With DC Chief Cathy Lanier announcing her retirement recently  to take over as NFL’s head of security, it seemed like the right time to take a fresh look at DC’s crime statistics. We here at District Ninja have largely stayed away from Crime as a subject as it is well covered by mainstream media, however with our new partners at B3 Analytics, we wanted to take a new approach to an otherwise well covered subject. While brain storming how to bring new relevancy to crime data we asked ourselves: how does crime effect  those who attend DC’s many sporting events?

So in leveraging data from OpenDataDC, CodeForDC, and data we cultivated from sports statistics sites, we were able to take a deeper look into this question: between attendees of both Washington Nationals and Washington Wizards home games, who is more likely to be a victim of a crime?

 

Overview

First the B3 team crunched raw data for each game, crime incidents within a half mile of the venue, and total attendance for each game from January 2011 to December 2014. By filtering down all crime that occurred 90 minutes before game start time until 90 minutes after game time and that also occurred within a half mile from each team’s stadium, we were able to determine with reasonable confidence which crime incidents were plausible to effect a game attendee. Below you can see a clustered map of all incidents found during our study period.

PlottedMap

After reviewing these numbers it is obvious that much like their record, when it comes to crime, things also look worse for Wizards fans than Nationals fans. Despite the Nationals having almost twice the amount of total games and almost four times the number of attendees, Wizards fans experienced 1.5x as many crime incidents as Nationals fans but this raw number barely scratches the surface when looking at this question.

B3 venue heat maps

In order to look a little deeper for applicable significance, we looked to see whether Wizards fans and Nationals fans are at potential risk for different types of crime. Based on highest occurrence, a fact that will shock few from the DC area, Nationals fans and Wizards fans both are at greatest risk to experience theft and theft from vehicles. That said, with each season, Wizards fans faced an increasing trend in the number of theft incidents year-over-year. In contrast, Nationals Park has never seen theft incident frequency like the Verizon Center and that amount is now decreasing as of last year. You can also see that with the exception of theft and theft from vehicles, none of the 6 other categories ever exceeded 20 total incidents in one season.

 

Next, we tried to see if there was any clustering of crimes based on distance from the sports venues. In other words: do crime incidents occur closer to the venues, further out from the venues, or are they distributed evenly across the half mile radius from the venues?

crime incident bar charts

In order to look at the distribution of crime based on distance from venue more closely, we broke down the total number of incidents by .04 mile intervals. Looking at this breakdown, we noticed that crimes near the Verizon Center were concentrated closer to the arena, while crime incidents decreased slightly as you move further from the arena. In contrast, the crime around Nationals Park peaks within 0.12 miles of the park, but the other crime incidents are concentrated at an average of 0.3 miles from the park.

Once we discovered the crime differences based on distance, we wanted to see if the day of the week had any significance.

scatter plot - Wizards

The chart above represents the distribution of crime events in the relevant study area for Wizards games broken down by day of the week and distance from the stadium. The Wizards rarely played on Thursdays and Sundays; therefore, we see low volumes of crime on those days. Based on this distribution, Wizards fans do not need to be concerned with any particular game day being worse for crime than another.

scatter plot - Nats

Unlike the Verizon Center, when looking around Nationals Stadium certain days of the week appear to be “safer” than others. Based on our study, the “safest” day of the week to attend a Nationals game is Thursday as there is 58% less crime near the stadium on Thursdays than other day of the week.

 

Conclusions

RiskofWizvsNat

With all of the caveats we have presented above, we wanted to determine the actual “risk” for Wizards fans versus Nationals fans when it comes to being a victim of crime. To do this, our friends at B3 ran various statistical tests to understand if it’s riskier for fans to attend a Wizards game or a Nationals game (more information on these tests in notes below). The raw number shows that there’s a 5.77x greater risk of being a victim of a crime attending a Wizards game than a Nationals game, but we wanted to see if there was a statistically significant difference. To see the tests we used to answer this question, check out the Ninja Notes at the end of the article.

Statistical testing showed that the rate of crime incidents per game attendance are not equal between Nationals and Wizards games. Further testing showed that crime rate per fan at Nationals Park is significantly less than the crime rate for fans at the Verizon Center.

crime incident forecast

With all this testing in mind, we wanted to try to forecast Nationals and Wizards crime incidents. Since the crime data ended in December 2014, the forecast starts in January 2015 and goes through May 2015. If we can get our hands on updated data, we can predict the frequency and location of incidents to assist with public safety.

 

Untitled2

Our previous conclusions above state that there is a statistically higher chance of being a victim of a crime at a Wizards game. Our conclusion could be questioned if there was considerably more crime across the District during Wizards season (November – April) versus Nationals season (April – October).  However, the graph above shows the seasonal aspect of crime across the District where crime occurs more frequently in warmer months.

Untitled3

Is there a higher chance of being involved in a crime on game day? Statistical testing indicated that there is no statistical difference in the crime rates on Nationals game days versus non-game days. The same can be said for Wizards game days and non-game days.

In sum, there is statistically significant differences between the crime rate per fan at a Wizards game and a Nationals game. When it comes to distance, we found a very low crime rate close to Nationals Park clustered around 0.3 miles. In contrast, crime around the Verizon center was concentrated closer to the arena and slightly decreases as distance increases from the arena. Finally, risk seems to remain with Wizards fans as they are 5.77 times more at risk of being a victim of crime on Wizards game days than Nationals fans on Nationals game days.

—————

 

Ninja Notes

A. Testing used to determine risk of Wizards vs. Nationals Games

Data used for statistical testing

Datapreparedforstatisticaltesting

Statistics for Table of Fans by Response

Statistical testing shows the Pearson chi-square test (listed as “Chi-Square” in the table). The small p-value for the test (p<0.0001) indicates that the null hypothesis of equal proportions can be rejected and that the proportions are unequal.

Fischerstest

For the chi-square test to be valid, the cell counts must not be too small.  When some cell counts are too small, the Fisher’s exact test can be used.  The Fisher test, while more conservative, also shows a significant difference between the proportions (p<0.0001 as show in the last row of the table).

1. MLB is 162 games long while NBA is 82 games long. Also, it is worth noting that the 2011-12 NBA season tipped off on Christmas Day due to a player lockout that season, thus that season has 66 total games.

2. We considered all crimes that occurred 90 minutes before game start time until 90 minutes after game end time. The average shifts, but MLB games that are 9 innings seem to be on average roughly 3 hours in length. The average non-overtime NBA game length is 2 hours and 15 minutes.

3. Nationals stadium seats 41,313 while Verizon Center holds 18,277.

Comments

  1. Great article. I really like the way this piece emphasizes exploring hypotheses over algorithm or technology bravado.

    I wonder if there is something about the physical structure of the venue that contributes to the differences in crime rates. I’ve seen that the design of an area (layout of streets, positions of lights, etc) can affect the rates of certain crimes. In a follow-up piece, I would be interested in seeing what happens when you pursue this question.

Leave a Reply

Your email address will not be published. Required fields are marked *