Skip to Main Content
Frequently Asked Questions
Submit an ETD
Global Search Box
Need Help?
Keyword Search
Participating Institutions
Advanced Search
School Logo
Files
File List
Petraglia dissertation - final document for uploading.pdf (3.02 MB)
ETD Abstract Container
Abstract Header
Estimating County-Level Aggravated Assault Rates by Combining Data from the National Crime Victimization Survey (NCVS) and the National Incident-Based Reporting System (NIBRS)
Author Info
Petraglia, Elizabeth Ellen
ORCID® Identifier
http://orcid.org/0000-0002-9042-4565
Permalink:
http://rave.ohiolink.edu/etdc/view?acc_num=osu1439027433
Abstract Details
Year and Degree
2015, Doctor of Philosophy, Ohio State University, Statistics.
Abstract
Crime rates for small geographic areas, or domains, are often of interest in research applications. However, survey data on victimization is often not reliable at the local level; the main source of crime data in the United States, the National Crime Victimization Survey (NCVS) is designed to produce national estimates only and the public-use data does not contain county identifiers. Even with county identifiers, crime is such a rare event that most sampled counties contain only a handful of victimizations at best. Crime data collected from police reports in the United States, such as in the Uniform Crime Reports (UCR) and the National Incident-Based Reporting System (NIBRS), is widely available at the county level but excludes crimes not reported to the police. The UCR and NIBRS are meant to be censuses of all police-reported crime, but are plagued by missing and incomplete data that makes estimation challenging. This dissertation presents two methods for estimating county-level crime rates that account for both crimes reported and not reported to the police. In both methods, we take police-reported crime data from NIBRS to calculate county-level crime rates including only crimes reported to the police, then use NCVS data to estimate the percentage of crimes reported to the police in each county of interest. The difference between the two methods is the mechanism for matching NCVS records to counties, since there is no natural way of linking the two. In the first method, a resampling technique is used on NCVS records. We use American Community Survey (ACS) five-year estimates to determine the population of each county in sex by race categories, then sample NCVS records according to these proportions. We then use the sampled records to estimate the county-level police reporting rate. A negative binomial distribution can then be used to model the true number of crimes committed in a county, taking the estimated police reporting rate as the probability of success and the number of police-reported crimes as the number of observed successes. The second method is an adaptation of a hierarchical Bayes model with county-level priors based on the demographic profile of each county to estimate county-level police reporting rates, updated with estimates based on the NCVS geographic identifiers available for each county. The method again uses a negative binomial to model the distribution of the number of all crimes committed in a county. Both methods are illustrated using the crime of aggravated assault, but can be extended to other crime types. These methods can be further extended to other scenarios when small area estimates are desired, if high-quality survey data cannot be used alone because of its limited coverage or sparsity and a large administrative dataset is available but may be biased or covers only a portion of the population of interest.
Committee
Elizabeth Stasny (Advisor)
Catherine Calder (Committee Member)
Eloise Kaizar (Committee Member)
Asuman Turkmen (Committee Member)
Pages
228 p.
Subject Headings
Criminology
;
Statistics
Keywords
National Crime Victimization Survey
;
National Incident-Based Reporting System
;
NCVS
;
NIBRS
;
small-area estimation
;
crime rates
;
simulation
;
administrative data
;
Uniform Crime Reports
;
UCR
Recommended Citations
Refworks
EndNote
RIS
Mendeley
Citations
Petraglia, E. E. (2015).
Estimating County-Level Aggravated Assault Rates by Combining Data from the National Crime Victimization Survey (NCVS) and the National Incident-Based Reporting System (NIBRS)
[Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1439027433
APA Style (7th edition)
Petraglia, Elizabeth.
Estimating County-Level Aggravated Assault Rates by Combining Data from the National Crime Victimization Survey (NCVS) and the National Incident-Based Reporting System (NIBRS).
2015. Ohio State University, Doctoral dissertation.
OhioLINK Electronic Theses and Dissertations Center
, http://rave.ohiolink.edu/etdc/view?acc_num=osu1439027433.
MLA Style (8th edition)
Petraglia, Elizabeth. "Estimating County-Level Aggravated Assault Rates by Combining Data from the National Crime Victimization Survey (NCVS) and the National Incident-Based Reporting System (NIBRS)." Doctoral dissertation, Ohio State University, 2015. http://rave.ohiolink.edu/etdc/view?acc_num=osu1439027433
Chicago Manual of Style (17th edition)
Abstract Footer
Document number:
osu1439027433
Download Count:
1,424
Copyright Info
© 2015, some rights reserved.
Estimating County-Level Aggravated Assault Rates by Combining Data from the National Crime Victimization Survey (NCVS) and the National Incident-Based Reporting System (NIBRS) by Elizabeth Ellen Petraglia is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by The Ohio State University and OhioLINK.