Using Social Media to Measure Temporal Ambient Population: Does it Help Explain Local Crime Rates?

Abstract

A challenge for studies assessing routine activities theory is accounting for the spatial and temporal confluence of offenders and targets given that people move about during the daytime and nighttime. We propose exploiting social media (Twitter) data to construct estimates of the population at various locations at different times of day, and assess whether these estimates help predict the amount of crime during two-hour time periods over the course of the day. We address these questions using crime data for 97,428 blocks in the Southern California region, along with geocoded information on tweets in the region over an eight month period. The results show that this measure of the temporal ambient population helps explain the level of crime in blocks during particular time periods. The use of social media data appear promising for testing various implications of routine activities and crime pattern theories, given their explicit spatial and temporal nature.

Keywords:

• spatial crime
• routine activities
• social media
• ambient population

Notes

¹ The region is defined as including five counties: San Bernardino, Riverside, Los Angeles, Orange and San Diego.

² https://dev.Twitter.com/streaming/

³ Given that only the percent single-parent households variable is available for blocks, we use synthetic estimation for ecological inference to impute the other variables (Cohen & Zhang, 1988; Steinberg, 1979). The imputation models use the following variables: racial composition, percent divorced households, percent households with children, percent owners, percent vacant units, population density, and age structure (percent aged: 0–4, 5–14, 20–24, 25–29, 30–44, 45–64, 65 and up, with percent 15–19 the reference category).

⁴ The land use data was obtained from the Southern California Association of Governments (SCAG).

⁵ Given that the independent variable is log transformed, a .10 change in this logged variable represents an approximate 10% change in the number of tweets.

⁶ http://www.pewinternet.org/2015/08/19/the-demographics-of-social-media-users/.

Additional information

Notes on contributors

John R. Hipp

John R. Hipp is a Professor in the Departments of Criminology, Law and Society, and Sociology, at the University of California Irvine. His research interests focus on how neighborhoods change over time, how that change both affects and is affected by neighborhood crime, and the role networks and institutions play in that change. He approaches these questions using quantitative methods as well as social network analysis.

Christopher Bates

Christopher Bates is a PhD student in the Department of Criminology, Law and Society, at the University of California, Irvine.

Moshe Lichman

Padhraic Smyth is a Professor in the Department of Computer Science, with a joint appointment in the Department of Statistics, at the University of California, Irvine. He is also the director since 2014 of the UCI Data Science Initiative. His research interests include machine learning, data mining, pattern recognition, and applied statistics and he has published over 160 papers on these topics.

Padhraic Smyth

Moshe Lichman is a PhD student in the Department of Computer Science at the University of California, Irvine.