Incident Statistics | University of Michigan DPS

University of Michigan’s Incident Data

Home Statistics About

Observations On Data Trends

It should be clear that the visualizations show not all days, weeks, incidents, hours, etc are created equal. For instance, the culmination in data points to a Saturday in September between 4 - 5PM being most likely time for an incident to be reported. By drilling down into specific types of incidents, it is revealed that none of them have this exact trend and detract in some manner. Below are a few observations on the different types of incidents.

Minors in possession of alcohol (hereafter referred to as MIPs) has to be the favorite visualization because it is so telling. Contrary to most other incidents, MIPs occur almost exclusively between Friday and Sunday and between the hours of 11PM and 5AM. No other incident has such an extreme bias. Also, Welcome Week, the week where students return to Ann Arbor for the fall semester, is by far and away responsible for the most MIPs reported. One can hypothesize that warm weather, no classes, and social nature of the week may induce students into bold behaviors. Other weeks popular with MIP incidents are Halloween and the week before Thanksgiving, which isn't hard to believe. A close companion to MIPs are ambulance requests, which also is concentrated around Welcome Week and Saturdays, but MIPs and ambulance requests differ in hours reported with ambulance requests predominantly being reported during business hours. Minors succumbing to alcohol poisioning, in part, can explain the similarities.

Larceny from buildings, which is an incident that involves reports of stolen laptops, wallets, coats, is a type of incident that is most likely to occur during business hours Monday through Friday while winter and fall semesters are in session. Explanation for this could be that during these times, the most people are on campus and in general, a higher concentration of people leads to a higher concentration of thefts.

Remarkably, there are incidents where the frequency does not fluxuate with the number of students. These include: property damage (accidental), assist other agency, and to some extent, traffic accident (off roadway). The only immediate explanation is that other agencies (Ann Arbor police department, fire department, etc) aren't influenced by the flux of students and so the university's police has a constant flow of activity from other agencies. Intuitive explanations for the other two types of incidents remain elusive.

Notes on Implementation

The close observer will be quick to point out the fact that not every year has the same number of days/weeks, and that days don't have a set week that they occur in. The point of these graphics isn't to be technically rigorous, but rather give general trends. The other option would be to display all data across all years, but collapsing all the data into a single year more easily gives rise to trend detection.