Implementation Plan for Modifying the National Crime Victimization Survey Crime Incident Report Questions about the Race of the Offender(s)
Currently, the NCVS asks victims of crime to identify the race(s) of the offender or the races of the offenders (when multiple offenders committed the crime.) The questions used to collect this information are set up as a single answer when the crime was committed by one offender and multiple entries when the crime was committed by more than one offender. Both questions have four answer categories: 1) White, 2) Black, 3) Other (Specify), and 4) Don’t know. The NCVS does not currently ask for the ethnicity of offenders. In reviewing the 2009 NCVS clearance for the survey, OMB determined that the race of offender questions do not conform to federal standards for collecting self-identified race and ethnicity information from the population. These standards specify the minimum number of categories for statistical reporting on race. OMB imposed a special condition on the NCVS clearance, requiring BJS to bring the race of offender questions into compliance with the standards for collecting self-identified race information.
In response to the special condition, in the fall of 2009, BJS and the Census Bureau began working on a plan to evaluate and modify the NCVS race of offender questions. At BJS’s request, the Census Bureau developed a research plan to study issues related to identification of observed race and ethnicity. The research plan included a literature review, and an experiment to assess respondents’ ability to accurately identify the race of other people using standard self-report questions. The research plan was provided to, and approved by OMB in October-November 2010.1 The research plan incorporated revising the race of offender questions, reordering the set of questions obtaining information on offender characteristics, and evaluating the questionnaire changes. The evaluation was to be accomplished through implementation in outgoing sample rotation groups who would be administered a debriefing questionnaire to determine whether they encountered any difficulties in answering the revised race and ethnicity of offender questions.
In February 2011 BJS decided to remove the experimental component of the implementation plan. The reasons for dropping the experiment include:
Compliance with OMB standards. BJS has been bringing all survey instruments into compliance with OMB standards for measuring race and ethnicity. With respect to the NCVS questions on race of offender, BJS could no longer justify a position of departing from OMB standards for the NCVS when BJS decided in its Police Public Contact Survey supplement (OMB No 1121-0260) to adopt OMB standards for race and ethnicity when asking respondents to describe the race of the police officers with whom they had contact. Dissemination of reports and data will note this reason for making the change in the race and ethnicity of offenders in the NCVS.
Most victims of violent crimes (the subset of victims asked the race of offender questions) either know their offender or know about their offender. Over 50% male victimizations and about 65% of female victimizations are committed by persons known to the victim.
In stranger perpetrated crime, victims are still able to utilize many non-visual cues that facilitate the assignment of a racial and ethnic origin. These include location of the crime, subsequent police follow-up reports and arrest, language, dress, statements made by the offender, and mannerisms.
The NCVS is not attempting to establish unique identities of offenders but their general characteristics. Eye witness research that indicates witness problems with identifying unique individuals points out that the problem is not one of identifying the race or ethnic group membership of an individual but between identifying specific individuals within a race or ethnic group.
Most crime is intra-racial and this facilitates victim assessments of race of offender.
Insufficient power to test differences between old and new question formats. The race of offender question is only administered to victims of crime who had contact with the offenders and these cases typically are limited to violent crimes. Relying on data from an outgoing rotation to conduct comparisons meant that the comparisons between the test and control groups (new and old question sets) would have been based on a small numbers of cases. Hence, in February 2011, BJS concluded that outgoing rotations would provide insufficient sample cases to have sufficient power to conduct a test in a timely manner.
The current plan calls for implementing the revised race and ethnicity questions in January 2012, and evaluating the data subsequent to implementation. The revised set of questions, which are included as an attachment (“Proposed Offender Characteristics Questions”) at the end of this memo, accomplish two goals. First, the race and ethnicity questions are worded to comply with the OMB Standard for these questions. Second, the revision reorganizes the questions related to offender characteristics to group the demographic questions together.
Taking these factors into consideration, the revised set of questions asks first about the offender(s)’ relationship to the victim, followed by questions about the offender(s)’ gender, age, ethnicity, and race; and finish with questions about gang involvement and drug or alcohol use. There are two modules; one for crimes committed by a lone offender and one for crimes committed by multiple offenders. As with the PPCS, since the questions ask respondents to identify the ethnicity and race of other people, the questions include “don’t know” categories that are not read to respondents to address instances in which the respondent is not able to identify the ethnicity or race of the offender(s).
In the NCVS and PPCS, whether it is the race of the offender or police officers, BJS has found that there is considerable difficulty for some respondents to recall exact racial and ethnic categories. As with the PPCS, asking race as a separate question with distinct categories preceded by a question specifically asking about Hispanic origin is preferable to asking respondents to report on race and ethnicity in a single question format. The two question format allows respondents to recall Hispanic origin and race as separate cognitive tasks and should provide higher quality data. Additionally, a two question format allows a respondent to provide a “don’t know” response (not read aloud) on the race or ethnicity question while still providing a valid response on the remaining question.
BJS had considered the inclusion of questions targeting stranger offenders in the debriefing portion of the previous implementation plan. The purpose of the proposed addition was to assess the extent of victims’ ability to determine race of the offender and identify any problems specifically related to identification of stranger perpetrated crimes. These additional targeted questions were dropped from the current plan. As part of the ongoing research to improve NCVS methodology, BJS has tasked a visiting fellow with examining how the current incident form may be reformulated to both streamline and add analytical value to the survey. A detailed examination of racial and ethnic identification of stranger crimes is more aligned with that research effort than as a component of the race of offender implementation and will be examined therein. Despite this change, all respondents, whether victimized by strangers or known persons, will be administered the new race and ethnicity questions.
Current Implementation Plan
The components of the current implementation plan are described below:
Programming the new questions will be accomplished in October, 2011. Once the instrument code has been updated staff will complete a series of tests to verify the instrument is working as expected in terms of skip patterns, valid values, and question text. The Census Bureau will test both situations where the revised questions should be asked and when they should not be asked to verify the questions are on path when appropriate.
Census will develop a self study for the FRs to complete prior to January 2012 interviewing. This self study will comprise of a review of the new questions, why the changes were made, as well as instructions for completing the questions with respondents. In addition, the FRs will have access to an updated instrument when completing the self study in order to help familiarize the interviewers with the new questions and skip patterns. As part of the expanded quality enhancement program being implemented for the NCVS, periodically the FR’s will be provided with practice interview modules to complete to ensure that they have experience with the full set of survey questions. These practice interviews will include administering the revised race of offender questions.
Evaluation
Evaluation of the revised questions will be accomplished in a number of ways. First, immediately following implementation of the revised question, BJS and Census Bureau staff will review incoming data from the field to ascertain whether issues in implementation are in evidence. Although testing of the questions will be completed by December 1st, unforeseen issues sometimes arise during full scale implementation. Any such issues will be addressed immediately, and additionally, local supervisory field representatives will be monitoring implementation in their areas to insure their FRs do not encounter problems.
Upon implementation monthly materials provided to the FR’s will include a set of questions asking the FR’s to identify any problems encountered in asking the questions. Continuous monitoring of the data in the revised questions will continue until BJS and the Census Bureau are confident any issues related to implementation have been identified and resolved.
Additionally, as part of the ongoing NCVS quality improvement program, the Census Bureau has developed a web-based tool for FR use in the field. The purpose of the tool is to obtain real-time feedback from the field on survey administration issues that the FRs experience. Although victim debriefing questions were dropped as part of the current implementation plan, the web-based tool will be utilized to provide Census headquarters with information on interviews in which victims struggle with the race and ethnicity of offender questions. This information will then be used to refine the questionnaire as needed during the evaluation process.
However, BJS does not anticipate that problems will arise during implementation. In 2003, NCVS question categories for race of the victim were brought into compliance with OMB Directive 15. The revised response categories were white, black, Asian, Native Hawaiian or other Pacific Islander, American Indian and Alaska Native, and two or more races. For reporting purposes BJS collapsed American Indian and Alaska Native, Asian and Native Hawaiian or other Pacific Islander, into an “other” category for annual victimization estimates as the survey had done in previous years. Because less than 1% of survey respondents identified two or more races, the impact on the victimization rates for each race category was small (Criminal Victimization, 2004. NCJ 210674).
BJS expects a similar effect when changes are made to the race of offender categories. In the months following implementation of the questions, the data from the questions will be evaluated as part of the NCVS program’s increased quality standards. This evaluation will include a comparison of data from the new and old questions and an examination of any anomalies. In the event that substantial shifts in the reporting of offender race categories is realized, BJS will explore additional analyses such as evaluating crime and victim characteristics in association with old and new race questions in an attempt to understand the origin and manner of the shifts in response categories and evaluate the effect of the change.
Analysis Plan
Beginning with data collected in January 2012, BJS will include analyses of offender ethnicity and race in reports and other data products that discuss offender characteristics. The change will be introduced to stakeholders as a modification that brings the NCVS into full compliance with OMB standards for the collection of data on race and ethnicity. Following the approach of the previous race of victim change, BJS will note the change in the race and ethnicity of the offender questions in reports released for two years following implementation. A fuller discussion of the change will be included in the technical report to be released by BJS on the change in offender race and ethnicity categories.
Similar to what has been done for other offender related variables; BJS will examine the distributions of data from the offender ethnicity and race items and develop analytic recodes for use in analyses. The level of detail for these variables will, as it is with all analytic variables for the NCVS, be dependent on the extent to which the data can support detailed presentation of individual characteristics. For some analyses, it may be necessary to aggregate categories to achieve sufficient numbers of cases to produce reliable estimates. Any aggregation of data will follow OMB guidelines, as have all previous data releases of aggregated data for victim racial categories. As in all BJS releases of NCVS racial data, these categories are—
White only,
Black only,
Asian only,
Native Hawaiian or other Pacific Islander only,
American Indian or Alaska Native only, and
Two or more races.
Annual estimates of victimization generally lack sufficient cases to present the six categories, and as a result, the Asian, Native Hawaiian or other Pacific Islander, American Indian or Alaska Native, and two or more races are aggregated into an ‘other’ category for purposes of presentation in reports. This type of aggregation conforms to Directive 15 while still allowing BJS to release information in a manner that utilizes data from all response categories. Otherwise, most if not all of the detailed categories would be flagged or suppressed. BJS publications not requiring annual estimates provide greater latitude for presenting estimates of race and ethnicity because multiple years of data can be grouped together to provide sufficient cases for presentation of more detailed response categories.
As mentioned previously, BJS will announce the change as one that was initiated to bring the NCVS into full compliance with federal standards on the reporting of race and ethnicity. BJS has made identical transitions for the NCVS race of victim categories and the PPCS categories for race of the officer.
Given the limited number of expected cases available for analysis, an extensive analysis of the data may not be possible prior to 2013. At a minimum, BJS requires one year of data to begin extensive analysis to allow for fluctuations in the data related to seasonality. However, BJS will begin exploratory analyses of the data immediately upon implementation in Quarter 1 of 2012 and continuing until such time that sufficient cases are obtained to support more detailed analyses. An interim data brief including a discussion of the change and preliminary exploratory analyses may be released during 2012 dependent on whether sufficient cases are present on which to base an analysis.
The current and proposed questions are also attached.
Finalization of the requirements for the implementing the new questions occurred in September 2011. These requirements included adjustments to question ordering and placement in the NCVS instrument. Modification of the instrument specifications in CAPI will begin in early October 2011 and conclude on December 22, 2011 when the Census Bureau releases the January 2012 instrument to field staff. During this time, Census Bureau staff from the Demographic Surveys Division and Technology Management Office will program and test the questions iteratively. As the requirements have been established, testing will focus primarily on insuring that the correct skip patterns and edit checks are operating correctly. During the same time period the training materials including the self-study will be developed and refined. Self-studies will be released to FRs by December 9, 2011, and FRs will have until December 29, 2011 to complete the self-study. Census does not anticipate issues with FRs not completing training prior the start of interviewing in January. The Census Bureau is currently ahead of schedule as compared to the timeline presented in the attachment.
1,2 Email from Rochelle W. Martinez to Michael Rand 11/10/2010
| File Type | application/vnd.openxmlformats-officedocument.wordprocessingml.document |
| Author | seama300 |
| File Modified | 0000-00-00 |
| File Created | 2021-01-31 |