ECP Memory Aid

Enhanced Contextual Priming as a Memory Aid

Obtaining accurate data on crime rates depends on complete reports from respondents on crimes they may have experienced. In the NCVS, respondents are asked a series of questions in the NCVS-1 crime screener to determine a count of crimes that may be in the scope of the survey. Each incident is then detailed in an incident report, which captures the information necessary to classify the type of crime that occurred.

Many factors can contribute to error in measurement of crime. For example, respondents may be reluctant to report a crime due to its sensitivity (for a review on reporting on sensitive topics, see Tourangeau, Rips, & Rasinski, Chapter 9). Further, they may not realize that a particular incident is a crime that should be reported (Lynch, 2006). A major factor that may contribute to error is forgetting that a crime occurred.

The encoding of an event may include information such as time and place of occurrence, emotional content, and images of the event (Tourangeau, Rips and Rasinski, 2000, Chapter 3). Cues may help respondents understand what the question is asking and determine the type of responses that are relevant. They may also help to directly access the memories that are to be reported in response to the question (Dashen and Fricker, 2001; Tulving and Pearlstone, 1966; Hudson and Davis 1972; Bellezza and Hartwell, 1981; Smith, 1990). Retrieval cues such as those incorporated in the NCVS-1 crime screener that refer to elements of the event may help direct the search of memory. Taylor and Rand (1995) found that the redesigned NCVS screener, which included many short cues to different kinds of crime, increased reports of some crimes.

In addition, prior questions within a survey can provide context that influences response to later questions (e.g., Tourangeau and Rasinski, 1988; Schuman & Presser, 1981).) Providing context through the use of cues and attitude questions can improve recall performance not only by priming the information to be recalled, but also by providing respondents with additional time for processing. Time on task is highly related to improved reporting (Sudman et al., 1996). Indeed, simply asking longer questions can give people more time to recall events, producing better data (Cannell, Oksenberg, & Converse, 1977; Bradburn et al., 1979). With the NCS, it was found that including attitude questions prior to the crime screening questions increased reporting of crime (Cowan, Murphy, and Wiener, 1978). Between 1972 and 1975, the NCS included supplemental attitude questions, which were administered to half the sample. The attitude questions appeared prior to the crime screening questions. The rationale for having the attitude supplement precede the crime screening was to avoid any influence of the victimization questions on attitudes to crime. In an analysis of the 1974 data, Cowan et al. (1978) compared crime rates for the half sample that received the attitude supplement to the half sample that did not. The major findings were:

• Personal crimes of violence: Robbery and rape showed no effects based on a t-test, but robbery showed an effect with a sign test. That is, reports of these crimes were not increased by the attitude supplement. The sign test suggests that the attitude supplement had a bigger effect on simple than on aggravated assault. The attitude supplement resulted in more reported crimes by strangers and more crimes by nonstrangers as compared to the no-supplement condition. There was a bigger effect of the supplement on crime reports for lower as opposed to higher age ranges; for younger respondents, the supplement yielded more reports as compared to no supplement.

• Personal crimes of theft: Higher reports of theft were found when the supplement was used. In the detailed crime categories, only personal larceny without contact showed higher reporting with the supplement; other detailed crimes did not show a difference by supplement condition. The supplement condition had slightly higher rates of reports to police than the no-supplement condition. Whites given the supplement reported more crime than whites not given the supplement—an effect not seen in black respondents.

• Household crimes: Reports of household crime are higher for households given the supplement. In detailed crime, larceny—completed and attempted—is the only significant category. Both white and blacks show significantly higher reporting with the supplement as compared to without.

The findings reported by Cowan et al. (1978) suggest that the attitude supplement could be increasing reports of crime in a number of ways. First, the supplement acts generally to prime thoughts about crime and activate memories of crime related events before recall of events is asked, thus increasing the accessibility of memories of particular incidents. It also provides cues for types of crimes, such as relatively less serious crimes that are more easily forgotten. The supplement may also clarify for respondents what counts as a crime and what should be reported.

The ECP reflects the role of context in influencing survey response. The ECP asks respondents about their feelings of safety at and near their home, the places they go, their trips away from home, their feelings of safety at the places they go, and their trust in others. These questions are intended to prompt respondents to think about crime-related topics, before answering the NCVS crime screener questions. The ECP takes a standard survey approach to providing context to respondents; that is, respondents are presented with a list of scripted interview questions. The ECP was cognitively tested; findings from the cognitive interviews are presented in the Appendix.