Dissertations, Theses, and Capstone Projects

Date of Degree

9-2024

Document Type

Dissertation

Degree Name

Ph.D.

Program

Criminal Justice

Advisor

Kevin T. Wolff

Committee Members

Amy Adamczyk

Gohar Petrossian

Eric Piza

Subject Categories

Criminology

Keywords

motor vehicle theft; Google Trends; crime statistics; victimization; UCR; NIBRS

Abstract

This dissertation examines the linear relationships between motor vehicle theft (MVT) estimates from Google Trends--a form of digital trace data--and official crime statistics, including the Uniform Crime Reports (UCR), National Crime Victimization Survey (NCVS), National Insurance Crime Bureau (NICB), National Incident-Based Reporting System (NIBRS), and 911 calls for service (CFS 911). It offers a comprehensive analysis of the limitations inherent in current crime statistics, discusses measurement errors, and explores statistical models to mitigate the impact of measurement error, applying these methods to validate Google Trends MVT data and other official MVT statistics. Furthermore, this study incorporates social disorganization theory and routine activity theory to test the concurrent validity of Google Trends MVT estimates across three different levels--state and year, Designated Market Area (DMA) and year, and DMA and month. The methodology employed includes generalized least squares (GLS), fixed effect models, natural experiment and interrupted time series, and instrumental variables, aiming to enhance the understanding of how user-generated digital trace data and its ability to make crime estimates. The findings confirm a linear relationship between MVT estimates from Google Trends and other official crime statistics, except for the NCVS. Moreover, MVT estimates from Google Trends show strong concurrent validity, mirroring trends and reacting similarly to crime covariates as other official statistics, except for temperature and precipitation. The conclusion discusses the further applications of Google Trends data, how population size influences crime data estimates, and the potentials and limitations of digital trace data such as Google Trends. To the best of the author's knowledge, this dissertation is the first to apply multiple-level crime statistics derived from Google Trends using robust and rigorous methods (multiple samples from GT) and to discuss in-depth GT's relationship with crime statistics and its applications in scientific studies. This dissertation not only examines the reliability of digital trace data but also explores its ability to illuminate the dark figure of crime.

Download

This work is embargoed and will be available for download on Wednesday, September 30, 2026

Included in

Criminology Commons

Share

COinS