Are there any trends in duration of police stops by race in Boulder Colorado?
Issues revolving around racial discrimination have been a important issue in the United States for quite some time now. CNN published a story in 2019 referring to a study from Stanford University titled “Researchers studied nearly 100 million traffic stops and found black motorists are more likely to be pulled over” (Willingham, 2019). This is no new idea, activists and people of color have spoken on the discrimination experienced when it comes to police. This article writes that the study found that black people are 20% more likely to be pulled over on average than white people (Willingham, 2019). We can see similar issues with some policies that have been in place before like the stop and frisk. Stop and frisk was somewhat recently found unconstitutional due to evidence showing that it was essentially weaponized against the people of color population of New York City. Here are a few alarming statistics from civilrights.org. Number 1: “between 2004 and 2012, the New York Police Department made 4.4 million stops under the citywide policy. More than 80 percent of those stopped were Black and Latino people” (2023). Second is “The likelihood a stop of an African-American New Yorker yielded a weapon was half that of White New Yorkers stopped, and the likelihood of finding contraband on an African American who was stopped was one-third that of White New Yorkers stopped” (2023). These statistics should turn heads in my opinion, this kind of discrimination that was allowed until recently is unacceptable.
This motivated me to look into the Boulder Colorado Police Department. Boulder is widley known as a quite progressive city and I believe that the city itself does identify with that as well. I feel that the majority of the Boulder population resonates with that as well. Not only is the city quite progressive, or at least presents as so, but it is a college town as well. College towns tend to be more democratically leaning as well. That said I wanted to see if the data I looked at reflected a similar story from a policing standpoint. I do believe that it is important to mention that Boulder compared to very many other cities is not very diverse and I think it will be interesting to think about that as I look through this data.
The dataset that I have chosen to explore is Boulder Colorado’s police stops demographics. This is a public dataset that has information on stops initiated by officers so it does not include stops where an officer is responding to a call. The data shows the sex, race, ethnicity, year of birth, whether they are a Boulder resident, and the duration of the stop rounded to five minute intervals.
I have focused on the race of the person stop and the duration of the stop to see if there are any possible biases that could be important to explore deeper. First I removed a couple of columns that were not important to the question I am asking and contained null values. I then wanted to get a general idea of what this data shows so I created a chart showing the average duration of stops for each race listed in the data set. For reference: A:Asian, B:Black, I:American Indian or Alaskan Native, U: Unknown, and W:White (as listed in the data dictionary provided with the dataset).
From this chart we can see that although white people do not have the shortest average duration per stop we can still see that black, american indian and alaskan native, and unknown race stops do have higher average durations. So there is still more to dig into.
I then made histograms for each race listed in the data to try and take a look at the frequency of stops in different duration bins.
It is hard to make any sort of concrete claims from these charts due to the massive range of number of people stopped of each race. If we look closely and consider the scales used on the x and y axis we can see that the majority of the stops fall into the shorter duration bins but we can’t really make any claims yet. From there I created a table to show percentages to get a better idea of if there are any biases.
This table shows duration of stops ranging from essentially 1 minute to 60 minutes. We can easily tell that the previous assumption I made from the histograms was pretty accurate. Most of the stops across all races fall within the twenty minutes or less bins. What I found interesting is that although the first visualization showed that stops of asian people had the lowest average duration per stop, but they have the highest percentage of stops in multiple of the higher duration bins. I do not see any glaringly obvious callouts from this table that would indicate any discrimination. I think that across all races from the data set that they are all distributed relatively similarly. Another point that I think about when looking at the previous visualizations and this chart is that I think the massive number of white people stopped compared to the other races in this dataset makes this analysis a bit more difficult to analyze. Since it is relative to Boulder specifically this is unfortunately going to be the nature of the data.
Finally I decided to create a boxplot to analyze the data further.
In this boxplot we can see that there are far more outliers and a much smaller IQR for white stops than any other group. Due to the larger number of instances for white stops I believe that this is impacting the measures of central tendencies in a huge way which is why we see such variation in white stops than other groups.
Unfortunately I believe that answering this question based on the data that I analyzed is very difficult. The incredibly overwhelming majority of the stops recorded are white people and when I say this, the difference is thousands. I will admit this is something that I did anticipate based on my knowledge of the city of Boulder’s demographic, but I did not think that it would be this severe. With the difference being so high I do not feel very confident in making any very concrete statements about the trends in policing. But based on what I have produced as far as visualizations and analysis I do not think that I have any strong evidence that Boulder Police Department is wildly biased towards any one group based on this data. That said I would like to reiterate that I do not feel very confident in making the statement that Boulder Police Department does not have any bias since this data was so skewed towards white stops. I think the strongest evidence produced probably comes from the percentages shown in the table above. If we look at those percentages as well as the histograms for just stops of white vs black people we can see that the data is distributed in a pretty similar fashion. Unfortunately even in the histograms for each group the number of stops is making it quite difficult to make any super close comparisons. I think my best bet at trying to make any sort of true analysis from what I have would be from the percentages.
Finally I would like to say purely based on the data that I have looked at I think that there is no reason to believe that the Boulder Police Department is showing very much if any bias toward any racial group considered in this dataset. I would say that the fact that the city itself and its residents tend to be more democratic or left leaning could be somewhat reflected in its police activity. I think that even with data from years dating back to 30 or even 50 years prior would not help in the case of trying to make a concrete statement from this data specifically, due to the nature of Boulder’s demographics. I believe that it would probably actually make it more difficult to analyze and probably skew the data even more in the way of white stops.
All of that said I would like to once again state that I am not very comfortable or confident in making any very serious assumptions or claims about Boulder PD based on what I found in this data. I would like to believe that Boulder PD does a pretty good job of remaining relatively unbiased, which I think if you really wanted to based on what is seen here you could try to. Although I feel that you would get a lot of pushback from someone who looked at this analysis with any sort of criticalness. So my final statement would be that I was not able to come to any real decisions or answer the driving question that I had about this data. In the future I would probably try to draw a different conclusion from this data specifically or pair it with data from a city that has a similar demographic profile to see if there are any differences to make more of a concrete statement.
