We've highlighted two dates that might be relevant. The first line marks the death of George Floyd, which sparked a massive series of protests across the nation. The second marks the date that the Seattle Chief of Police sent out a statement promising a "complete re-envisioning of community safety and the police department's role in it".

Where did this data come from?

This data comes from the City of Seattle open data portal. It records all incidents where members of the Seattle Police Department used force on a member of the public, starting from January 2014 and continuing up to the present. It includes information on location, date and time of the incident, race and gender of the subject, and the level of force. You can find more information about the dataset here.

The Use of Force (UoF) dataset recognizes four levels of force:

• Type I: force that causes transitory pain or the complaint of transitory pain, not including handcuff discomfort, which is reported separately.
• Type II: force that causes or is reasonably expected to cause physical injury greater than transitory pain but less than great or substantial bodily harm.
• Type III: force that causes or is reasonably expected to cause great bodily harm, substantial bodily harm, loss of consciousness, or death.
• Type IV: officer-involved shooting.

Zoom in to see how the frequency of different levels of force has changed over time

From the two time graphs above, it's worth noting how much the trend changes depending on what period of time you look at. Over the past six years, we have seen a gradual decrease in use of force among the SPD.

However, the issue of police violence certainly hasn't gone away. From a distance, it may look like the use of force following the George Floyd protests shot down dramatically, but a closer look reveals that there's much more nuance. In reality, the data simply got more erratic, alternating between large peaks and periods of relative inactivity. With the data we have so far, it's hard to tell whether the change in UoF frequency since the death of George Floyd represents a significant departure from the previous trend.

Clearly, not all races are affected equally by police violence. But maybe that can be explained by the demographics of the neighborhoods? What about economic factors, do those play a role?

A closer look at the context:

So what does all this mean?

Comparing the proportion of different races in the Seattle population with the proportion of different races in the Use of Force data:

Select different menu options to flip between population data, UoF data, and the difference between them:

In theory, if race had no effect on how likely someone is to be the subject of a Use of Force incident, we'd expect people of each race to be subjects of UoF incidents at about the same rate as they are residents of the area. In other words, the proportion of UoF subjects of each race should be about the same as the proportion of the Seattle population of each race.

When we see large differences between those proportions, that means that certain racial groups are disproportionately affected by police violence. African Americans, for instance, are only about 3% of the population, yet over 30% of Use of Force subjects: ten times as likely to be UoF subjects than we would expect if race had no impact on the Use of Force. On the other hand, white people are over 70% of the population and only 40% of UoF subjects; white people are targeted almost half as often as we would expect.

Since SPD Chief Best's Statement on June 22nd, 2020 promising "Re-Enivisioning Public Safety" after the Black Lives Matter protests in May and June, have things gotten better?

When we compare the graphs from the entire dataset vs. from after the SPD statement, we see that the percentages of Black/African Americans experiencing use of force has decreased, getting closer to what we'd expect based on their percentage of the population. However, we can also note that the number of white people involved in UoF incidents has decreased as well, making white people even less likely to be involved in UoF incidents despite being a majority of the Seattle population. Instead we see a rise in incidents where the subject race is not specified, which is a pattern we can see in the very first visualization as well.