BJS responses to OMB questions

ICR 200801-1121-002, Police Public Contact Survey

1. Given the break in series for the main NCVS, what is the BJS plan for describing these data in historical context?

At this point, it is not clear how the recent changes to the NCVS will affect the PPCS. To determine whether the break in series for the main NCVS had an impact on the PPCS supplement, BJS plans to compare a number of national and subnational estimates of rates of police-public contact and rates of police use of force from the 2005 and the 2008 PPCS. Subnational estimates to be evaluated include whether the respondent lives in an urban, suburban, or rural area, the region of the country in which the respondent lives, and various demographic characteristics such as age, race, and sex.

Based on knowledge gained by evaluating the impact of the break in series for the NCVS, BJS will also compare 2005 and 2008 rates of police-public contact and police use of force for particular subgroups identified during the NCVS analysis as particularly vulnerable to survey changes. In addition, BJS will also compare 2005 and 2008 contact and force rates among persons who were victims of crime, as an additional layer of analysis to help determine the potential impact on the PPCS of the break in series for the main NCVS.

BJS will ensure that changes in NCVS methodology and issues regarding comparability with previous data are discussed in the final report. Also, BJS will include findings of the above-described analyses in the final report generated from the 2008 data collection.

2. Given the schedule for reporting to Congress, it appears that this survey would be better conducted during the first ½ of the year rather than the second. Why is it on the current schedule?

With the exception of the 1996 pilot study, all of the past PPCS data collections (1999, 2002, and 2005) have been conducted during the last 6 months of the corresponding calendar year. Results have generally been disseminated 12-18 months following the end of the data collection. For example, the 2005 data collection occurred from July 1 to December 31, 2005, and the final report was published in April 2007. The April 2007 report helped to fulfill the mandate for 2007 from the Violent Crime Control and Law Enforcement Act of 1994.

The current schedule for collecting data from the PPCS does not impact the ability of BJS to gather annual data on police use of force. BJS activities to compile information on that topic go beyond the PPCS. For instance, the annual publication “Homicide Trends in the United States” includes the latest national statistics on justifiable homicides by police. In addition, BJS has published findings from “Traffic Stop Data Collection Policies for State Police,” which includes information on the number of state police agencies that routinely keep administrative records on police use of force in traffic stops. BJS also released a report in 2006 on citizen complaints about police use of force, scheduled to be updated in 2009.

3. Given the break in series and the schedule, what quality assurance plans are in place to ensure that the very preliminary data reported to Congress are done so accurately?

BJS has no plans to publish preliminary data from the PPCS. The language in the supporting statement has been revised to accurately reflect the dissemination plan for PPCS data.

The data file goes through a rigorous quality control process both prior to and following the analysis phase. Over the years, BJS has developed an extensive number of data quality checks to examine the data for inconsistencies and errors in the file. These checks are performed prior to analysis of the data. In addition, the final report goes through a stringent verification process before the results are released to Congress and the public.

4. Given the anticipated response rate, please provide BJS’s nonresponse bias analysis plan.

The three basic ways a person can be a nonrespondent to the PPCS include:

• The entire sample household may not respond to NCVS.

• A person within an interviewed sample household may not respond to the NCVS. Since it is a requirement that a person completes their core NCVS interview first, any persons who do not respond to the NCVS are, by default, nonrespondents for the supplement.

• A person who responded to NCVS may not respond to the PPCS.

BJS plans to calculate three separate unit response rates: NCVS alone, PPCS alone, and a combined NCVS and PPCS response rate. The three response rates will measure the overall impact of nonresponse and identify how much each of the areas impact the overall response rate. The attachment entitled “Nonresponse Bias Analysis Plan for the 2008 Police Public Contact Survey” contains more detailed information on the current nonresponse bias analysis plan.

In addition, BJS is considering other alternatives and studying the issues surrounding nonresponse bias. To that end, BJS is funding research to more fully evaluate NCVS nonresponse and explore development of methods to better reduce or adjust for nonresponse. Results from these efforts will provide information on the impact of nonresponse on estimates from the PPCS.

5. Part B describes the bounding interview in a manner that appears outdated, given the new NCVS procedures for using those data as the first interview. Please update.

Until July 2007, estimates from the NCVS excluded the bounding interview. In an effort to reduce program costs, the NCVS sample was reduced by 14% in July 2007. At that time, to offset this reduction in sample and maintain an equivalent sample size for the 2007 NCVS estimates, BJS began including the bounding interview as part of the overall victimization estimates. Adjustment factors were introduced to mitigate any over-counting of crime due to the use of first time interviews. The language in the supporting statement will be revised to accurately reflect these new procedures.

For the PPCS, however, no weighting adjustment is necessary because past enumerations from the supplement have used the bounding interviews in producing estimates related to police contacts.

6. Is the policy to not accept proxy interviews for PPCS the same or different from NCVS?  If different, why?

Approximately 2% of NCVS interviews are conducted by proxy. The PPCS excludes these proxy interviews. Given the nature of the questions, BJS determined that proxy respondents would have difficulty describing the details of contacts between police and the sampled respondents. For instance, respondents are asked whether they felt the force was excessive; a proxy interview would not know how the sampled respondent felt about the use of force.

7. In reference to Item 3, Part B, how does the interviewer obtain characteristics of noninterview persons and households?

For the PPCS, demographic characteristics are obtained from the NCVS. Information on the characteristics of noninterviews is obtained in two different ways: 1) from sampling frame information and 2) from demographic information provided in previous interviews.

8. Since this survey is not self-administered, why is all of the PRA and confidentiality information written on the questionnaire in a manner that appears to be excluded from the interviewer script?  Please provide all materials, including advance letters or FAQs, used to provide PRA and confidentiality information to respondents of the PPCS.

Advance letters are sent to NCVS households and provide advance notice to sample households that a Census representative will be contacting them in the near future for the NCVS. These letters contain information on a number of frequently asked questions, including the confidential nature of the NCVS data collection and that the participation of respondents is purely voluntary. The advance letter content is identical across all regions except for the regional office contact information. Attached are examples of advance letters for the Atlanta region.

BJS has directed the Census Bureau to include a burden statement in the initial introduction to the PPCS. The revised questionnaire is attached reflecting the inclusion of the burden statement.