obÌåÓý

Overview

The NCVS, NISVS, YRBSS, and UCR each have strengths and limitations in the information on sexual victimization they can provide. While these collections together contribute to a more comprehensive view of sexual victimization in our nation, measurement differences across the collections lead to differences in the stories they tell about the level of victimization experienced by U.S. residents.

Scope and context of data collection

The NCVS and the UCR are crime-based collections, whereas information from the YRBSS and NISVS are collected in a public health context. From a public health perspective, certain types of sexual victimization, such as the non-contact unwanted sexual experiences measured by NISVS, may be harmful and important to measure, although they do not necessarily rise to the level of criminal behavior. Using a broad range of behaviors that constitute sexual victimization in a public health data collection compared to a criminal justice collection could result in higher estimates of the prevalence of sexual victimization. Context (e.g., crime versus public health) can additionally be important because if the respondent does not consider what happened to be a crime, he or she may be less likely to report that incident in an interview presented in the context of crime. This may be particularly true for sexual victimization given most perpetrators are known to the victim. Survey context differences could partially explain the lower rates of sexual victimization in crime surveys compared to public health surveys.

Question wording and design

The language of questions in a survey may affect whether a respondent indicates that an incident occurred. The three surveys use different approaches to asking about experiences with rape and other sexual victimization. The NCVS and YRBSS ask questions about sexual victimization using more traditional criminal justice terminology. The NCVS questions include the terms ‘rape,’ ‘attempted rape,’ and ‘forced or unwanted sexual acts,’ and the YRBSS asks about ‘forced sexual intercourse.’ These terms are not explicitly defined for the respondent and if the victim does not feel that his or her experience can be described by those terms, the question may not be answered affirmatively. On the other hand, the NISVS uses behaviorally specific questions to ask whether these types of experiences occurred. For instance, rather than asking women if they have been raped, NISVS asks whether they experienced physically forced “vaginal, anal, or oral penetration by a male using his penis or a male or female using their fingers, or an object.” With this approach the exact acts that are within scope are described and defined for the respondent. One downside is that more questions are needed to fully capture the behaviors measured which can create additional respondent burden. However, additional questions may also facilitate rapport and increase opportunities for disclosure of victimization.

The three surveys also differ in whether they use a one- or two-stage approach to measuring sexual victimization experiences. The NISVS and YRBSS use a one-stage approach meaning that an affirmative response to a question like the example provided above results in that respondent being counted as a victim of rape. The NCVS on the other hand, uses a two-stage approach to identifying incidents of rape and sexual assault. Initially, a screener is administered, with cues designed to help the respondent’s recollection of events and ascertain whether the respondent experienced any criminal victimization during the reference period. If the respondent answers a screening question affirmatively, he or she then completes an incident form about each incident. The incident form captures detailed information about the incident and is used for classifying the type of crime that occurred. Even if the respondent does not respond affirmatively to the specific screeners on rape and unwanted sexual contact, the respondent could still be classified as a rape or sexual assault victim if a rape or unwanted sexual contact is reported during the stage-two incident report for another type of crime. The positives and negatives of the one and two stage approaches have been well documented in the literature, but it is unclear the extent to which these approaches alone impact estimates of sexual victimization or how they impact the estimates.

Study population

Another important consideration in national sexual violence estimates is who is included in the study population. Although all the systems are intended to be representative of the populations they are designed to cover, each system necessarily excludes certain categories of victims. For instance, the four collections vary in the ages for eligibility. NISVS includes adults ages 18 and older; NCVS includes anyone who is 12 years or older; YRBSS includes students in grades 9-12, and UCR covers any reported crime regardless of the victim’s age. Aside from age, it is important to consider who else is not represented in the study population. For example, the NCVS sample is address- and household-based and does not cover those who are homeless, living in military barracks, or institutionalized; YRBSS is school-based, so excludes youth not in school; the UCR is based on police reports, so excludes victims who do not report to the police; and NISVS is a landline/cell phone based survey, thus, does not capture those who live in institutions or do not have telephones.

Most of data collection and response rates

For resident surveys like the NCVS, NISVS, and YRBSS, the mode of data collection can impact both respondent participation in the survey and victim disclosure of sexual victimization. Self-administered surveys, such as the YRBSS, are not subject to interviewer effects and afford respondents the greatest privacy for reporting on sensitive topics. However, some victims may appreciate the opportunity to talk about what happened to them and may be more inclined to provide information about an incident to an interviewer than on a survey form. Interviewer-administered surveys like the NCVS and NISVS, can be subject to interviewer effects if respondents are more or less likely to talk about their experiences depending on characteristics of the interviewer. For instance, female respondents may be more comfortable talking about sexual victimization with female interviewers than male interviewers. For this reason, NISVS only uses female interviewers given previous research suggesting that female interviewers are more likely to create conditions that are conducive to disclosure of both male and female respondents. Additionally, in-person versus telephone interviews may result in differences in disclosure. For example, a respondent may not be comfortable disclosing sensitive information in-person to an interviewer in their home where other family members may be present. While both in-person and telephone interviewers are trained to instruct respondents to answer survey questions in private, there is always a possibility that respondents may be deterred from providing accurate survey responses if they think that someone nearby could overhear.

Data collection mode can also impact response rates and influence nonresponse bias. Surveys with low response rates may result in estimates that are subject to nonresponse bias. Nonresponse bias means that those who participated in the survey may differ in important ways from those who did not participate, which could in turn impact the survey findings. In the NCVS, first interviews are conducted in person to develop rapport with respondents and NCVS response rates have historically been above 80%. The NISVS faces challenges with reaching respondents by phone. The 2015 NISVS response rate was 26.4%. However, when contact was made with an eligible telephone respondent, the rate of cooperation was 89.6%. In 2015, the self-administered, national YRBSS had a school response rate of 69%, a student response rate of 86%, and an overall response rate (school response rate x student response rate) of 60%.

While the UCR is an administrative records collection, the data are also impacted by response rates in terms of whether law enforcement jurisdictions provide the summary statistics and how complete their reporting is. In 2015, law enforcement agencies active in the UCR Program represented 98% of the total population. The coverage amounted to 99% of the population in Metropolitan Statistical Areas, 92% of the population in cities outside metropolitan areas, and 93% of the population in non-metropolitan counties.

Data on the characteristics of the violence

Another difference between these data systems is the additional data that are collected on the characteristics of sexual violence. The NCVS and NISVS collect rich data about the sexual violence experience, such as the victim’s relationship with the perpetrator, when and where the violence occurred, the frequency of victimization, and associated impacts. For instance, NCVS includes, weapon use, injury, socio-economic consequences of the crime, and victim help-seeking behaviors. NISVS includes detailed information about the health impacts associated with victimization including acquiring an STD, getting pregnant, injury, need for services, and various chronic health conditions. NISVS also captures the age at first victimization of rape and being made to penetrate. The YRBSS captures whether the events occurred, but no further details are collected about the violence because the primary focus of the survey is general health versus sexual violence specifically. However, the YRBSS includes many health risk behaviors that can be examined in relationship to the violence. The UCR SRS data does not include any contextual information.