inmates was obtained just prior to the start of interviewing at each facility. Inmates under age 18 and inmates who had not been arraigned were removed from the roster. Each eligible inmate was assigned a random number and sorted in ascending order. Inmates were selected from the list up to the expected number of inmates determined by the sampling criteria. Due to the dynamic nature of jail populations, a second roster of inmates was obtained on the first day of data collection. Eligible inmates who appeared on the second roster but who had not appeared on the initial roster were identified. These inmates had been arraigned since the initial roster was created or were newly admitted to the facility and arraigned. A random sample of these new inmates was chosen using the same probability of selection used to sample from the first roster. A total of 81,306 jail inmates was selected. After selection, an additional 9,490 ineligible inmates were excluded-7,844 (9.7%) were released or transferred to another facility before interviewing began, 455 (0.6%) were mentally or physically unable to be interviewed, 144 (0.2%) were under age 18 or their age could not be obtained during the interview process, 308 (0.4%) were selected in error (i.e., an inmate was incorrectly listed on the facility roster), and 739 (0.9%) were on unsupervised work release or only served time on weekends. Of all selected inmates, 17% refused to participate in the survey, 4% were not available to be interviewed (e.g., in court, in medical segregation, etermined by the facility to be too violent to be interviewed, or restricted from participation by another legal jurisdiction), and 8% were not interviewed due to survey logistics (e.g., language barriers, releases, and transfers to another facility after interviewing began). Overall, 48,066 jail inmates participated in the survey, yielding a response rate of 68%. Approximately 95% of the participating inmates (45,126) received the sexual assault survey. (See appendix table 5 for the number of participating inmates in each jail facility.) Selection of special confinement facilities A sample of 11 special facilities was drawn to represent the inmate populations in military, Indian country, and ICE facilities. Three military, 3 Indian country, and 5 ICE facilities were included. The selected military facilities were the largest Army, Navy, and Marine facilities, including the U.S. Disciplinary Barracks (Leavenworth, KS), the Naval Consolidated Brig. (Miramar, CA), and the Marine Corps Base Brig. (Camp Pendleton, CA). The selected Indian country facilities were the three largest facilities identified in the 2007 Survey of Jails in Indian Country. One facility, Tohono O'odham Detention Center (AZ) was undergoing major renovations to accommodate increased capacity and could not participate in the survey during the data collection period. This facility will be included in the NIS-3 data collection. The Navajo Department of Corrections Window Rock (AZ), participated in the survey but held fewer inmates in 2009 (14) than reported in the 2007 survey (99). The 5 ICE facilities were sequentially sampled from the 22 facilities run by ICE with probabilities of selection proportionate to size (as measured by the number of persons held at yearend 2008). Facilities were sorted by region and state. Selection of inmates in special confinement facilities For purposes of inmate selection, military facilities were treated as prisons, and Indian country and ICE facilities were treated like jails. The assumptions used to determine the sample size within a prison or jail and the corresponding selection procedures were used. However, in ICE facilities, a second sample of newly admitted inmates was not drawn due to an inability to identify new inmates on the ICE rosters. In addition, inmates in ICE facilities who did not speak English or Spanish were defined as ineligible for the study. Overall, 2,494 inmates were selected, including 546 in military facilities, 161 in Indian country facilities, and 1,787 in ICE facilities. After selection, 409 ineligible inmates were excluded 225 (9%) were released or transferred to another facility before interviewing began, 10 (0.4%) were mentally or sysically unable to be interviewed, 4 (0.2%) were on unsupervised work release, and 170 (7%) in ICE facilities did not speak English or Spanish. A total of 16 inmates were excluded in military facilities, 15 in Indian country facilities, and 378 in ICE facilities. Of all selected inmates in special facilities, 16% refused to participate in the survey, 0.2% were not available to be interviewed (e.g., in court, in medical segregation, determined by the facility to be too violent to be interviewed, or restricted from participation by another legal jurisdiction), and 7% were not interviewed due to survey logistics (e.g., language barriers, releases, and transfers to another facility after interviewing began). Overall, 1,471 inmates participated in the survey (399 in military, 115 in Indian country, and 957 in ICE facilities), yielding a response rate of 71% (75% in military, 79% in Indian country, and 68% in ICE facilities). Approximately 95% of the participating inmates (1,379) received the sexual assault survey (379 in military, 107 in Indian country, and 893 in ICE facilities). (See appendix table 9 for the number of participating inmates in each special confinement facility.) Weighting and non-response adjustments Responses from interviewed inmates were weighted to provide national-level and facility-level estimates. Each interviewed inmate was assigned an initial weight corresponding to the inverse of the probability of selection within each sampled facility. A series of adjustment factors was applied to the initial weight to minimize potential bias due to non-response and to provide national estimates. Bias occurs when the estimated prevalence is different from the actual prevalence for a given facility. In each facility, bias could result if the random sample of inmates did not accurately represent the facility population. Bias could also result if the non-respondents were different from the respondents. Post-stratification and non-response adjustments were made to the data to compensate for these two possibilities. These adjustments included-- 1. calibration of the weights of the responding inmates within each facility so that the estimates accurately reflected the facility's entire population in terms of known demographic characteristics. These characteristics included distributions by inmate age, sex, race, time since admission, and sentence length. This adjustment ensures that the estimates better reflect the entire population of the facility and not just the inmates who were randomly sampled. 2. calibration of the weights so that the weight from a non-responding inmate is assigned to a responding inmate with similar demographic characteristics. This adjustment ensures that the estimates accurately reflect the full sample, rather than only the inmates who responded. For each inmate, these adjustments were based on a generalized exponential model, developed by Folsom and Singh, and applied to the sexual assault survey respondents.*** Footnote 3 3Folsom, Jr., R.E., and A.C. Singh, (2002). "The Generalized Exponential Model for Sampling Weight Calibration for Extreme Values, Nonresponse, and Poststratification," Proceedings of the American Statistical Association, Section on Survey Research Methods, pp. 598-603.*** A final ratio adjustment to each inmate weight was made to provide national-level estimates for the total number of inmates ages 18 or older who were held at midyear 2008. These ratios represented the estimated number of inmates by sex (from BJS's 2008 Annual Survey of Jails and 2008 National Prisoner Statistics, Midyear) divided by the 3Folsom, Jr., R.E., and A.C. Singh, (2002). "The Generalized Exponential Model for Sampling Weight Calibration for Extreme Values, Nonresponse, and Poststratification," Proceedings of the American Statistical Association, Section on Survey Research Methods, pp. 598-603. number of inmates by sex in the NIS-2 after calibration for sampling and non-response. The national estimates for state prisons were 1,178,916 males and 88,518 females; for federal prisons, 178,153 males and 12,120 females; and for jails (with an average daily population of 6 or more inmates), 678,136 males and 99,096 females. Final ratio adjustments were not applied to inmate weights in military, Indian country, and ICE facilities. Estimates for special confinement facilities were made at the facility level only. Standard errors and tests of significance The NIS-2 is statistically unable to provide an exact ranking for all facilities as required under PREA. As with any survey, the NIS estimates are subject to error arising from the fact that they are based on a sample rather than a complete enumeration. Within each facility, the estimated sampling error varies by the size of the estimate, the number of completed interviews, and the size of the facility. A common way to express this sampling variability is to construct a 95%-confidence interval around each survey estimate. Typically, multiplying the standard error by 1.96 and then adding or subtracting the result from the estimate produces the confidence interval. This interval expresses the range of values that could result among 95% of the different samples that could be drawn. For small samples and estimates close to 0%, as is the case with sexual victimization in most prisons and jails, the use of the standard error to construct the 95%-confidence interval may not be reliable. An alternative developed by Wilson has been shown to perform better than the traditional method when constructing a confidence interval. (See note 1 on page 7.) This method produces an asymmetrical confidence interval around the facility estimates in which the lower bound is constrained to be greater than or equal to 0%. It also provides confidence intervals for facilities in which the survey estimates are zero (but other similarly conducted surveys could yield non-zero estimates). (See tables 2, 4, and 5 and appendix tables 5, 6, 8, and 9.) When applied to large samples, the traditional and the Wilson confidence intervals are nearly identical. As a result, the tables that show national estimates display traditional standard errors. (See table 1.) The traditional standard errors have also been used to compare estimates of sexual victimization among selected groups of inmates that have been defined by type of incident, demographic subgroup, sexual history, and criminal justice status. (See tables 6 through 12 and tables 15 through 19.) To facilitate the analysis, rather than provide the detailed estimates for every standard error, differences in the estimates of sexual victimization for subgroups in these tables have been tested and notated for significance at the 95%-level confidence. For example, the difference in the rate of inmate-on-inmate sexual victimization among female prisoners (4.7%), compared to male prisoners (1.9%), is statistically signifi-cant at the 95% level of confidence (table 6). In all tables providing detailed comparisons, statistically significant dif-ferences at the 95% level of confidence or greater have been designated with two asterisks (**). Exposure period To calculate comparative rates of sexual victimization, respondents were asked to provide the most recent date of admission to the current facility. If the date of admission was at least 12 months prior to the date of the interview, inmates were asked questions related to their experiences during the past 12 months. If the admission date was less than 12 months prior to the interview, inmates were asked about their experiences since they had arrived at the facility. The average exposure period of inmates participating in the sexual victimization survey was-- * 9.0 months for federal prisoners * 7.9 months for state prisoners * 3.4 months for jail inmates * 8.4 months for inmates in military facilities * 3.0 months for inmates in ICE facilities * 4.4 months for inmates in Indian country facilities. Measurement of sexual victimization The survey of sexual victimization relied on inmates reporting their direct experience, rather than inmates reporting on the experience of other inmates. Questions related to inmate-on-inmate sexual activity were asked separately from questions related to staff sexual misconduct. (For specific survey questions see appendices 1 and 2.) The ACASI survey began with a series of questions that screened for specific sexual activities without restriction, including both wanted and unwanted sex and sexual contacts with other inmates. To fully measure all sexual activities, questions related to the touching of body parts in a sexual way were followed by questions related to manual stimulation and questions related to acts involving oral, anal, and vaginal sex. The nature of coercion (including use of physical force, pressure, and other forms of coercion) was measured for each type of reported sexual activity. ACASI survey items related to staff sexual misconduct were asked in a different order. Inmates were first asked about being pressured or being made to feel they had to have sex or sexual contact with the staff and then asked about being physically forced. In addition, inmates were asked if any facility staff had offered favors or special privileges in exchange for sex. Finally, inmates were asked if they willingly had sex or sexual contact with staff. All reports of sex or sexual contact between an inmate and facility staff, regardless of the level of coercion, were classified as staff sexual misconduct. The ACASI survey included additional questions related to both inmate-on-inmate sexual victimization and staff sexual misconduct. These questions, known as latent class measures, were included to assess the reliability of the survey questionnaire. After being asked detailed questions, all inmates were asked a series of general questions to determine if they had experienced any type of unwanted sex or sexual contact with another inmate or had any sex or sexual contact with staff. (See appendix 3.) The entire ACASI questionnaire (listed as the National In-mate Survey-2) and the shorter paper and pencil survey form (PAPI) are available on the BJS web site at . Interviews checked for inconsistent response patterns Once data collection was completed, individual response patterns were assessed to identify interviewer error, inter-views that had been completed in too short of time, and incomplete interviews. In 133 interviews, the interviewers administered sex-specific survey items inconsistent with the sex of the inmate. In 208 interviews, the inmate either failed to complete enough questions to be considered a completed interview or completed the survey in too short of time. These interviews were excluded from the calculations of sexual victimization. Interviews were also examined for inconsistent response patterns. A list of 18 indicators were developed based on in-mate characteristics (e.g., education, age, marital status, and time since admission) and items related to victimization (e.g., number of times, injuries, willing contact with staff, sex of staff perpetrator, and reporting of victimization). Indicators compared responses to initial questions with responses to detailed follow up questions. The results were combined into a count of the total number of inconsistent responses for each inmate. Overall, the results revealed very high levels of consis-tency in survey responses. Of 45,126 completed interviews of jail inmates, 94.4% had no inconsistent responses, 5.2% had 1, and 0.4% had 2 or more. Of 29,954 completed interviews of prison inmates, 93.6% had no inconsistent responses, 5.9% had one, and 0.5% had two or more. Definition of terms Sexual victimization--all types of sexual activity, e.g., oral, anal, or vaginal penetration; hand jobs; touching of the in-mate's buttocks, thighs, penis, breasts, or vagina in a sexual way; abusive sexual contacts; and both willing and unwill-ing sexual activity with staff. Nonconsensual sexual acts--unwanted contacts with another inmate or any contacts with staff that involved oral, anal, vaginal penetration, hand jobs, and other sexual acts. Abusive sexual contacts only--unwanted contacts with another inmate or any contacts with staff that involved touching of the inmate's buttocks, thigh, penis, breasts, or vagina in a sexual way. Unwilling activity--incidents of unwanted sexual contacts with another inmate or staff. Willing activity--incidents of willing sexual contacts with staff. These contacts are characterized by the reporting inmates as willing; however, all sexual contacts between inmates and staff are legally nonconsensual. Staff sexual misconduct--includes all incidents of will-ing and unwilling sexual contact with facility staff and all incidents of sexual activity that involved oral, anal, vaginal penetration, hand jobs, blow jobs, and other sexual acts with facility staff. Logistic regression models Multivariate logistic regression estimation is a modeling technique used to determine what characteristics are statistically significant for predicting a dichotomous outcome (e.g., victimized or not victimized) while controlling for all the other characteristics in the model. NIS-2 used this technique to determine what inmate-level characteristics were significant predictors of inmate-on-inmate sexual victimization and staff sexual misconduct. For each outcome, inmate-level characteristics were divided into 3 categories: demographic characteristics, sexual orientation and history, and criminal justice status and history. For each category a logistic model was iteratively run under a backwards selection technique until only predictors that were significant at the 95% level of confidence remained (tables 7, 9, 11). Each reduced model was then combined to provide 4 models (for each type of sexual victimization and facility type). Backwards selection was conducted on each of the combined models until only predictors significant at the 95% level of confidence remained. Results for each model are displayed in terms of their conditional predicted probability (table 12). In each of the logistic regression models, the condi-tional predicted probability represents the probability that an inmate with a particular characteristic has experienced sexual victimization (by type) conditional on the inmate having the mean value for all other predictors in the model. For example, based on demographic characteristics only, a female prison inmate has a 3.8% chance of being victimized by another inmate given that she was at the mean of the joint distribution of race or Hispanic origin, education level, and marital status. (See table 7 and note 2 on page 13.) Predicting facility-level rates of sexual victimization Estimates of the expected rate of inmate-on-inmate and staff sexual misconduct in each high rate prison and jail were calculated based on the characteristics of the inmates housed in the facility and the estimated rates of victimization associated with each characteristic. For each level of a characteristic, the logistic models provide an estimate of the odds that an inmate was sexually victimized given that the inmate had that characteristic. (For presentation purposes, these odds have been converted into conditional probabilities.) Overall, the predicted odds that a particular inmate has been victimized is the sum of the odds that cor-respond to that inmate's set of characteristics. For example, the inmate-on-inmate sexual victimization model for prison inmates consists of 7 inmate characteristics: sex, race or Hispanic origin, marital status, sexual orientation, prior sexual assault status, most serious offense, and sentence length. The model provides estimates of the odds for each level of each of these characteristics. The odds of victimization for an inmate who is male, white, never married, heterosexual, with no prior sexual assault, held for a property offense, and sentenced to 1 to 5 years in prison is the sum of the individual odds for each of those characteristics. Once the overall odds for an inmate has been calculated, it can be converted to a probability. A predicted facility-level victimization rate is the weighted average of probabilities for all inmates in the facility. This weighted average in each high rate facility was calculated by summing across all inmates the product of the inmate's probability of victimization and the adjusted inmate sampling weights and dividing it by the eligible inmate population in the survey. Five BJS reports on sexual victimization in prisons and jails: Sexual Violence Reported by Correctional Authorities, 2004 (NCJ 210333) Sexual Violence Reported by Correctional Authorities, 2005 (NCJ 214646) Sexual Violence Reported by Correctional Authorities, 2006 (NCJ 218914) Sexual Victimization in State and Federal Prisons Reported by Inmates, 2007 (NCJ 219414) Sexual Victimization in Local Jails Reported by Inmates, 2007 (NCJ 221946) An overview of all of the BJS prison rape collections: PREA Data Collection Activities, 2010 (NCJ 230448) These reports are available online at . BJS Office of Justice Programs Innovation  Partnerships  Safer Neighborhoods http://www.ojp.usdog.gov