Is the recent increase in national homicide abnormal? Testing the application of...

Introduction

Recent increases in homicides continually receive attention from media outlets, and have been a focus of a growing body of research (Davey & Smith, 2015; McDowall, 2019; Rosenfeld & Fox, 2019; Rosenfeld, Spivak, & Muhlhausen, 2017; Wheeler & Kovandzic, 2018). The New York Times sparked a debate on recent changes in crime when they reported that, in a sample consisting of 10 big cities, including Philadelphia, New York, Milwaukee, and St. Louis, homicides increased by 32.8% on average from 2014 to 2015. In a report on homicide trends commissioned by the National Institute of Justice, Rosenfeld et al. (2017) argued that national homicide rates in 2015 increased by a substantial amount, and such an uptick necessitates more attention. James Comey, the Federal Bureau of Investigations (FBI) director during the period, acknowledged the recent increases in violent crimes in major U.S. cities and suggested that the increase could be attributed to police officer's reluctance to engage in law enforcement activities (Schmidt & Apuzzo, 2015). Their presumption is that police officers desire to avoid controversy over their use of force against minorities and other members of society, thus reducing their enforcement of the law and allowing more violent individuals to remain unrestrained – and then commit future violent crime. Others have proposed that civilian perceptions of police legitimacy have decreased as a result of high profile incidents of police brutality, thus leading to individuals acting brazenly with disregard for the law. Research on these two takes on the “Ferguson Effect” attempts to tease out any negative consequences stemming from de-policing (reductions in police presence, traffic stops, stop and frisks) on violent crime. Both scholars and practitioners assumed the recent change in crime statistics was substantial and warranted more scrutiny for public safety.

The upward spike in homicides provides the backdrop to this paper's research question: do increases in homicide in recent years show abnormal patterns when compared to historical crime trends? Despite high levels of attention from members of the media, public, and scholarly society, academic research has just begun to examine the change in recent homicide trends (Lauritsen & Lentz, 2019; McDowall, 2019; Pyrooz, Decker, Wolfe, & Shjarback, 2016; Rosenfeld & Fox, 2019; Wheeler & Kovandzic, 2018). Such questions are not new, and past increases in homicides prompted similar questions of whether crime was waving (Barnett, Kleitman, & Larson, 1975).

One of the issues in monitoring crime trends is the use of appropriate methods to identify abnormal patterns in changes over time. Prior studies examining temporal crime patterns relied on two methods: descriptive analysis of percent changes over time, or a structural break test, also known as trend analysis (Cook & Cook, 2011; McDowall, 2014; McDowall & Loftin, 2005; Pyrooz et al., 2016; Rosenfeld, 2016; Rosenfeld, 2018; Rosenfeld et al., 2017). Although each technique has their own utility, they have a limited ability to identify outliers in recent crime patterns due to potential bias embedded in percent change or a lack of post-observations required for the trend analysis. The limitations of these two conventional methods prompted the motivation for this work, to develop more rigorous methods to examine the abnormality of crime rates over time.

Recent studies offered an alternative approach for monitoring crime trends; the use of forecasts and fan charts (Wheeler, 2016; Wheeler & Kovandzic, 2018). The procedure to identify outlying patterns in crime trends can be described as follows: (a) specify the statistical model reflecting the series of crime data, (b) use a forecasting method to estimate prediction intervals around crime during the period of time of interest, (c) determine whether observations of interest fall inside or outside the forecasted prediction intervals. Some researchers have employed this method and showed its utility in identifying abnormality in recent changes in crime patterns (Wheeler, 2016; Wheeler & Kovandzic, 2018). These studies focused on temporal crime patterns at the city level, with little attention paid to national crime patterns. Given the fact that researchers and law enforcement agencies found homicide increases across the United States (Davey & Smith, 2015; Pyrooz et al., 2016; Rosenfeld et al., 2017), examination of the potential abnormality in recent national homicide patterns is needed.

The present study can contribute to the literature on temporal patterns in homicide in two ways. First, this study employs a forecasting method and uses fan charts to graph national level homicide data to examine whether recent homicide increases are significant compared to prediction intervals generated from historical homicide patterns. By considering national level accounts, it is possible to better understand whether the recent increase in homicide is abnormal or a result of random fluctuation. Second, we will compare the recent change in homicide with those for robbery, rape, and aggravated assault. This allows us to assess the potential for abnormality in recent increase in homicide by assessing whether the change is pronounced only in the homicide series or if a significant change occurs across the other UCR part 1 violent crimes.