Which of the following would be considered a limitation of victimization studies?

Types of Crime Covered

NCVS crime coverage includes both violent and property crimes associated with households and individual victims. Household property crimes include attempted and completed burglary, motor vehicle theft, and household theft (e.g., theft from the yard). Personal crimes (or crimes against individuals) include attempted and completed violent crimes of rape, sexual assault, robbery, aggravated and simple assault, and personal thefts (such as purse snatching). Homicide is excluded due to the victim-based nature of the survey. Arson is excluded as well. Crimes against children under age 12 are not measured because the Census Bureau has determined that the instrument is not valid for younger respondents. The NCVS is able to measure crime more broadly than the UCR because incidents not reported to police are included.

The determination of whether an event is a crime and if so, the type of crime, is not made during the interview. Rather, the characteristics of each reported event are detailed during the interview. Final classification of the crime type occurs during data processing and is based on information provided in the incident report. This is done to minimize definitional variations and to ensure that the use of legal terms such as burglary or robbery matches definitions used by the police and the UCR. Additionally, events are classified according to the most serious aspect of the victimization. For example, if a respondent reports that they were hit (assaulted) and had something taken from them by the use of force (robbed) during the same event, a “hierarchy rule” is applied and the event is classified as a robbery. Specific victimization definitions can be found in Bureau of Justice Statistics (BJS) victimization reports.

Survey Data Weighted for Crime Severity

Increasingly, surveys of victims, such as the NCVS administered by the Bureau of Justice Statistics, U.S. Department of Justice, have come to occupy a central place in the measurement of crime. By interviewing respondents in a national probability sample of households, researchers can generate highly valuable estimates of the incidence of crime, information that does not depend on whether the victim reported the crime to the police or the manner in which the police responded to crimes that were reported. From these surveys, criminologists have learned that a considerable amount of crime actually occurs but is not reported to the police and cannot be included in crime rate calculations. Clearly, therefore, victimization surveys are an important adjunct to police statistics. Similarly, self-report surveys are used to illicit data on the hidden dimension of crime—the crimes that people commit for which there may not have been a report or there was no arrest. These data not only address the incidence of crime but also provide information about the prevalence of criminality across sociodemographic correlates of official statistics versus hidden crime.

It is not only feasible but also highly desirable to apply the Sellin–Wolfgang severity scaling procedures to these surveys. In so doing, research could address topics other than just the incidence of unreported victimization or the prevalence of hidden crime. For example, comparisons could be made between the severity of offenses that victims report to the police and those that victims choose to let go unreported. Are there strong severity differences that explain the reporting phenomenon? Furthermore, research could investigate whether different segments of the population experience different degrees of harm. In terms of self-report data, the application of severity scoring procedures provides the opportunity to compare the severity of offenses that are known to the police with those for which the offender was never caught. In any event, victimization and self-report data weighted for severity components would enhance the important function that these measurement approaches serve in augmenting official crime statistics.

3.2 Evidence from Victimization Studies

To determine the extent of discrimination in international justice systems, some scholars have looked to victimization surveys, reasoning that if arrest data matches victimization reports, the racial disparities evidenced in arrests could be explained by disproportionate offending rather than by bias.

Among Western countries, only the US, England, Canada, and The Netherlands conduct long-term national victimization studies. US data shows that minorities fall victim to violent crime at a rate that exceeds their representation in the population (Fig. 1). US data also shows that violent crime is largely an intraracial event. From 1976 to 1996, 94 percent of cases involving Black murder victims involved a Black offender. The corresponding percentage was 85.6 for Whites.

The victimization data largely points away from invidiously or blatantly discriminatory targeting of offenders by the justice system. Instead, the data suggests that differential rates of offending by racial and ethnic groups contribute a great deal to the disparity in the US justice system. Hindelang's (1978) classic study shows significant correspondence between the distribution of the perceived race of offenders from victimization survey results and the distribution of arrestees by race from official statistics. While there are well-documented limitations to this approach, existing research has prompted Tonry (1997) to conclude that ‘in countries in which research has been conducted on the causes of racial and ethnic disparities in imprisonment, group differences in offending, not invidious bias, appear to be the principle cause.’

5.1 Measuring Delinquency

Debates have raged for the last few decades on the proper way to measure delinquency and test the classical theoretical perspectives. There are three major sources of data that have been used, self-reports of delinquent behavior, victimization surveys, and official accounts (e.g., arrests, court records). Unfortunately, these sources of data do not always agree. In general, survey-reports of delinquency show weaker associations between social status (e.g., poverty, race, gender) and delinquency than official records. Whether or not the differential correlations of social factors with official records and self-reports reflect bias by the State or methodological weaknesses in survey research is still subject to debate. There has emerged a consensus, however, that the two sources of data yield reasonably similar patterns when the object of inquiry is serious and persistent delinquency (Hindelang et al. 1981).

Victimization

A related way of measuring crime is by asking respondents questions about their victimization experiences. In the same way that respondents to a self-report delinquency survey are asked about behaviors that they have committed, a self-report victimization survey asks respondents to self-report crimes that they have experienced. Again, similar to delinquency surveys, victimization surveys allow for the inclusion of multiple follow-up questions that can give the researcher a fuller understanding of the circumstances surrounding any given victimization experience.

Since 1973, the Census Bureau has conducted the National Crime Victimization Survey (NCVS). Prior to 1992, this survey was known as the National Crime Survey. The Census Bureau interviews approximately 50,000 households annually; each household selected for the NCVS is interviewed a total of seven times over a 3-year period. The NCVS obtains detailed characteristics of a criminal event after the respondent (or another member of the household) claims to have been a victim of a crime. The survey is organized in such a way that a series of screening questions is used to identify any victimization experiences. Once the respondent (or other household member) reports a victimization, there is an additional series of questions that is asked to provide detailed information on the context of the victimization experience. These follow-up questions focus on characteristics such as the time and the location of the crime, the level of physical harm (if any), the value of property damaged or stolen (if any), and the perceived characteristics of the offender (e.g., age, sex, and race) if the offender was viewed by the victim.

One of the key benefits to the use of victimization survey data is that it has allowed researchers to gain a better understanding of the conditions or the circumstances that make a crime event more or less likely. As victimization data have become more accessible to researchers, analyses have helped to further develop theoretical frameworks for understanding victimization risk. Newer research relying on victimization survey data has focused on the dynamics of crime, with a particular interest in the interactions among the victim, the offender, and the social and physical space of the criminal event.

Crime Reporting2

We are going to discuss three annual publications that document crime statistics in the United States:

Crime in the United States—Uniform Crime Reporting (UCR) The Summary System

The National Incident-Based Reporting System (NIBRS)

National Crime Victimization Survey (NCVS)

Events Occurring in Settings Not Subject to Surveillance

Many events of interest to researchers occur in settings that are not routinely subject to surveillance. For example, crimes such as assaults, burglaries, robberies, and rapes occur in settings and ways that make identifying and apprehending offenders difficult. They only become known if the victim reports the crime to the police or if he or she happens to be a respondent in a crime victimization survey.

Of course, while settings are subject to routine surveillance is increasingly becoming an open question. For example in September, 2002, Madelyne Toogood turned herself in to the police because she was observed beating her child by a surveillance camera in a store parking lot. The fact that the video was carried on national television contributed to her turning herself in.

There are also events that occur in protected settings such as homes. For example, for decades women and children were physically assaulted and otherwise victimized in their homes without outside attention. Changes in legislation required reporting of child abuse and more careful accounting of spousal violence. Additionally, abused and battered women now have alternatives to remaining in an abusive relationships such as hotlines and women's shelters. Such changes led to the creation and storage of records that can be used for research.

4 Other Sources of Error

There are other sources of nonsampling error that can effect survey data. Errors caused by interviewer bias, panel rotation, survey mode, coding, and data entry are all errors that should be avoided in conducting surveys. We will briefly consider panel rotation and survey mode errors.

Panel rotation error occurs when responses to a survey change depending on the length of time the panel has been included in the survey. Bailar (1975) described the effects of time-in-sample for the US Current Population Survey. In general, the longer a person has been in the sample, the less likely the person is to report being unemployed. A similar effect is found in the US National Crime Victimization Survey; persons in the survey for a longer time are less likely to report a victimization. Possible explanations for this phenomenon include that respondents learn to answer the survey so as to minimize the amount of time it takes to complete the survey or that differential dropout leads to a biased sample over time.

Survey modes include in-person, telephone, self-administered paper-and-pencil questionnaires and, most recently, self-administered Web questionnaires. The same questions asked using different survey modes can elicit different responses (see Sample Surveys: Cognitive Aspects of Survey Design). For example, respondents are more concerned about their self-images in personal interviews than on paper-and-pencil surveys. Thus, they may be more likely to report personal problems on a self-administered questionnaire than in a face-to-face interview. On the other hand, security and confidentiality concerns may make respondents less likely to respond truthfully to surveys on the Web. The accuracy of responses to Web surveys is an area of ongoing research.

Additional information on these other sources of nonsampling error can be found in references such as Groves (1989) and Biemer et al. (1991). An excellent example of a survey for which all sources of error, both sampling and nonsampling, are explored in detail is the US Current Population Survey (see US Department of Commerce 2000).

Essential outcomes

The Hetrick family was fortunate to hear about their son being bullied. Most families will not. Not from victims. The bullying victim is ashamed to talk about it. And not from the bully. The bully knows that his or her behavior is wrong and will not talk about it.

That is not the whole picture. An unacceptable percentage of K–12 students across your district have the same problems but will not ask for help. At the same time, percentages of bystanders are living in fear that they may be victimized ... or even more fearful they may be questioned about what they witnessed. Your job is to identify both the victims and the victimizers–and help them.

Prevention is the only option – key to student success and wellness.1

Bullying behavior continues to present a tragedy of disturbing proportions. According to the UNESCO Institute of Statistics, “one-third of the globe's youth is bullied;”2 according to the National Crime Victimization Survey (School Crime Supplement/SCA), 2013 data reveals about 22% of students 12–18 reported being bullied at school during the school year3; and, according to the 2019 Indicators of School Crime and Safety report, the percentage of students 12–18 who reported being bullied at school between 2005 and 2017 decreased from 29% to 20%, with no measurable change between 2015–17. Based on these statistics, about one in five children are still reporting bullying incidents whereas one in four bullying incidents was the norm prior to 2013. Either norm is unacceptable, yet such a gain is encouraging. When historical trends were compared from 2007, the data indicates a consistent reduction in student reports from as high as 32%. And while generational and cultural fluctuations will evolve over time, prevention protocols are working and remain necessary to the well-being of our students emotionally, physically, academically.

Reality doesn't change itself. We need to act.

Margaret Wheatley4

Providing a safe, thriving, and welcoming environment that fosters both academic student growth and socio-emotional wellness is the foremost goal of any school setting, and for all school and community stakeholders. This can be accomplished by developing, communicating, and implementing a visionary, yet practical strategic plan that details action steps for preparedness and response, and when needed, threat assessment and/or intervention. Criminologist Stuart Henry states:

Our nation's schools should be safe havens for teaching and learning, free of crime and violence. Any instance of crime or violence at school not only affects the individuals involved but also may disrupt the educational process and affect bystanders, the school itself, and the surrounding community. (https://www.air.org/resource/indicators-school-crime-and-safety-2007).

Students are more likely to experience success and growth in all areas of their life if they feel safe to learn. For many students, the safest place for them to be is at school. So, as school stakeholders, shouldn't we prepare and provide our students with the safest learning environment possible? Researcher Karen M. Hawkins states:

The goal of any school is to be a place of learning and for students to learn, they must feel emotionally and physically secure. In an arena where lethal shootings can share headlines with teasing and bullying, safety for school children is a critical challenge of educators.5

This chapter will focus on providing insight into the nature of bullying, bullycide, cyberbullying, sexting, dating violence, sexual assault and bystanders. It also discusses viable strategies for creating a schoolwide discipline plan and building a safe school environment. Practical suggestions, resources, and best practices are also included for your consideration and potential implementation.

Aided and Unaided Recall

In general, unaided (or free) recall produces less complete reporting than aided recall. It may also produce fewer erroneous reports. Cues and reminders serve to define the scope of eligible events and stimulate recall of relevant instances. A cuing approach was employed to improve victimization reporting in a 1980s redesign of the U.S. crime victimization survey. Redesigned screening questions were structured around multiple frames of reference (acts, locales, activities, weapons, and things stolen) and included numerous cues to stimulate recall, including recall for underreported, sensitive, and nonstereotypical crimes. The result was much higher rates of reporting.

Although cuing improves recall, it can also introduce error because it leads to an increase in reporting of ineligible incidents as well as eligible ones. In addition, the specific cues can influence the kinds of events that are reported. The crime survey redesign again is illustrative. Several crime screener formats were tested experimentally. The cues in different screeners emphasized different domains of experience, with one including more reminders of street crimes and another placing more emphasis on activities around the home. Although the screeners produced the same overall rates of victimization, there were large differences in the characteristics of crime incidents reported. More street crimes and many more incidents involving strangers as offenders were elicited by the first screener.

Dramatic cuing effects such as this may result from the effects of two kinds of retrieval interference. Part-set cuing occurs when specific cues interfere with recall of noncued items in the same category. For example, giving “knife” as a weapons cue would make respondents less likely to think of “poison” or “bomb” and (by inference) less likely to recall incidents in which these noncued items were used as weapons. The effect would be doubly biasing if (as is true in experimental studies of learning) retrieval in surveys is enhanced for cued items and depressed for noncued items.

A second type of interference is a retrieval block that occurs when cues remind respondents of details of events already mentioned rather than triggering recall of new events. Recalling one incident may block retrieval of others because a respondent in effect keeps recalling the same incident. Retrieval blocks imply underreporting of multiple incidents. Early cues influence which event is recalled first, and once an event is recalled, it inhibits recall for additional events. Therefore, screen questions or cues asked first may unduly influence the character of events reported in a survey.

Another illustration of cuing or example effects comes from the ancestry question in the U.S. census. “English” appeared first in the list of examples following the ancestry question in 1980 but was dropped in 1990. There was a corresponding decrease from 1980 to 1990 of approximately 17 million persons reporting English ancestry. There were also large increases in the numbers reporting German, Acadian/Cajun, or French-Canadian ancestry, apparently due to the listing of these ancestries as examples in 1990 but not 1980, or their greater prominence in the 1990 list. These effects of examples, and their order, may occur because respondents write in the first ancestry listed that applies to them. In a related question, examples did not have the same effect. Providing examples in the Hispanic origin item increased reporting of specific Hispanic origin groups, both of example groups and of groups not listed as examples, apparently because examples helped communicate the intent of the question.