Guidelines for Selecting Victimization Questionnaires
Do I want to find out about a particular form of victimization or about many forms?
Anyone wanting to measure children’s victimization should consider a number of basic questions.
Do I want to find out about a particular form of victimization or about many forms?
Several questionnaires focus on a single form of victimization. An example of a focused questionnaire would be the Parent-Child version of the Conflict Tactics Scales (Straus et al., 1998), which exclusively measures child maltreatment. Other questionnaires ask about more than one form of violence. For example, the Survey of Children’s Exposure to Violence (Richters and Martinez, 1993) includes items on community (as opposed to family) assault, property crimes, and witnessing violence. The advantage of focused scales is that they cover a topic in depth and often identify more cases of that particular form of victimization than would a more wide-ranging scale. The advantage of multivictimization scales is that they provide data on the incidence and co-occurrence of several forms of victimization. Since many kinds of victimization co-occur (for example, family violence and community violence), unless one gets information on both, one can mistakenly conclude that a child’s distress or problem behavior stems from one kind of victimization, when another or both are behind the problem.
Do I need results that correspond to official crime or child maltreatment categories?
Some studies need only general measures of victimization. For example, a study correlating level of victimization with level of posttraumatic stress symptoms or depression could probably use many sound questionnaires. A study that is trying to show pretest to posttest improvement after participation in a therapy, social services, or prevention program may find useful any questionnaire that asks about the type(s) of victimization targeted by the program. Other projects, however, may require questionnaires that define certain victimizations, such as aggravated assault or physical abuse, in the same way police or child protective services (CPS) would. For example, a study that is trying to identify the number of abused and neglected children in a community and compare that number with the number who have been reported to CPS may need the community measure to match child maltreatment categories of abuse and neglect quite closely. Similarly, a study of crime underreporting would probably require a measure that would easily allow the researchers to label unreported crimes according to conventional categories of theft, robbery, assault, and so on. Researchers who must present their results to legislators or other policymakers may find it helpful if questionnaires can be described using official categories that are more likely to be familiar to professionals in fields other than children’s services.
Few existing measures provide results that closely correspond to crime or CPS categories. The National Crime Victimization Survey (NCVS) is one measure that corresponds to crime categories, but it has not been used with children younger than 12 and does not measure CPS categories of abuse and neglect. NCVS also has more complicated followup questions than most questionnaires. The Parent-Child version of the Conflict Tactics Scales (Straus et al., 1998) has sections for each of the major forms of abuse and neglect, but the threshold for abuse is not clear on many scales. For example, the physical violence scale of the Parent-Child version of the Conflict Tactics Scales includes spanking and other forms of legal physical discipline. Some new questionnaires are in development that will provide closer mapping onto CPS categories (Runyan, personal communication, 1999; Walsh and MacMillan, 1999). One new questionnaire, the Juvenile Victimization Questionnaire (JVQ), categorizes victimization by both conventional crime and CPS categories (Hamby and Finkelhor, 1999).
How often the terms “crime” or “abuse” are mentioned also affects what respondents will report, because questionnaire wording suggests or implies the interviewer’s apparent interests. In other words, the language used in a questionnaire helps to create a context that may influence the respondent’s answers. Questionnaires about “crime” tend to elicit reports that fit stereotypical perceptions of what constitutes criminal activity. Thus, disclosures are more likely to involve nonsexual assaults, stranger perpetrators, and incidents that were reported to the police (Hamby and Finkelhor, 2000). Questionnaires that do not focus exclusively on crime obtain generally higher rates of victimization, including many kinds of serious episodes that victims do not consider as crime. Very broad contexts, however, may result in the reporting of trivial events as crimes.
Context is created by a number of questionnaire features, including not only the wording of the victimization questions but also the preamble to the victimization survey, the sequence of questions within the victimization questionnaire, the other questions that are asked (for example, whether the questions are about family or home security), and the placement of the victimization questions in relation to other items and questionnaires. Although there is no standard prescription for the best context for victimization research, all of these issues should be considered in the design of any study.
Am I interested in obtaining rates on recent violence (e.g., last year), or do I want rates on lifetime exposure?
An incident rate provides an estimate of the amount of victimization youth have experienced in a given period, such as within the last year or since the introduction of an intervention program. A lifetime prevalence rate provides an estimate of the youth who have ever experienced the type of victimization under study. There are advantages and disadvantages to each approach.
Incident rates. One advantage to incident rates is that there is less reliance on the respondent’s long-term memory. Studies have shown that people tend to underreport victimizations that happened longer ago (U.S. Department of Justice, 1974). Although they may recall the incident if reminded of it, most people do not go around with a tally of all the victimizations they have experienced and thus may not have a precise count available when they come upon a question in a questionnaire. The longer the time period asked about, the more inaccurate the reporting of the total frequency of incidents and the details about specific incidents is likely to be. Another advantage of incident rates is that the same time period can be used for all respondents. For children, even a seemingly small age difference can lead to large differences in the chances of being exposed to victimization. For example, an 8-year-old has a 33 percent greater period of exposure than a 6-year-old. Thus, it would be impractical to compare first graders and third graders using lifetime exposure to violence, because one would assume the lifetime rate for third graders is higher anyway.
Lifetime exposure rates. The main advantage of a lifetime exposure rate is that more victimization experiences will be identified. This can be important if your sample of respondents is small or the type of victimization under study is rare. The rarity of victimization changes with context. For example, in the United States, exposure to war or terrorist violence is rare, but in other parts of the world it is common. Many inner-city communities experience more violence than most rural communities. Researchers studying a school program or interviewing children in a community setting will probably find much lower victimization rates than those evaluating a group of children from a mental health clinic or a juvenile detention center. Ideally, if the sample of respondents is at least of moderate size (200 or more), at least 10 percent of the sample should report the type of victimization being evaluated in order to compare victimized and nonvictimized groups statistically. If the rate of victimization falls below 10 percent, then more respondents will be needed so that at least 20 or 30 respondents are in the victimized group. Otherwise, any findings would be generalizations from the experiences of a very small number of individuals.
Do I want to use the questionnaire as an interview or in a self-administered format?
Interviews. Interviews, including those conducted over the telephone, pose fewer problems with comprehension, as long as complex or technical vocabulary is not used (for example, terms like “aggravated assault”). They also have the advantage of allowing more specific followup to a respondent’s answers, because an interviewer can be trained to ask some questions only if others have been answered in a certain way. Interviewers are also able to attend to children while they respond to the questions and can address issues of misunderstanding or discomfort. The disadvantages of interviews are that they can be extremely labor intensive and some children may feel less comfortable disclosing a victimization to an actual person rather than doing so privately on a questionnaire. A recently developed interview-administered questionnaire with good attention to reliability and validity is My Exposure to Violence (Selner-O’Hagan et al., 1998).
Self-administered questionnaires. Self-administered questionnaires (SAQ’s) have the advantage of requiring less labor from researchers and clinicians and can even be administered in group settings (e.g., school). Because respondents are expected to complete them on their own, however, SAQ’s must follow the simplest format possible in order to be easy to understand. Ideally, the reading level required for SAQ’s will be low. At a minimum, questionnaires should require less than an eighth-grade reading level for adolescents and parents and less than a fifth-grade reading level for middle school students. The ability of elementary school students to complete SAQ’s on victimization has not been adequately tested, although sometimes questionnaires are read aloud to groups of younger students, who then fill in their own answers. Self-administered questionnaires are also limited in the amount of information they can obtain on specific incidents because followup questions, which often need to be tailored to an individual’s responses, can easily lead to a dizzyingly complex pattern of skipping among questions. For example, the seriousness of a property victimization needs to be measured by a followup question about the value of the loss, whereas the seriousness of an assault needs to be followed up with a question about the extent of any injuries suffered. One example of an SAQ measure of victimization is the Children’s Report of Exposure to Violence, which has been shown to have good reliability and validity in samples of 9- to 18-year-olds (Cooley, Turner, and Beidel, 1995; Cooley-Quille, 1998).
Audio computer-assisted self-interviews. The newest technology used to administer questionnaires is called audio computer-assisted self-interview (audio CASI). With audio CASI, a child hears the question and the possible answers through a headset and touches a computer screen to respond. Studies have shown that, at least for adolescents and adults, many respondents disclose more sensitive information using an audio-CASI format (O’Reilly et al., 1994; Percy and Mayhew, 1997). This appears to be due to the increased privacy of the audio-CASI setting. Turner and colleagues (1998) conducted one of the most thorough tests of the new methodology with adolescents to date. They compared audio-CASI responses to those obtained using a pencil-and-paper self-administered questionnaire. They found increased reporting for a variety of sensitive behaviors, including male-male sexual behavior, substance use, violent perpetration, and violent victimization. Although the results for this method are extremely promising, audio-CASI equipment is very expensive and is typically used only when face-to-face contact with individual respondents is possible (because an interviewer must set up the equipment and show the child how it works). Nonetheless, it is likely that this method will be increasingly popular in the future.
What ages of children do I want to survey?
Adolescents. Adolescents, especially those in high school, will have language and comprehension abilities that are similar to those of adults. They are also much more likely than younger children to have encountered classes in sex education and to have been exposed to media that address sensitive victimization issues. Thus, questionnaires for high school students are often similar to (or even the same as) those used with adults. Self-administered questionnaires for adolescents should not require a reading level any higher than eighth grade (lower is preferable), but this is less of a concern for orally administered questionnaires.
Younger children. Younger children require much more developmentally sensitive questions. The vocabulary needs to be simpler, and many more terms need to be reviewed for their age-appropriateness. For example, many experts suggest using “private parts” for questions about sexual assault rather than more medical or legal terms (Everson and Boat, 1994). The Things I Have Seen and Heard questionnaire is one that has been used with children in first grade (Richters and Martinez, 1993). The Violence Exposure Scale for Children (VEX), which circumvents language issues by presenting cartoon images of various victimizations, has been used with preschoolers (Fox and Leavitt, 1995). Both have simplified language, omit some forms of victimization that may not be understood by young children, and use pictorial response categories to make responding easier.
Do I want to be able to compare my results with national norms or other
Questionnaires that supply national or community norms or have been used in other studies have several advantages. National or community norms allow researchers to compare the results from the group of respondents under study with those of a more general sample of youth. Thus, researchers can determine if the respondents in the sample have experienced more or less victimization than others. This can be helpful in terms of presenting your results to policymakers and funding agencies. Some scales also offer normative data for specific populations, such as CPS clients, therapy clients, or juvenile offenders. Some examples of questionnaires with national norms are the National Crime Victimization Survey (Kindermann, Lynch, and Cantor, 1997), the Monitoring the Future victimization questions (Wells and Rankin, 1995; Elliott, Huizinga, and Ageton, 1985), and the Parent-Child version of the Conflict Tactics Scales (Straus et al., 1998). Obtaining national norms is very expensive, however, and many well-constructed scales exist that do not offer normative data. Furthermore, because of the expense, many questionnaires with national or community norms are briefer and more general screening instruments than questionnaires that do not provide norms. Previously developed questionnaires, especially those in wide usage, offer the advantage of being able to compare new findings with earlier results. This is often true even if national norms are not available. In any situation in which longitudinal data are being collected, it is important to use the same questionnaire (or, at a minimum, very similar questionnaires) during each assessment.