Temperature, Light, and Crime Risk: A Seasonal Approach

Crime seasonality is fundamentally the observation that the occurrence of crimes (and therefore crime risk) is not static throughout the year. Observations and statistics to support this have existed since the 1800s — most crime types increase in frequency during the summer months and recede in the winter.

The finer mechanics of what produces these changes have been a source of lively debate. Temperature Aggression Theory states that hotter temperatures produce greater aggression and, therefore, fuel crime. Routine Activities Theory suggests that the summer months’ higher degree of public life: bustling downtowns, street festivals, and longer days, all provide motivated offenders with a greater number of targets and criminal opportunities, and this abundance of activities strains guardianship resources.

Certain research on crime seasonality has also observed that some property crimes increase in the winter months. However, this trend is not as statistically consistent as the spikes across most crime types observed in the summer months. Examining the Pinkerton Crime Index’s national aggregate for property crime seasonality, based on five years of data, we see a clear concentration in the summer months, with a slight bump in November, while January through April show significant reductions from “average” crime levels."

Comparing Crime Patterns in Diverse U.S. Counties

Zooming in on specific counties and cities, the story complicates further. The counties encompassing many major American metro areas feature notable spikes in property crime observed in October and November. We have paired the Pinkerton Crime Index’s measurement of property crime seasonality with average daily temperatures for each month to observe the severity of seasonal change and the impact that might have on property crime.

Here is the seasonal variation for property crime in Cook County (Chicago, Illinois), known for its severe winters. The average temperature in November is significantly lower than the summer months but does not reach the lows of December through March, where we see significantly lower than average property crime. Noting the y-axis to this plot, Cook County experiences strong seasonal variation, with property crime 15% lower than average in February and almost 13% higher than average in August.

Fulton County (Atlanta, Georgia) reflects normal seasonality except for a particularly sharp spike in property crime in November. However, all of Fulton County’s seasonal variation is not particularly pronounced — March shows the strongest variation at -10.2% and November shows an increase of 6.2%

Harris County (Houston, Texas) shows the second-warmest year-round temperatures of our sampled counties and features normal seasonality except for November’s modest increase in property crime. The seasonal shifts in Harris County property crime are relatively mild — February and March drop as much as -4.5% below average and spike 3.2% in June. Harris County experiences an average of 10.6 hours of daylight in November, the most average daylight of any of our sampled counties.

Los Angeles County (Los Angeles, California) features our warmest year-round average temperatures and one of our more striking dissimilarities from average national seasonality. Specifically worth noting is a relatively active November. November is warm in Los Angeles, and among our sample of counties, it has the second-longest average daylight hours for the month, with 10.4.

Multnomah County (Portland, Oregon) shows similar differences in seasonality as Los Angeles but with a stronger effect. Multnomah County’s property crime score is up 6.1% over the national average in November, the strongest deviation it experiences all year.

Multnomah County has the 4th highest average temperature in November, but the shortest average daylight with 9.5 hours. Understood through the mechanism of Routine Activity Theory, the mild temperatures may encourage evening activity and nightlife, while relatively long hours of darkness are advantageous to potential offenders.

Unpacking Daily and Hourly Crime Trends

This sample of counties containing major American cities highlights the high-level ways in which seasonality can be a strong rule of thumb, but far from a hard and fast rule. Varying climatological facets, such as particularly mild or severe winters, affect routines and human movement in ways critical to the delicate relationships between potential offenders and their targets.

The Pinkerton Crime Index also features county-level heatmaps unpacking crime risk 24/7, producing 168 scores of county risk sensitive to the day and the hour, for violent and property crime risk. Driven by statistical machinery sensitive to unique dynamics of offense counts and population flows, these graphs provide another granular view of local risk. Below, the heatmap breaks down hourly property crime risk through the week, for October in Cook County, Illinois.

Cook County’s property crime risk remains lowest in the early mornings, rising in the afternoons. This heatmap captures acute risk heightening on Fridays from the afternoon into the evening. It also picks up elevated property crime risk in the early morning hours of Saturday and Sunday, associated with people leaving bars. These maps update monthly and capture the way crime risk is driven both by routine human movements and factors like daylight and temperature. The scene at 8 PM in most communities varies greatly from summer to winter, and Pinkerton’s statistical machinery is designed to factor in these seasonal changes and their relationship to other facets of risk.

The value of monthly crime data updates

Where a single annual indicator can fail to depict the sensitivity of crime risk, our time series tools provide insights into how these risks evolve over the year within communities, highlighting the importance of monthly updates to crime scores and detailed time-specific analyses.

Because the city is more than just one city — it is many cities. Every city is multifaceted, capable of feeling like different places depending on the timing of observation and the specific areas analyzed. Questions like “Is this city safe?” or “How is traffic in this city?” reveal diverse answers shaped by changing conditions. The dynamics of suitable targets, motivated offenders, and guardianship shift seasonally, influenced by various societal and environmental factors.

The Pinkerton Crime Index blends a variety of prediction models, deconstructing real crime data and the underlying criminogenic factors that make up risk. Some of these models focus on fast-moving, noisy data, while others emphasize the glacial shifts of socioeconomic circumstance. These cutting-edge models are backstopped by the perspectives and understanding of real-life risk professionals.

The Pinkerton Crime Index updates risk scores monthly. Dashboard users can set alerts to be notified when a risk score has changed by a given percentage (%). Report clients will note the variances in their new monthly crime reports. These tools facilitate clear-eyed, up-to-date decision-making.