National Crime Victimization Survey
The National Crime Victimization Survey (NCVS) is a national household survey, conducted on an ongoing basis by the Census Bureau for the Bureau of Justice Statistics (BJS). It is the major tool used by BJS to provide statistics on criminal victimizations, covering rape, sexual assault, robbery, assault, theft, household burglary, and motor vehicle theft. It is the source of data for an annual BJS publication, Criminal Victimization, and numerous other regular and special reports.
Over its history, the NCVS (and its predecessor the National Crime Survey) has been a uniquely valuable source of information on the “dark figure of crime”—those crimes not reported to police (Baumer and Lauritsen, 2010; Biderman and Reiss, 1967; Lauritsen and Heimer, 2008; National Research Council, 2008). The NCVS is complementary to, and frequently compared to, data from the Uniform Crime Reports (UCR) discussed in Chapter 3. The NCVS enables BJS to provide statistics on victimization for the U.S. population (12 years of age and older) as a whole and also for important subpopulations, such as women, the elderly, members of various racial groups, city dwellers, and other groups definable from survey data. It also focuses attention on the characteristics of the victims of crime, including the nature of their victimizations.
This chapter first briefly describes the history and development of the NCVS, including a major redesign in 1991. It then provides some descriptive details about the survey, including the target population and sample design, the data collection process, the survey instruments, the definitions used for rape and sexual assault, the estimation process, and the products. This chapter is descriptive in nature: it does not offer any assessment of
the quality of the survey. (The panel’s assessment of various features of the survey is the subject of Chapters 7, 8, and 9.) In reading this chapter, it is important to keep in mind that the NCVS is an omnibus survey, covering many types of criminal victimizations, not just rape and sexual assault. However, to the extent possible and in keeping with the panel’s charge, this chapter and the entire report are focused on the issues relevant to those two victimizations.
Origins and Early Development: 1967-1991
The NCVS has its roots in a report from President Johnson’s Commission on Law Enforcement and Administration of Justice in 1967. The commission reported that the UCR was useful in many ways, but it noted several critical concerns with having just one source of crime statistics. One concern was that the UCR was a summary of only those crimes reported to (and recorded by) police and not all crimes (see Chapter 3). Secondly, the commission found that the UCR administrative statistics were open to possible manipulation and misrepresentation. Finally, the commission said, the UCR lacked information about the victims, the victimization incidents, and the offenders that is needed to develop effective policy choices. The commission recommended the development of a national crime survey (President’s Commission on Law Enforcement and the Administration of Justice, 1967; Rennison and Rand, 2007).
Work on the National Crime Survey (NCS), predecessor of the NCVS, began soon afterwards with small-scale field tests of questions asked of victims of crimes that had been reported to police. A questionnaire was developed from these tests, which was initially implemented as a supplement to the Census Bureau’s Quarterly Household Survey in 1971. These supplements served as a way to further develop concepts and questions; they were not used to produce BJS published reports.
In July 1972 the new NCS was fielded. The core had a sample of 72,000 households and noninstitutionalized group quarters.1 In addition, the suite of surveys included a national sample of 15,000 businesses, as well as city-based samples in each of 26 major cities (12,000 households and 2,000 businesses) to support local-area estimates of victimization. These
1Group quarters included group living arrangements, generally of individuals who are not related to each other, such as college residence halls, residential treatment centers, etc.
“cities surveys” began in eight “impact” cities and were expanded to other cities over the next 3 years.2
A change in design was recommended a few years later in Surveying Crime (National Research Council, 1976). One of the recommended changes was to abandon the “cities surveys” to provide resources to continue to support the core national household sample. BJS established a redesign consortium, and small-scale changes to the survey and its instruments were implemented in 1986.
Considerable cognitive work was undertaken as part of this redesign, much of which focused on understanding and testing screener strategies to increase reported incidents. In the NCS the screener directly asked a respondent about several (but not all) types of possible victimizations. According to the Bureau of Justice Statistics (1989, pp. 11-12):
The major proposed changes that were ultimately tested [for this omnibus survey] included asking respondents about victimizations that occurred in various life “domains” such as work and leisure, providing many short cues to help trigger memory of incidents in these life contexts, and attempting to evoke the sort of emotional states that might result from a crime incident (such as anger or fear) before administering the screen questions. The aim of these innovations was both to elicit increased reporting of crime incidents and to structure the recall task to a greater degree, so that cognitive and subcultural differences among respondents would have a smaller impact on the reporting of crime incidents.
Other topics, such as bounding, reference period, interview-to-interview recounting, ways to enhance the reliability of dating incidents, and series crimes were analyzed during this redesign process. More details are available from Bureau of Justice Statistics (1989). Following this development, a three-wave national pretest of the redesigned victimization survey was fielded in 1989.
Development Since the 1991 Redesign
In July 1991, the name of the redesigned survey was changed to the National Crime Victimization Survey. The new survey questionnaire included revisions that were viewed as a major improvement over its predecessor (Bachman and Taylor, 1994). Specifically, the redesigned questionnaire asked respondents about experiencing forced or coerced sexual behavior in addition to having been “attacked” or “threatened.” The redesign process
2The impact cities were Atlanta, Baltimore, Cleveland, Dallas, Denver, Newark, Portland (Oregon), and St. Louis. More information about this early version of the survey can be found in Lynch and Addington (2007).
also examined the appropriate reference period to use. Based on cognitive research demonstrating that long recall periods yielded poorer reports and that a reference period of 12 months generally reduced the reporting of all criminal incidents by approximately 30 percent (Cantor and Lynch, 2000), a 6-month reference period was retained.
The NCVS was fully implemented in 1993 with 70,707 households and 136,747 individuals interviewed. Because the cost of conducting the NCVS grew while congressional funding remained essentially flat for BJS, sample size cuts were implemented to the NCVS over the subsequent 15 years: 12 percent in 1996, 4 percent in 2002, 16 percent in 2006, and 14 percent in 2007 (National Research Council, 2008).
Surveying Victims: Options for Conducting the National Crime Victimization Survey (National Research Council, 2008, p. 78) raised questions about the funding levels of the NCVS: “As currently configured and funded, the NCVS is not achieving and cannot achieve BJS’s legislatively mandated goal.” The report then made two strong recommendations (National Research Council, 2008, p. 79):
1. BJS must ensure that the nation has quality annual estimates of levels and changes in criminal victimization.
2. Congress and the administration should ensure that BJS has a budget that is adequate to field a survey that satisfies that goal.
BJS received increased funding for the NCVS, and sample sizes were raised. In 2011, 79,802 households and 143,122 individuals were interviewed.
TARGET POPULATION AND SAMPLE DESIGN
The target population of the NCVS is the noninstitutionalized population of the United States, 12 years of age or older. It includes (Bureau of Justice Statistics, 2008b, p. 1)
residents living throughout the United States, including persons living in group quarters, such as dormitories, rooming houses, and religious group dwellings. Crew members of merchant vessels, Armed Forces personnel living in military barracks, and institutionalized persons, such as correctional facility inmates, were not included in the scope of this survey. Similarly U.S. citizens residing abroad and foreign visitors to this country were excluded.
The design challenge for the NCVS is to estimate the number of victimizations (for the target population) that have occurred within the past year and the change from the previous year. The victimizations include
events that range from statistically common, such as theft (104.2 per 1,000 households in 2011) to events that are uncommon, such as total violent crime (22.5 per 1,000 people [12+ years] in 2011) to events that are statistically very rare, such as rape and sexual assault (0.9 per 1,000 people [12+ years] in 2011). This type of situation is discussed in a classic textbook on sampling (Hansen, Hurwitz, and Madow, 1953, p. 107): “the particular population under consideration is buried in a very much larger population, and at best it will be an expensive proposition to obtain a sample that will give estimates … that will be of a high relative precision.”
To produce estimates for a wide range of victimizations, the NCVS uses a classical area sampling design, with selection using multiple stages with stratification and clustering (see detailed description below). The result is a nationally representative sample of individuals identified for data collection.
Table 4-1 provides a summary of the two main stages of this multistage design. The primary sampling units (PSUs) in the first selection stage are small groups of neighboring counties, or large individual counties or metropolitan areas. The largest PSUs (in terms of population size) are included in the NCVS sample with certainty. The remaining PSUs are grouped into strata with similar geographic and demographic characteristics3 as determined by the most recent census, and then sampled with probability proportional to their population size.
Secondary sampling units (SSUs) are individual residential units, which are sampled within each selected PSU separately from each of four nonoverlapping list frames of eligible units. The four frames (called unit, permit, area, and noninstitutionalized group quarters) are used to maximize the population coverage of the NCVS sample. “Unit” listings are those residential units included in the Master Address File developed for the 2000 census.4 “Permit” listings represent new construction and are taken from building permits for new residential units since the 2000 census. “Area” listings are generated from periodic canvassing of selected census blocks within the PSUs and consist of identified residential units that are not on the other three frames. Noninstitutionalized “group quarters” are those living quarters identified for the 2000 census “where people live or stay, in a group living arrangement, that is owned or managed by an entity or organization providing housing and/or services for the resident…. People
3These stratification characteristics include geographic region, population density, rate of growth, population, principal industry, and type of agriculture.
4The Master Address File developed for and used in the decennial census is essentially the listing of all known living quarters in the United States, but excluding group quarters. Address listings from the 2000 census, rather than the 1990 census, began to be used in the NCVS sample in 2005.
TABLE 4-1 NCVS Sample Design Summary
|Stage||Sampling Unit and Frame Source||Stratification|
|1||Primary Sampling Unit (PSU): Counties, groups of counties, or large metropolitan areas. The largest PSUs are self-representing (SR) PSUs, chosen with certainty into the PSU sample; all other PSUs are non-SR (NSR) and thus sampled with probabilities less than 1.||Each SR PSU is considered to be a unique stratum.|
|NSR PSUs are stratified by similar geographic and demographic characteristics.|
|Frame: Listing of all counties, groups of counties, or large metropolitan areas in the entire United States. (Military bases and external territories are not included.)|
|2||Secondary Sampling Unit (SSU): “Measures” defined as clusters of four neighboring housing units from the unit listing in a specific frame. When one “measure” is selected systematically, the next six “measures” in the listing are also selected.||Explicit Stratification: By type of frame.|
|Frame: Four nonoverlapping lists of the following types were created in each sample PSU: unit, area, permit, and noninstitutional group quarters. “Unit” listings are those residential units included in the Master Address File developed for the 2000 census. “Permit” listings represent new construction and are taken from building permits for new residential construction since the 2000 census. “Area” listings are generated from periodic canvassing of selected census blocks within the PSUs and consist of identified residential units that are not on the other three frames. Noninstitutional “group quarters” are those living quarters identified for the 2000 census “where people live or stay, in a group living arrangement, that is owned or managed by an entity or organization providing housing and/or services for the resident.”|
NOTES: Target population is the noninstitutionalized U.S. residents aged 12+ years in all 50 states. The noninstitutionalized population includes people living throughout the United States, including persons living in group quarters, such as dormitories, rooming houses, and religious group dwellings. Not included are crew members of merchant vessels; Armed Forces personnel living in military barracks; institutionalized persons, such as correctional facility inmates; U.S. citizens residing abroad; and foreign visitors to the United States.
SOURCE: Information from Bureau of Justice Statistics (2008b).
|Sample Selection||Sample Size|
|SR PSUs selected with probability equal to 1.||
• Originally, the total PSU sample size was 245.
o 93 SR PSUs were identified.
o 152 NSR PSUs were randomly selected.
|In each NSR stratum:|
• 1 PSU is randomly chosen from each stratum with probability proportional to size (PPS).
• In October 1996, the PSU sample size was reduced to 203.
o The same 93 SR PSUs were retained.
• The measure of size for PPS selection is the PSU’s population size.
o aThe number of NSR PSUs was reduced to 110.
• Within a selected PSU, each of the four frames is sampled separately. The housing units in a given frame are grouped together into clusters of four units called “measures.” The second stage sample is drawn with a systematic selection of “measures” from each frame. When one “measure” is selected, the sample includes that “measure” and the next six “measures” in the listing.
• Number of participating households varies from year to year but is currently about 80,000 households.
• The number of survey respondents varies per year, ranging from 134,000 to 181,000. Currently it is approximately 145,000.
• Annual overall response rates have varied between 86%-91%; currently it is approximately 87%.
• The household sample is divided into six “rotation groups,” and each rotation group is in turn randomly divided into six “panels.”
• The different panel of households (from each rotation group) is interviewed each month during a 6-month data collection period.
• All residents 12+ years at each address for each type of listing are selected for interview.
living in group quarters are usually not related to each other” (U.S. Census Bureau, 2010).5
Within a selected PSU, each of the four frames discussed above is sampled separately.6 The housing units in a given frame are grouped together into clusters of four units. The second stage sample is drawn with a systematic selection of these clusters from each frame.7 When one cluster is selected, the sample includes that cluster and the next six clusters in the listing. This sampling process is similar to that used for the Current Population Survey.8 The actual number of households and persons interviewed in the NCVS sample varies slightly from year to year (see Table 4-2).
For field interviewing, the overall sample is divided into six rotation groups, and each rotation group is further divided into six equal collection panels. A single panel from each rotation group is contacted each month, allowing the entire rotation group to be contacted within 6 months with an equal distribution across months. Table 4-3 displays the rotation pattern for a single rotation group, showing the interview month (I) and the months (x) that are included in the reference period for that interview. Because the survey is continuous, newly constructed housing units are selected as described and assigned to rotation groups and panels for subsequent incorporation into the sample. A new rotation group enters the sample every 6 months, replacing a group phased out. With this overall design, each household in a rotation group is interviewed once every 6 months for a total of seven interviews.9 The initial interview is used to bound information reported in subsequent interviews.10 Prior to 2007 data from the initial interview were
5Noninstitutionalized group quarters include such living quarters as college residence halls, group homes, emergency and transitional shelters, religious group homes, and workers’ dormitories (U.S. Census Bureau, 2010). The NCVS sample excludes other specified institutionalized group quarters, such as military barracks, merchant vessels, correctional facilities, halfway houses, and skilled nursing facilities (Bureau of Justice Statistics, 2008b).
6It was unclear from the documentation whether these frames were sampled proportionately, or disproportionately based on some criteria. As the report discusses later, a disproportionate sample weighted toward certain group quarters or PSUs in more at risk areas may be appropriate for measuring rape and sexual assault.
7Before sampling, the group quarters frame “is converted to housing unit equivalents because Census addresses of individual group quarters or people within a group quarter are not used in the sampling. The number of housing unit equivalents is computed by dividing the Census 2000 group quarters population by the average number of people per household” (U.S. Census Bureau, 2006, pp. 3-8).
8More documentation of the process can be found in U.S. Census Bureau (2006).
9Respondents are contacted every 6 months, and because the data collection is retrospective, they are asked for a total of 3.5 years of data over the 3 years of contact.
10Bounding recall procedures are used to minimize forward telescoping. Forward telescoping describes a pattern of reporting events as having occurred more recently that they actually did. Bounded recall requires a panel survey; in every interview after the first, the interviewer checks whether any incidents the respondent reports were also reported in the previous interview in order to eliminate double reporting.
TABLE 4-2 Sample Size and Response Rates for Households and Persons in the NCVS: 1993-2011
|Year||Eligible Householdsa||Household Interviewed||Household Response Rate (%)||Eligible Personsb||Persons Interviewedc||Response Rate for Persons (%)d|
aThe number of eligible households in which there was a household response at least once during that year.
bThe number of persons in responding households.
cThe number of eligible household members who responded at least once during the year.
dThe rate for person response in a responding household.
SOURCE: Data provided to panel by Bureau of Justice Statistics (personal communication).
TABLE 4-3 Panel Design for the NCVS
|Interview Month (I) and Reference Months for Reporting (x) by Data Collection Panels (P1-P6) Within a Rotation Group|
|Months in Year 1||Months in Year 2|
|Months in Year 3||Months in Year 4|
SOURCE: Data from Bureau of Justice Statistics (2008b).
not used in estimation, but data from wave 1 have been used since that time with an adjustment to account for potential telescoping (see below, under Estimation).
DATA COLLECTION AND SURVEY MODE
As just noted, each address selected for the NCVS remains in the sample for 3 years, with seven interviews of household members taking place at 6-month intervals. The sample is address based, so that if a household or household member moves during the years of the panel, the interviews continue with the current household members who reside at the sampled address.
The initial contact and interview (wave 1) on the NCVS are conducted in person by a Census Bureau field representative. For subsequent interviews, the field representative may, and most often does, conduct the interview by telephone; personal interviews are conducted subsequently on the basis of a respondent’s request and as needed to maintain continued response. Both in-person and telephone interviewing use computer-assisted survey instruments. Overall, 55 percent of person-level interviews in 2011 were conducted over the telephone.
The NCVS data collection protocol calls for the direct interviewing of each person 12 years or older in the household, with interviews held separately for each household member. Proxy interviewing instead of direct interviewing is permissible for minors (when a knowledgeable household member insists they not be interviewed directly); incapacitated persons; and individuals absent from the household during the entire period of field interviewing.
Three survey instruments support the NCVS: the control card, the basic screen questionnaire, and the incident report.11 Respondents are first screened to determine whether they experienced an incident in which they were victimized during the past 6-month period. In the second part of the interview, detailed information about any incident reported during the screening process is collected. The process is described in more detail below.
The control card is used to build a roster of household members, to determine eligibility to be interviewed, and to obtain basic information about the sample unit, including a record of visits, telephone calls, interviews, and reasons for any noninterview. It lists the name, age, gender, marital status,
11The survey instruments (both screening questionnaire and incident report) are displayed on the BJS website. Appendix C provides the URLs where these instruments can be found.
education, and relationships of all persons residing in a household. The importance of the control card lies in its brief description of all victimizations reported by household members in each interview. It serves as a quick reference during future interviews to ensure that victimizations previously counted are not reported again in error.
The basic screen questionnaire is used to screen for incidents that occurred during the 6-month reference period in which the individual household member may have been victimized. This is a very critical part of the interview (see “Cue Screening Questions” below).
An incident report is completed for each event recorded in the basic screen questionnaire. The report collects detailed information about each victimization, including the time, date, and place; information about the offender, including any relationship to the victim; circumstances of the incident, such as whether a weapon was used; consequences to the victim, including injury and loss of or damage to property; and whether the incident was reported to police. The incident report, along with the screening report, is subsequently used to categorize the incident as to whether it is a crime, and if so, what type of crime. This process is discussed in greater detail in Chapter 7.
Cue Screening Questions
During the screening phase, respondents are asked both directly and indirectly about potential victimizations through short cue screening questions. In an omnibus victimization survey like the NCVS, it would be difficult (and burdensome) to list each specific possible type of victimization in the screening questionnaire. Instead the screener contains cues designed to trigger memories of victimizations through cues related to particular contexts, locations, weapons, and offenders. A respondent may report any victimization at any point during the screening process. Several cue screening questions deal specifically with physical attacks, threats, or sexual activity.
Since [end date for 6-month reference period], were you attacked or threatened OR did you have something stolen from you: (a) at home including the porch or yard; (b) at or near a friend’s, relative’s, or neighbor’s home; (c) at work or school; (d) in places such as a storage shed or laundry room, a shopping mall, restaurant, bank, or airport; (e) while riding in any vehicle; (f) on the street or in a parking lot; (g) at such places as a party, theater, gym, picnic area, bowling lanes, or while fishing or hunting; OR (h) did anyone attempt to attack or attempt to steal anything belonging to you from any of these places?
(Other than any incidents already mentioned,) has anyone attacked or threatened you in any of these ways: (a) with any weapon, for instance, a gun or knife; (b) with anything like a baseball bat, frying pan, scissors, or stick; (c) by something thrown, such as a rock or bottle; (d) include any grabbing, punching, or choking; (e) any rape, attempted rape or other type of sexual attack; (f) any face-to-face threats; OR (g) any attack or threat or use of force by anyone at all? Please mention it even if you are not certain it was a crime.
People often don’t think of incidents committed by someone they know. (Other than any incidents already mentioned,) did you have something stolen from you OR were you attacked or threatened by (a) someone at work or school, (b) a neighbor or friend, (c) a relative or family member, (d) any other person you’ve met or known?
Incidents involving forced or unwanted sexual acts are often difficult to talk about. (Other than any incidents already mentioned,) have you been forced or coerced to engage in unwanted sexual activity by (a) someone you didn’t know before, (b) a casual acquaintance? OR (c) someone you know well?
During the last 6 months [other than any incidents already mentioned), did you call the police to report something that happened to YOU which you thought was a crime?
If the respondent replies yes to any of the cue screening questions, then he or she is asked to briefly describe in his or her own words what happened.
In the incident report, the respondent is further queried in detail about each incident reported in the screener. There are questions that relate to the location of the incident, weapon used (if any), injuries, medical care and expenses, distress as a victim, others present during the incident, etc. Listed below are questions that deal specifically with physical attacks, threats, or sexual activity.
Did the offender hit you, knock you down or actually attack you in any way? Did the offender TRY to attack you? Did the offender THREATEN you with harm in any way? What actually happened?
If the respondent says an attack was attempted or threatened, then she or he is asked:
How did the offender TRY to attack you? How were you threatened? Any other way?
If the respondent reports an unwanted sexual contact, then she or he is asked:
You mentioned some type of unwanted sexual contact with force. Do you mean forced or coerced sexual intercourse including attempts?
If the respondent mentions rape, then he or she is asked:
You mentioned rape. Do you mean forced or coerced sexual intercourse?
If the response is no, then he or she is asked:
What do you mean?
The same queries are used for attempted rapes.
During processing, BJS staff, working with Census Bureau staff, classifies reported incidents into categories using a type-of-crime algorithm. Incidents involving sexual violence or unwanted activity are coded into one of several distinct categories: completed rape, attempted rape, sexual assault with serious assault, sexual assault with minor assault, sexual assault without injury, unwanted sexual contact without force, verbal threat of rape, and verbal threat of sexual assault. Figure 7-4 in Chapter 7 shows a flow chart of this process.
Even if the respondent reported “no” to the first query on rape in the cue questions, the respondent would still be classified as a rape victim if she or he reports a rape at any point in the incident report or at any time during the interview. Thus, classifying a respondent as a victim of rape or sexual assault is a two-stage process in the NCVS methodology, and it can occur through numerous avenues.
BJS uses the following definitions for rape and sexual assault (Bureau of Justice Statistics, n.d.-b):
Rape: Forced sexual intercourse including both psychological coercion as well as physical force. Forced sexual intercourse means vaginal, anal or oral penetration by the offender(s). This category also includes incidents where the penetration is from a foreign object such as a bottle. Includes attempted rapes, male as well as female victims, and both heterosexual and homosexual rape. Attempted rape includes verbal threats of rape.
Sexual assault: A wide range of victimizations, separate from rape or attempted rape. These crimes include attacks or attempted attacks generally involving unwanted sexual contact between victim and offender. Sexual assaults may or may not involve force and include such things as grabbing or fondling. Sexual assault also includes verbal threats.
This section first briefly describes post collection weighting and adjustments in the NCVS. It then describes a recent change in the way that series victimization (repeated victimization of one victim) is handled in summarization. Lastly, the section briefly highlights some of the important public data products that come from the NCVS.
The estimation procedures12 for the NCVS begin with assigning a base weight to each selected unit, which is the reciprocal of the probability of the unit’s selection for the sample. This base weight provides a rough measure of the population numbers represented by each unit in the sample. Next, an adjustment is made to the weight to account for nonresponse, which includes both a household nonresponse adjustment and a within-household adjustment. These adjustments are standard weighing procedures for household surveys.
BJS makes two additional adjustments. First, data from responders in a first interview (both for the household and each individual respondent) are adjusted to minimize the effect of telescoping on what is an unbounded interview.13 The adjustment factor is calculated based on the following ratio. TIS stands for “time in sample.” Thus TIS 1 refers to respondents who are
12The estimation procedures for the NCVS are described in Survey Methodology for Criminal Victimization in the United States (Bureau of Justice Statistics, 2008b). Unfortunately this documentation is out of date (2008), so the panel depended on several internal memoranda to complete its description of the estimation process.
13“Telescoping” is the tendency of respondents to report events that occurred prior to the start of the reference period. In this case, it would be victimizations that occurred more than 6 months ago. The initial interview then forms a “bound” that helps to minimize telescoping in subsequent interviews.
being interviewed for the first time, and TIS 2 refers to respondents who are being interviewed for the second time.
That is, the potential overreporting of crimes in TIS 1 is adjusted to the average over the other times in sample.
A similar adjustment is made for initial interviews in other TIS groups, 2 through 6, denoted as a for reinstated sample cases:
The final weight also includes ratio adjustments to known population totals based on the adjusted counts from the most recent census (currently 2010). The adjustments are applied at two stages:
• In stage 1, ratio adjustments are applied to the data on the basis of PSU-level estimates of individuals by race and zone of residence to more closely match the census PSU estimates for those population totals. PSUs that are included with certainty (self-representing) receive an adjustment ratio of 1.
• In stage 2, ratio adjustments are applied to the weighted counts of individuals from the NCVS to census estimates (adjusted for the undercount) of population totals by sex, race, and age categories.
Thus, for estimation, the final person weight is the product of the values of the base weight and the adjustments described above. The final household weight is the product of all components except the within-household noninterview adjustment component.
Annual estimates of the levels and rates of victimization are derived by accumulating four quarterly estimates (which are not publicly available). The weights of all crimes reported during interviews in that year are summed, regardless of when the crime occurred. The base weight for personal crime is the sum of all person weights.
Series victimization is a category used when a respondent reports that six or more separate but similar victimizations have occurred during the 6-month reference period, but is unable to recall enough details of each incident to distinguish them from one another. Until recently, the NCVS estimates published in Criminal Victimization excluded series victimizations (or included them as a single victimization in certain special reports), which clearly undercounted the total number of all types of victimizations, including rape and sexual assault.
Beginning with Criminal Victimizations, 2011, BJS began including series victimizations directly in its estimates. The NCVS uses the victim’s report of the number of similar victimizations, with a maximum of 10, and collects (and applies to each victimization) detailed information only for the most recent victimization. These new procedures are being applied to all types of victimizations, including rape and sexual assault (Bureau of Justice Statistics, 2012a, p. 13):
BJS now includes series victimizations using the victim’s estimate of the number of times the victimizations occurred over the past 6 months, capping the number of victimizations within each series at a maximum of 10. This strategy for counting series victimizations balances the desire to estimate national rates and account for the experience of persons with repeat victimizations while noting that some estimation errors exist in the number of times these victimizations occurred. This bulletin is the first to include series victimizations throughout the entire report, and all victimizations estimates in this report reflect this new count strategy.
A technical report provides findings on the extent and nature of series victimization (Lauritsen et al., 2012, p. iii):
Including series victimizations in national rates results in rather large increases in the level of violent victimizations; however, trends in violence are generally similar regardless of whether series victimizations are included. The impact of including series victimizations may vary across years and crime types, in part reflecting the relative rarity of the offense type under consideration.
BJS has revised estimates back to 1993 in its online database, accessible through the NCVS victimization analysis tool.16 The effects of this change on the estimates of rape and sexual assault are so substantial that the panel decided to include both estimates—one based on the new method of including up to 10 victimizations in series, and one in which series victimizations are excluded—in our comparisons in Chapters 6 and 7. The revised estimation process means that a very small number of reports have a major impact on the estimates. (This issue is further discussed in Chapters 6, 7, and 10.)
Products and Statistics
The NCVS provides a major database of information about criminal victimizations. BJS’s major annual publication is Criminal Victimization, which includes estimates of rape and sexual assault. Figure 4-1 and Table 4-4 provide the historical trend for the number of victimizations, which shows a steady downward direction with considerable year-to-year instability.17 As explained above, the estimates that include series victimizations and the estimates excluding these (in Table 4-4) are substantially different.
The NCVS has three primary rotating supplements that are conducted on a periodic basis: on school crime, identity theft, and on police-public contact. BJS also pools NCVS data across years to produce special reports, such as Female Victims of Sexual Violence, 1994-2010 (Planty et al., 2013).
In addition to its publications, BJS assists and encourages researchers to undertake their own analysis using the online NCVS victimization analysis tool (see footnote 16), which allows data users to produce custom tables. BJS also offers a visiting fellows program, which broadens the collaboration between the agency and academic scholars. Data users are able to conduct research in conjunction with BJS researchers and staff and have access to facilities and microdata18 for that research. BJS also supports the archiving of NCVS data in the National Archive of Criminal Justice Data that is part of the Inter-university Consortium for Political and Social Research (ICPSR).
16The tool is available at http://bjs.gov/index.cfm?ty=nvat [May 2013].
17It is useful to note that the decline in rape and sexual assault victimizations between 1993 and 2011 is consistent with the pattern of change in all violent crime during these years. The number of all violent crime victimizations in 2011 was only 35 percent of the number estimated for 1993 (calculated using the NCVS victimization analysis tool, Bureau of Justice Statistics, n.d.-a). This trend appears in the United States and in other industrialized nations.
18In this context, microdata are data provided at the individual respondent level rather than summarized data.
TABLE 4-4 Statistics on Rape and Sexual Assault Criminal Victimizations, NCVS, 1990-2011
|Series Criminal Victimization Excluded|
|Year||Number of Criminal Victimizations||Standard Error for Number of Criminal Victimizations||Criminal Victimization Rate per 1,000 People, Aged 12+||Standard Error for Rate|
aCalculated based on formula provided in Bureau of Justice Statistics (BJS) Criminal Victimization, 1990.
bThe estimates published in BJS Criminal Victimization, 1993, 1994, 1995 were revised in 1996 to reflect a methodology change to estimate victimizations for the “collection year” rather than the year in which the victimization occurred.
cBased on errata issued by BJS on June 16, 2011.
SOURCES: Bureau of Justice Statistics (1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002a, 2003, 2004, 2005, 2006, 2007, 2008a, 2009, 2010, 2011, 2012a, n.d.-a).
|Series Criminal Victimization Counted up to Ten Incidents|
|Number of Criminal Victimizations||Standard Error for Number of Criminal Victimizations||Criminal Victimization Rate per 1,000 People, Aged 12+||Standard Error for Rate|