Simple indicators of crime by time of day
Introduction
Crime varies more by hour of day than by any other predictor we know. Such variation is analyzed all too seldom. Perhaps one reason for this neglect is that hourly data produce too many categories, 168 h per week. The result is too few cases per cell (this loss of degrees of freedom impairs statistical analysis) and too many cells (this leads to very large tables that are hard to understand).
This paper provides some simple indicators that help gain a solution to these problems. However, a larger problem needs additional work—how to think about hourly variations in crime.
Section snippets
Background
Hägerstrand (1973) showed how the individual traverses a path through space–time in the course of a day. The importance of these movements was explained in social psychological terms by Bandura (1985), who coined the term, “the psychology of chance encounters.” Bandura described the intersection of individual paths in the course of a day and how these chance intersections can change individual lives and even history.
However, human ecology teaches us that many encounters are not so random as one
The first task
What is the first hour of the day? From the clock viewpoint, one starts with 12:00 to 12:59 AM, but that would ignore what we know about crime. At that hour, many people are not yet straggling out of urban bars, and parties are for some at a high swing. In many places, alcoholic beverage consumption accelerates after midnight in anticipation of closing hours, and food consumption may well decline. A majority of those driving cars or taking public transit in the early hours of the morning might
The second task
In order to study hourly crime patterns, a criminologist must decide what offenses to summarize and for what broader time period. For example, one might wish to describe the hourly patterns for all armed robberies in the city of Houston from 1990 to 1999, or one might wish to compare New York City's hourly aggravated assault patterns for September versus October of 2001.
The second task is more complicated than meets the eye. Many offenses are not readily reported to the police, so their hourly
The median minute of crime
Having selected 5:00 AM as the first moment of the day, we can now devise several simple indicators for hourly patterns of crime. The first is the median minute of crime, namely, that minute of the day by which exactly half of the crimes have occurred. For example, if the median minute of robbery is 6:13 PM, that means that exactly half the daily robberies occur from 5:00 AM to 6:13 PM, and the rest from 6:13 PM to 4:59 AM the next morning. This simple measure of central tendency tells us a
Crime quartiles
Measures of central tendency of course miss the dispersion over the hours of the day. Of course, one could calculate a standard deviation about the mean mentioned above. We think that quartiles offer a simpler and more cogent way to study hourly dispersion of crime and are more appropriate to the problem at hand. We suggest that the most direct and clearest way to study that dispersion is to find the quartile minutes. After the median minute of crime is known, the first half of the crime day is
Crime's daily timespan
Once we know the quartile minutes, it is elementary to calculate crime's daily timespan. This is the number of minutes between the first and third quartile minute. Where crime is more dispersed over the day, the daily timespan is higher. A narrow daily timespan will be expected for smaller cities with less extended nightlife. The median minute of crime and the daily timespan together tell us a lot of information, even though they are but two numbers. High school students appear to have an early
The 5-to-5 share of offenses
We have presented so far four summary indicators of how crime distributes over the course of a day. To take a different tack, we now ask what share of offenses have occurred by a particular time. We pick 5:00 PM as a cutoff time, since that vaguely tells us when evening begins. What percent of offenses occur by that time? We call this the 5-to-5 share of offenses. As evening and nighttime crime take over, this indicator will decline. Technically speaking, this number represents the percent of
Demonstration
The police departments of 13 middle-sized American cities have provided us with robbery data for the years 1999–2001 or parts of those periods. These cities include Akron, OH; Albany, NY; Cincinnati, OH; Evansville, IN; Fort Wayne, IN; Hartford, CT; Lincoln, NE; and Lowell, MA; Plano, TX; Rockford, IL; South Bend, IN; Springfield, IL; and Tampa, FL. The 2000 Census indicates that the largest of these cities is Cincinnati, with a population of 331,285. The smallest is Albany, with 95,658
General implications
We believe that these descriptive indicators serve as useful tools for describing hourly patterns of crime and making comparisons. We only considered one crime and a limited range of mid-sized cities, but we believe that these indicators can in the future assist researchers in describing and predicting how crime distributes over time.
Implications for forecasting
Some years ago, the senior author discovered that forecasting crime from 1963 to 1975 depended on studying trends in crime settings and crime timing (Cohen & Felson, 1979). The dispersion of activities away from family and household settings produced a major crime wave. That paper pointed towards a forecasting strategy that emphasized time patterns of activity in spirit, but lacked the data to carry out such forecasting directly. In recent decades, substantially more data on crime by hour of
Biographies: Marcus FELSON is author of Crime and Everyday Life, (Sage Publications), now in its third edition, and has developed the “routine activity approach” to crime analysis. He is also co-author (with Ronald V. Clarke) of Opportunity Makes The Thief, published by the British Home Office. Professor Felson graduated from University of Chicago and received his graduate degrees from the University of Michigan. He is Professor of Criminal Justice at Rutgers University.
References (9)
The psychology of chance encounters and life patterns
American Psychologist
(1985)- et al.
Nodes, paths and edges: Considerations on the complexity of crime and the physical environment
Journal of Environmental Psychology
(1993) - et al.
Social change and crime rate trends: A routine activity approach
American Sociological Review
(1979) Crime and everyday life
(2002)
Cited by (132)
Comparing crowding perception and sensor counts at the Ghent festivities
2024, Safety ScienceHot spots and burning times: A spatiotemporal analysis of calls for service to establish police demand
2022, Applied GeographyCitation Excerpt :Last, the 5-to-5 share of offences analyses the proportion of CFS occurring between 5:00 a.m. and 4:59 p.m. This is used to determine the time pattern throughout the day; a lower proportion of CFS indicates a later incident pattern (Ashby & Bowers, 2013; Felson & Poulsen, 2003). To analyse potential variability in the time patterns of the days of the week, we extend Felson and Poulsen's (2003) approach by calculating the parameters for each day separately.
Face Recognition Under Adverse Viewing Conditions: Implications for Eyewitness Testimony
2023, Policy Insights from the Behavioral and Brain Sciences
Biographies: Marcus FELSON is author of Crime and Everyday Life, (Sage Publications), now in its third edition, and has developed the “routine activity approach” to crime analysis. He is also co-author (with Ronald V. Clarke) of Opportunity Makes The Thief, published by the British Home Office. Professor Felson graduated from University of Chicago and received his graduate degrees from the University of Michigan. He is Professor of Criminal Justice at Rutgers University.
Erika POULSEN is a PhD candidate in the Geography Department at Rutgers University, and is the research director for the Crime Mapping Research Lab in the School of Criminal Justice, Rutgers University. Her research involves applying geographic techniques and methodologies for the spatial analysis of crime.