Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 177
Surveying Victims: Options for Conducting the National Crime Victimization Survey –E– Other Victimization Surveys: International and U.S. State and Local Experience Walker (2006) summarizes the basic design features of 62 victimization-related surveys conducted by various countries, as part of an inventory sponsored by the United Nations Economic Commission for Europe and the United Nations Office on Drugs and Crime. About two-thirds of these are standalone surveys specifically focusing on victimization, and the remainder are crime and victimization components of more general, omnibus surveys. All told, these surveys ranged in size from about 400 to 60,000 households (or up to 75,000 people); 7 countries’ surveys include 10,000 or more persons and households, although several of these are omnibus population surveys that include a victimization component. The survey design features of selected international victimization surveys—including the National Crime Victimization Survey (NCVS)—are summarized in Table E-1. Farrington et al. (2004) provide descriptions of both official-report and survey data sources in several countries, and Lynch (2006) summarizes international victimization surveys as a source for cross-national comparison. Some countries’ victimization surveys clearly use the well-established U.S. NCVS as a template; so, too, do several of the victimization surveys that have been fielded by individual states in the United States. Due to their cost, state victimization surveys tend to be one-shot efforts, although some states (e.g., Minnesota) have conducted several replications.
OCR for page 178
Surveying Victims: Options for Conducting the National Crime Victimization Survey Table E-1 Design Features of Nation-Specific Victimization Surveys Design Feature USAa Australiab Canadac England & Walesd Swedene Netherlandsf Scotlandg Switzerlandh Irelandi Sample Cross-Section No Yes Yes (rotated) Yes (rotated) Yes Yes Yes (rotated) Yes Yes RDD No No Yes No No No No No No Clustered Yes Yes Screening Unified No Yes Yes No Yes Yes No Yes Yes Dense cuing Yes No No No No No No No Bounding Prior Interview No No No No No No Short/long ref period No Reference period 6 mo. 12 mo. 12 mo. 12 mo. 12 mo. 12 mo. 12 mo. 12 mo. 12 mo. Mode In-person Yes No No Yes Yes No Yes No Yes Phone Yes No Yes No No No No yes No Central CATI Yesj No Yes No No No No Yes No Self-administered No Yes No No No Yes No No No Series Yes No No Yes Yes No Yes Yes No Free-standing survey Yes No No Yes No Yes Yes Yes No a National Crime Victimization Survey. b Summarizes three national crime victim surveys; see Carcase (2004). c General Social Survey–Victimization. d British Crime Survey. e Victimization component of annual Survey of Living Conditions. f Victim survey conducted by Statistics Netherlands; other victim surveys are conducted by the police, as described by Bijleveld and Smit (2004). g Scottish Crime Survey; for 2004, the survey was renamed the Scottish Crime and Victimisation Survey. h Swiss crime survey, which was used as a basis for the International Crime Victimization Survey; see Killias et al. (2004). i Crime and Victimization component of Quarterly National Household Survey; see Central Statistics Office (2007). j Calls from Census Bureau CATI centers to be eliminated from NCVS in 2007. NOTES: RDD, random digit dialing. CATI, computer-assisted telephone interviewing. SOURCE: Adapted from Lynch (2006:Table 2) and Farrington et al. (2004).
OCR for page 179
Surveying Victims: Options for Conducting the National Crime Victimization Survey In this appendix, we provide additional detail on selected international victimization surveys that are particularly relevant to our discussion of NCVS design features in the main body of the report. We describe the British Crime Survey and the International Crime Victimization Survey in particular. We also provide a fuller description of subnational victimization surveys that have been conducted in the United States. (One long-standing source of some crime victimization information at the state level—the Texas Crime Poll—is not described here, but is summarized in Box 4-2.) E–1 BRITISH CRIME SURVEY The British Crime Survey (BCS) measures criminal victimization among the population age 16 and older in England and Wales; Scotland was included in the earliest versions of the survey, but it now conducts its own victimization survey (as does Northern Ireland). The BCS became an annual data collection in 2001, having previously been conducted on a roughly biennial basis (1982, 1984, 1988, 1992, 1994, 1996, 1998, and 2000). The BCS was originally conceived as “a research tool” designed to“obtain a better count of crime,” “identify risk factors in victimization,” and “examine people’s worry about crime and their perceptions of and contact with the police.” Only with the passage of time did the goal of “provid[ing] more reliable information about trends in crime” become a principal focus of the survey (Jansson, 2007:4). The BCS is administered by the Home Office of the United Kingdom, the duties of which correspond—in part—to those of the U.S. Department of Justice. Unlike the NCVS, for which the Census Bureau is the data collection agent, the United Kingdom’s principal statistical agency and data collector—the Office for National Statistics—plays no role in the conduct of the BCS. (However, the placement of the BCS between the Home Office and the Office for National Statistics has been debated in reviews of the British government’s crime statistics programs, discussed in Section E–1.c.) An external research organization, BMRB Social Research, has been engaged as the data collection agent for the BCS since 2001. However, the contract to conduct the BCS is retendered every three years (Walker, 2006:3–4). Owing to its annual nature, the BCS employs a 12-month reference period; interviews are conducted continuously throughout the year, so that 1-year window shifts depending on the interview date. The BCS target population consists of “households in England and Wales living in private residential accommodation” and the “adults aged 16 and over living in such households”; accordingly, the BCS does not attempt to interview institutional populations (Grant et al., 2007:§2.2).
OCR for page 180
Surveying Victims: Options for Conducting the National Crime Victimization Survey E–1.a Sample Design The BCS uses the British Postcode Address File as its frame; “whole postcode sectors” are the primary sampling units (PSUs) for the survey and are selected after stratifying by basic demographic variables and Police Force Area (PFA) (Grant et al., 2007:§2.4–2.6). The current design of the BCS includes a base sample as well as two specialized “boost” components. As of 2004, the BCS base sample is designed to ensure representation—through at least 1,000 interviews—in each designated PFA in England and Wales. The Home Office’s decision to make the BCS an annual survey beginning in 2001 was accompanied by a doubling of sample size, since the generation of BCS estimates at the PFA level was also taken as a goal for the survey (Smith, 2006:3). Hence, for 2005–2006, the total BCS base sample size was approximately 47,000 (Grant et al., 2007). Sampling targets are still constructed so that the larger jurisdictions are allocated higher numbers of interviews—e.g., the “Metropolitan” PFA consisting of the Metropolitan (London) Police and the City of London Police was targeted for about 3,500 interviews in 2005–2006. To promote some continuity, half of the PSUs used in one year’s sample (e.g., 2004–2005) are retained for use in the next year’s sample (2005–2006), although fresh addresses are selected from those PSUs (and lists are checked so that no address is selected two years in a row). Over time, the BCS response rate has ranged from 73 to 83 percent; it has been at about 75 percent since 2001 (Jansson, 2007:5). E–1.b Supplements in the BCS From 2003 to 2006, the Home Office conducted a conceptual complement to the self-response victimization survey of the BCS: the Offending, Crime and Justice Survey (OCJS), which is a national self-report survey of offending behavior. (The survey also covers other antisocial behaviors, such as drug and alcohol use; questions on perceptions of antisocial behavior had earlier been fielded as a module in the 1992 survey.) The OCJS is particularly oriented at measuring juvenile delinquency and so relaxes the BCS age constraint, collecting information from respondents as young as 10 years old and oversampling persons ages 10–25 so that they represent roughly half of the basic sample. The OCJS is designed as a longitudinal study, with multiple contacts of the same people, in order to permit measures of trajectories of violence within the study period. The size of the core sample beginning in 2003 was just over 10,000 people; by 2006, about 4,100 of the respondents had been contacted in previous waves of the survey, and a fresh sample brought the total sample size to about 5,000. The OCJS featured multiple response modes within the same interview: standard computer-assisted personal interviewing (CAPI) was used for most questions, but computer-
OCR for page 181
Surveying Victims: Options for Conducting the National Crime Victimization Survey assisted self-interviewing (CASI)—in which respondents read questions from a laptop computer screen and answered them directly—was used for more sensitive topics. Audio-CASI, in which respondents could hear the questions through headphones, was used for some respondents with language difficulties. Other modules or supplements that have been added to the BCS over time include (Jansson, 2007): Perceptions of and confidence in the criminal justice system, including judges and the probation service (conducted regularly since 1996); Interpersonal violence, using CASI methods to try to promote fuller reporting of sensitive incident types, such as domestic violence, sexual victimization, and stalking (first conducted in 1994 and regularly thereafter); and Separate modules of questions on mobile phone theft (first in 2001), fraud and technology crimes (first in 2002), and identify theft (first in 2005). E–1.c Reviews of British Crime Statistics As part of a larger review of official statistics collected in the United Kingdom and their effectiveness in meeting the needs of users, a Statistics Commission (2006) was established and issued a final report on crime statistics in July 2006. Separately, the Home Secretary tasked an independent review board headed by Smith (2006:1) as follows: The Home Secretary is concerned that public trust in the crime statistics produced by the Home Office has declined to such an extent that it is no longer possible to have a debate about alternative criminal justice policies on the basis of agreed facts about the trends in crime. He wishes to be advised on what changes could be made to the production and release of crime statistics so that public trust is re-established. In addition, he wants the Review Group to examine the key issues raised by the Statistics Commission about crime statistics and to make practical recommendations to the Home Secretary as to what changes are needed to address those issues…. The Smith (2006) review was published in November 2006. In their focus on user needs, the basic tasks of these two review efforts in the United Kingdom are similar to our panel’s task, although their mandates are considerably broader. The conduct of the program for official reports to police was in scope for both studies (while administration of the Uniform Crime Reporting program is not part of ours), let alone the Statistics Commission’s broader task of reviewing data collections on other social and
OCR for page 182
Surveying Victims: Options for Conducting the National Crime Victimization Survey Box E-1 Recommendations of the United Kingdom Statistics Commission (2006) Responsibility for the compilation and publication of crime statistics should be located at arm’s length from Home Office policy functions and with clear accountability within the evolving framework of the government statistical service. Treasury and Home Office Ministers should consider together a fully developed business case for moving responsibility for the British Crime Survey to the Office for National Statistics and should publish their agreed view with supporting arguments. The Home Office, and others as appropriate, should make changes to the presentation of the recorded crime figures in order to communicate better the main messages. These steps include: changing the definition of violent crime; greater distinction between British Crime Survey results and police recorded crime data and the uses for which each source is appropriate; ensuring regular reviews of statistical classifications. Existing local data should be better used to improve the quality and range of statistics on crime. This could be achieved through police forces agreeing to publish, in a co-ordinated way, standardised comparable analyses at a local level. These analyses need not necessarily be drawn together and published as official statistics by the Home Office but must be consistent with those that are. Comparability of crime statistics between the various countries within the UK should be improved, identifying and addressing areas of statistics where there are problems. Technical research should be carried out (to a published timetable) to develop a set of weighted index measures of ‘total crime’ and promote debate on which, if any, of these measures should be adopted alongside the current basic count. SOURCE: Excerpted from Statistics Commission (2006). demographic characteristics. For reference, the six formal recommendations of the Statistics Commission are listed in Box E-1, and selected recommendations (those most relevant to the BCS and the topics in this report) of the Smith commission are listed in Box E-2. The Statistics Commission (2006:7) expressed particular concern “that the broad statistical messages about crime” from both the BCS and the official police report data “were being lost against a backdrop of confused reporting.” Hence, its Recommendation 3 stresses the need for statistical communications to be clear about the nature and limitations of their data sources. The Smith independent review noted the value of complementary information provided by the BCS and official police reports, recommending that results from the two series continue to be published together (2.7) and that main messages from both series (and others, as appropriate) be explored
OCR for page 183
Surveying Victims: Options for Conducting the National Crime Victimization Survey Box E-2 Selected Recommendations of the Smith (2006) Independent Review of British Crime Statistics 2.1 The British Crime Survey sample frame should be extended to include those under 16 and those living in group residences as soon as practical after taking the advice of those with relevant expertise and piloting the changes. In addition, research should be carried out on the victimisation of homeless and institutionalised populations. 2.2 We recommend that the Home Office should carry out a survey of commercial and industrial victimisation every two years. 2.3 We recommend that the Home Office should publish within 12 months an action plan for what it proposes to do to measure those crimes which are either not included in the present crime statistics or are poorly measured by them. 2.4 We recommend that the Home Office set up a standing panel of independent experts to provide regular review of and comment on methodological and analytic issues relating to the BCS and its other crime surveys. 2.7 The Home Office should continue to publish police recorded crime data and the BCS together. 2.8 We recommend that national crime statistics should be published annually and include a full commentary on the state of crime, drawing on all appropriate data sources. 2.10 We recommend that whenever Home Office statistical reports include interpretation or assumptions on the part of the authors these should be flagged frankly and openly on their first appearance in the report and the basis of those judgements should be referenced and made available. 2.11 We recommend that the Home Office should attach to each of its statistical series a statement clearly identifying the strengths and weaknesses of the particular series and the aspects about which professional judgements may need to be made. 2.12 In order to build trust, the Home Office should ensure that the release and statistical commentary on national crime statistics are quite clearly separated from political judgements or ministerial comments and should ensure the accuracy of any statements made about the statistics, whether in press releases or ministerial comments. 2.13 We recommend that the Home Office redefine violent crime in crime statistics to only include those crimes which actually cause physical injury or where the threat to inflict such injury is likely to frighten a reasonable person. 3.1 The Home Office should make the provision of local crime information a central part of its crime communication strategy and not just rely on publishing national crime statistics. 4.1 We recommend that the Home Secretary should put in place a regulatory environment which ensures that there is an actual and perceived separation between those who produce statistical data and commentary on crime (a “Back Office” function) and those who are responsible for policy advice and will be judged on the basis of the data (the “Front Office”) be they in a police force, the Home Office or elsewhere. SOURCE: Excerpted from Smith (2006).
OCR for page 184
Surveying Victims: Options for Conducting the National Crime Victimization Survey by analysts (2.8), while taking care not to attribute findings or conclusions to the government or the Home Office generally (2.10). Several of the commission’s recommendations are driven by a concern about crime statistics being issued by the program agency responsible for developing policy on crime. That is, the commission “[did] not believe trust [in crime statistics] can be built up whilst the same Ministers, advisers and senior officials are directly involved both in publishing the figures and in setting out the Government’s position” (Statistics Commission, 2006:12). The Smith (2006:2) review summarized the basic problem: Every Home Secretary [has] accepted that a major purpose of Home Office policy is to reduce crime and for the present government this has meant setting performance targets for crime reduction. As a result, crime statistics have become a key metric for judging the performance of the Home Office and therefore central to debates between government and opposition. This has meant that crime statistics have been subject to a quite new degree of scrutiny and their release and handling have become politically much more sensitive. Hence, the commission’s Recommendation 1 emphasizes the need for an arm’s length distance between the Home Office’s statistical operations and its policy operations. For the BCS, in particular, the commission’s Recommendation 2 suggested a way for a more-than-arm’s-length separation: transferring authority for the BCS from the Home Office to the Office for National Statistics (ONS), particularly if efforts to provide ONS greater independence as a nonministerial department came to fruition.1 The Smith independent review reiterated the need for a clear divide between statistical releases and policy decisions (Recommendation 2.12) but disagreed on the administrative placement of the BCS. “We believe [the BCS] should remain in the Home Office because as well as being a source for the national crime statistics, it is one of the most important research tools and sources of information for the Home Office to manage the crime problem” (Smith, 2006:6). The Smith review argued for expanding BCS coverage to persons under age 16 as well as to people in group housing and, possibly, institutions (Recommendation 2.1). Smith (2006:10) recognized that these changes “would be methodologically difficult” and would require careful development and piloting. The interviewing of children, in particular, raises parental consent issues that would take great care; the review text takes the more nuanced position that “interviews should be extended as far below 16 as proves prac- 1 “Although we see the case for transferring responsibility as strong, the Commission does not have available to it all the relevant information on costs and capacity to make a firm recommendation”—hence the recommendation’s call for a “business case” for transferring authority for the BCS (Statistics Commission, 2006:13).
OCR for page 185
Surveying Victims: Options for Conducting the National Crime Victimization Survey tical in a regular household-based survey.” Echoing the commercial victimization surveys that were originally part of the U.S. National Crime Survey—but dropped on the recommendation of the National Research Council (1976b)—the Smith review recommended that consideration be given to a biennial standalone study of commercial and industrial victimization (2.2). Both the Statistical Commission and the Smith review observed the “substantial demand from local delivery organisations for more and better local level information on crime” (Statistics Commission, 2006:15). E–2 INTERNATIONAL CRIME VICTIMIZATION SURVEY Administered by the United Nations Interregional Crime and Justice Research Institute (UNICRI), the International Crime Victimization Survey (ICVS) is a set of surveys that aspires to provide standardized measures of victimization for cross-national comparison. A first round of the ICVS was fielded in 1989, after two years of development by a working group; subsequent rounds were fielded in 1992, 1996–1997, and 2000. The survey began with about 15 countries (and 2 separate cities), while the 1996–1997 round was said to interview 135,465 people in 56 countries. The ICVS is conducted through computer-assisted telephone interviewing (CATI) supplemented with personal interviewing; the face-to-face personal interviews are typically restricted to a sample of people in a participant nation’s capital city.2 A survey firm from the Netherlands, Interview, coordinated the work in most of the participating industrialized countries, subcontracting with firms or companies in those countries for sample selection and fieldwork. ICVS participants bear the costs of fieldwork in their countries; they use a core set of questions in their interviews—pursuant to the goal of standardization—but have some flexibility for adapting the sample or the survey instrument based on their own needs. For instance, in Australia, the Australian Institute of Criminology oversaw the 7,000 ICVS interviews conducted in that country in the 2004 wave of the survey. The institute customized the ICVS to attend to matters of interest to the Australian government by making 1,000 of the interviews a “booster sample” of migrants from Vietnam and the Middle East, as well as by adding groups of questions on such issues as licensing and storage of firearms and perceptions of safety while using public transportation. They also dropped questions on sexual assault, judging that the institute’s separate Violence Against Women Survey provided more reliable results (Johnson, 2005). 2 Exceptions—cases in which personal interviews are favored over CATI—include places in which telephone penetration was not deemed to be high, including Malta, Northern Ireland, and rural Spain. ICVS work in Japan was also done face-to-face (http://www.unicri.it/wwd/analysis/icvs/methodology.php).
OCR for page 186
Surveying Victims: Options for Conducting the National Crime Victimization Survey The ICVS covers three crimes against persons (assaults and threats, robbery, and personal theft) and six crimes against households (burglary, attempted burglary, motor vehicle theft, theft from motor vehicles, motorcycle theft, and bicycle theft). Due to the frequency of the survey, the ICVS uses a 5-year reference period, asking respondents to recall incidents within that range; however, it also generally asks about events that occurred within the last 12 months, so that 1-year estimates may also be produced. As described on the ICVS methodology web page,3 ICVS sample selection typically includes 1,000–2,000 households in participant households, reached through random-digit dialing (RDD) methods. A randomly selected person from each household is chosen to complete the interview (and is not replaced if that person refuses to participate). Contacts continue until the desired number of interviews is completed. Overall, this CATI component had a 67 percent response rate in 11 industrialized countries in the 1996 ICVS round, with rates varying from 40 percent (in the United States) to 80 percent or greater. When face-to-face interviewing was used, procedures tended to be somewhat more ad hoc. Effort was typically concentrated in the country’s capital city, with the intent of obtaining about 1,000 respondents. The ICVS methodology page indicates that the sampling for face-to-face interviews “was generally hierarchical,” starting “with identifying administrative areas within the city, followed by a step-by-step procedure aiming at identifying areas, streets, blocks, households and finally the household member aged over 16 whose birthday is next.” E–3 VICTIMIZATION SURVEYS AS PART OF A BROADER SOCIAL SURVEY: CANADA AND AUSTRALIA Some countries that do not have standalone surveys of victimization still gather related information through major supplements to other, more omnibus surveys. Two examples are Canada and Australia. Canada’s system for the measurement of crime and victimization is similar to that of the United States. Official counts of crimes reported to the police have been collected annually since 1962 by the Uniform Crime Reporting Surveys (e.g. Gannon, 2006), while victimization measures are obtained as part of Statistics Canada’s omnibus General Social Survey.4 Since 1988, a victimization component in the General Social Survey has been in- 3 See http://www.unicri.it/wwd/analysis/icvs/methodology.php. 4 Despite the “Surveys” part of the name, the Canadian UCR is intended as a complete census of Canadian law enforcement agencies. As a further parallel to the U.S. model, the Canadian Uniform Crime Reporting Surveys include a component dubbed UCR2—gathering detailed incident-based information—that is similar to the developing American National Incident-Based Reporting System (NIBRS; see Section D–2). As of 2005, 127 Canadian police depart
OCR for page 187
Surveying Victims: Options for Conducting the National Crime Victimization Survey cluded on a quinquennial basis (however, the third round was conducted in 1999, not 1998). Starting with the 1999 administration, the survey was renamed the General Social Survey–Victimization (GSS-V, replacing General Social Survey–Personal Risk). The GSS-V is strictly a telephone survey, conducted by RDD from three centralized CATI centers in Statistics Canada regional offices. A Statistics Canada web page describing the survey’s methodology notes that the RDD approach excludes persons in households without telephones (estimated as representing “less than 2% of the target population” of Canadians 15 years of age or older) and persons with only cellular telephone service (“again, this group makes up a very small proportion of the population, less than 3%”).5 Consistent with the survey’s frequency, the GSS-V asks respondents to recall incidents within the past five years. Working with this long reference period can require repeated sets of questions, since circumstances may have changed within the time window. For instance, the first set of questions related to physical or sexual abuse by a spouse or partner asks “whether, in the past 5 years, your current spouse/partner has done any of the following to you” (emphasis in original). A follow-up question asks whether such acts of violence have occurred more than one time; if so, respondents are asked how many times the abuse has taken place in 12 months. In either event, the specific month and year is sought for the most recent incident. Subsequently, another module of questions asks about incidents of abuse at the hands of one’s previous spouse or partner. Like the NCVS, the GSS-V has included topical modules sponsored by other agencies: specifically, the Interdepartmental Working Group on Family Violence funded questions on domestic and elder abuse, and the Solicitor General Canada funded questions on public perceptions on imprisonment in the 1999 administration of the survey. However, both of these sponsored supplements were dropped from the 2004 GSS-V. The GSS-V sample was selected based on 27 strata, with major cities (census metropolitan areas) representing their own strata and partitioning nonmetropolitan areas into about 10 strata. Within each strata, respondents were contacted through RDD after eliminating “non-working banks” (telephone exchanges known not to work). When a household was contacted, basic demographic information on all household members was elicited; one household member age 16 or older was then randomly selected from this list to complete the interview. ments completed the UCR2 survey, “represent[ing] 62% of the national volume of reported” crimes (Gannon, 2006:14). 5 See http://www.statcan.ca/cgi-bin/imdb/p2SV.pl?Function=getSurvey&SDDS=4504&lang=en&db=IMDB&dbg=f&adm=8&dis=2; as of June 21, 2007, this page showed a “last modified” of October 27, 2005.
OCR for page 188
Surveying Victims: Options for Conducting the National Crime Victimization Survey For the 2004 administration of the Canadian survey, the overall response rate was estimated as 75 percent. The Australia Crime and Safety Survey—administered by the Australian Bureau of Statistics—is a periodic supplement to the omnibus Monthly Population Survey. Specifically, it is part of the April Labour Force Survey, itself a large supplement to the Monthly Population Survey. Most recently, the Crime and Safety Survey was conducted between April and July 2005; it had previously been fielded at the national level in 1975, 1983, 1993, 1998, and 2002 (Australian Bureau of Statistics, 2006). E–4 STATE AND LOCAL VICTIMIZATION SURVEYS Requests for proposals (RFPs) under the Bureau of Justice Statistics (BJS) State Justice Statistics Program for Statistical Analysis Centers (SACs) in 2000 identified local victimization surveys as a high priority. Specifically, the RFP indicated that “SACs receiving funds under this theme must agree to use the BJS developed Crime Victimization Survey software, which can be easily modified to meet State/local priorities and requirements” (Alaska Justice Statistical Analysis Center, 2002:1). E–4.a Alaska The Alaska SAC—the Justice Center at the University of Alaska, Anchorage—secured BJS funding in fall 2000 to conduct a localized version of the NCVS, with the intent of providing victimization estimates for the city of Anchorage. At the time of the original contract award, a planning group was to consider methods for conducting a victimization survey for rural Alaska; formal work in that direction does not appear to have occurred. More recently, the Alaska SAC has switched focus from a dedicated victimization survey to a general Alaska Community Survey. E–4.b Illinois In 2002, the Illinois Criminal Justice Information Authority contracted with Bronner Group, LLC, to conduct its first victimization survey. The survey was sent by mail to about 7,500 residents; the sample was drawn by the Illinois secretary of state’s office from its database of holders of driver’s licenses and state-issued identification cards. Unlike Minnesota’s victimization survey, described below, the sample was apparently filtered to include only persons age 18 or older. The survey methodology and results are documented by Hiselman et al. (2005); of the initial sample, some 23 percent proved to be unreachable (e.g., the person had moved or was deceased). In
OCR for page 189
Surveying Victims: Options for Conducting the National Crime Victimization Survey all, 1,602 completed questionnaires were received, for a response rate of 28 percent (Hiselman et al., 2005:v). E–4.c Maine The first Maine Crime Victimization Survey was conducted in fall 2006 by the University of Southern Maine’s Muskie School of Public Service, which houses the state’s SAC. A telephone questionnaire was administered to 803 adults (RDD until target sample size reached) between August and December 2006. The survey report (Rubin, 2007) indicates that the effort in Maine was modeled after Utah’s victimization survey (Haddon and Christenson, 2005), described below. The Maine survey also adopted questions on identity theft from the 2004 NCVS. E–4.d Minnesota The Minnesota Crime Survey is one of the few state victimization surveys to be fielded multiple times on a semiregular basis (1993, 1996, 1999, 2002, and 2005). This three-year cycle has allowed the content of the survey to evolve over time; for instance, the 2002 survey added a question about perceived fear of a terrorist attack as well as additional questions on domestic abuse (Minnesota Justice Statistics Center, 2003). Conducted by Minnesota Planning, the survey is conducted by mail using driver’s license (or state-issued identification card) rolls as the sampling frame. Generally, the use of this frame means that the survey covers persons age 16 and older, since that is the legal age for obtaining a driver’s license; however, since identification cards can be issued to persons of any age, it is possible for someone under 16 to be selected for the survey. The Minnesota survey used an advance-notice postcard to alert sampled respondents that a questionnaire should soon arrive, and a reminder postcard was issued if no response was received within three weeks. Survey collection ended after five weeks. In 2002, the survey obtained a 41.6 percent response rate on a mailout of 10,013 questionnaires. E–4.e Utah Utah’s BJS-affiliated SAC is the Research and Data Unit of the state Commission on Criminal and Juvenile Justice, in the office of the governor. In 2005, the Utah SAC conducted its third Utah Crime Victimization Survey; it used a 12-month reference period, so that respondents were asked to report crimes that occurred in calendar year 2004. However, the 2005 version of the survey also added a “lifetime victimization” question to each of the specific crime types covered by the survey. The 2005 survey also expanded a battery of questions about attitudes and perceptions of crime (e.g., “When
OCR for page 190
Surveying Victims: Options for Conducting the National Crime Victimization Survey you leave your home, how often do you think about it being broken into or vandalized?”). Unlike its predecessors—covering crimes in 2000 and 2002—the 2005 administration of the Utah victimization survey was the first to be conducted by telephone. Households were reached by RDD, with calls made until a target sample size was achieved. The 2005 administration of the survey included 2,002 interviews and found that 41.3 percent of the respondents had been a victim of one of the crimes covered by the survey in 2004, a slight increase from 2002’s estimate of 36.6 percent (Haddon and Christenson, 2005:4). The Utah SAC conducted a separate study—focused on women age 18 and older—specific to rape and sexual violence. Comparing this survey to the state victimization survey, Haddon and Christenson (2005:18–19) note that the separate study’s findings about the likelihood of a respondent’s having been raped sometime during her lifetime were “not unlike those of the 2004 victimization survey”; however, the standalone survey suggested a much lower level of reporting to the police. They conclude that the lower reporting total “is likely more accurate in that [the rape survey] included a much larger group of individuals who had been sexually victimized.” E–4.f Wyoming With support from BJS, the Wyoming Statistical Analysis Center (2004) and the Wyoming Division of Victim Services fielded a victimization survey in October 2003. The survey was conducted by random-digit dialing with an extension list stratified by county with selection probability proportional to population size. The survey yielded 1,439 interviews. Anticipating a relatively low victimization rate, the survey was structured so that interviews did not simply end if a respondent indicated no incidents in the screener questions. Instead, those respondents were routed through a set of questions on attitudes toward criminal justice and awareness of the Division of Victim Services, and those respondents who did experience victimizations were guided through the analog of the NCVS incident form.