The Effect of Prior Police Contact on Victimization Reporting: Results From the Police-Public Contact and National Crime Victimization Surveys, United States, 2002-2011 (ICPSR 36370)

Time Period(s) View help for Time Period(s)

Date of Collection View help for Date of Collection

Study Purpose View help for Study Purpose

The study centered around five main objectives. First, link people interviewed in the 2002, 2008, and 2011 Police-Public Contact Survey, 2002 (PPCS) to their corresponding records in the National Crime Victimization Survey (NCVS). Second, Identify the extent and nature of prior police contact for victims versus non-victims. Third, assess whether there is a relationship between prior face-to-face contact with the police and subsequent reporting of victimizations, controlling for characteristics of the victimization, victim, and offender. Several facets of police contact are examined including prevalence, frequency, type (e.g., citizen-initiated, police-initiated routine traffic stop, and more invasive contacts in which the respondent was arrested, handcuffed, searched, frisked, or the suspect in a crime), and perceptions of whether the police behaved properly. Fourth, examine if the extent and nature of victims' vicarious contact with the police (i.e., household members' police experiences) are associated with crime reporting. Fifth, explore whether the effects of personal and vicarious contact on victimization reporting are moderated by victim characteristics (age, race, gender, and poverty).

While the NCVS surveys households on crime victimization, it does not include information on police-related experiences. The study compensated for this by linking PPCS to participants of the NCVS. The NCVS is administered yearly, with questions regarding respondent and household characteristics, along with victimization experiences from the past six months. The PPCS is conducted every three years as a supplement to the NCVS and designed to document face-to-face contacts between the public and the police. Respondents are asked about their contacts with the police in the past year, and detailed data are collected on the most recent encounter including what happened, the outcome, and whether the respondent thought the officer's actions were proper. In 2002, 76,910 individuals completed both the PPCS and NCVS. The number of respondents who took part in both interviews decreased to 57,978 individuals in 2008 and 49,246 in 2011.

Study Design View help for Study Design

A dataset was created by merging NCVS and the PPCS to create a longitudinal person-level file containing information on the past-year face-to-face police encounters and subsequent victimization experiences of 105,273 people ages 16 and older living in the United States. The following PPCS and NCVS studies used for analysis and their corresponding ICPSR study numbers are listed below:

• Police-Public Contact Survey, 2002 [United States] (ICPSR 4273)
• Police-Public Contact Survey, 2008 (ICPSR 32022)
• Police-Public Contact Survey, 2011 (ICPSR 34276)
• National Crime Victimization Survey, Concatenated File, 1992-2014 (ICPSR 36143)
• National Crime Victimization Survey: Unbounded Data, 2002 (ICPSR 4447)
• National Crime Victimization Survey: Unbounded Data, 2003 (ICPSR 4448)
• National Crime Victimization Survey: Unbounded Data, 2004 (ICPSR 4449)
• National Crime Victimization Survey: Unbounded Data, 2005 [Record-Type Files] (ICPSR 22341)

The NCVS and PPCS do not contain common identifiers that would allow cases to be linked across datasets. Variables that were common to NCVS and PPCS were concatenated to create unique identifiers which could link the two datasets. Some variables required little to no recoding to create identifiers, while others needed significant manipulation. This created unique identifiers for approximately 77 to 90 percent of PPCS cases, depending on the year. Unique identifiers for other cases were generated by manipulating the sampling weights and using the person sequence number. After matching, the PPCS files were merged with the NCVS person, household, and incident files containing data that were collected at the time of the PPCS interview.

Each person in the matched file was then linked longitudinally to his/her corresponding subsequent records in the NCVS. Approximately a quarter of the cases could not be linked longitudinally and are not included in the dataset. The majority of these cases involved households rotating out of the study. Other cases could not be matched because of household non-response, individual non-response, and mismatched demographic characteristics across files (i.e., cases with the same NCVS identification number that changed sex, race, or age groups from one wave to the next), something that has been documented in prior studies linking NCVS files over time. Finally, household data and incident-level data on the first personal and household victimization occurring subsequent to the PPCS interview were merged into the person-level dataset. In a few cases, multiple incidents of the same type occurred in the same month and were the first incidents subsequent to the PPCS, making it impossible to determine which incident was most proximal to PPCS interview. When this occurred, the first incident was selected at random. This affected 39 (.6%) of the personal incidents and 443 (3.9%) of the respondents with household incidents.

Time Method View help for Time Method

Universe View help for Universe

Individuals who completed the National Crime Victimization Survey and the Police-Public Contact Survey in 2002, 2008, and 2011.

Unit(s) of Observation View help for Unit(s) of Observation

Data Type(s) View help for Data Type(s)

Response Rates View help for Response Rates

Not applicable.

Presence of Common Scales View help for Presence of Common Scales

None.

Original Release Date View help for Original Release Date

2021-11-30

Weight View help for Weight

A weight variable was derived from the PPCS that can be used to generate national population estimates. Users should note that not all PPCS interviews could be matched, therefore population estimates will not necessarily be representative of the U.S. population. Also, 2011 cases have higher weights due to sampling design, and thus have a disproportionate effect on the results.