Race + Data Journalism = Angry Readers

by Categorized: Data, Media, Public Safety Date:

There are few things – if there are any – that are more difficult to talk about in American society than issues of race and ethnicity. That was evident in the visceral reaction many readers had to the Courant’s story Sunday on traffic-stop data. Our report, based on data police departments are required to collect and that the state was supposed to have been analyzing for years, revealed that blacks and Hispanics were significantly more likely to receive a citation after a motor-vehicle stop, compared to white drivers pulled over for the same offense.

Those numbers struck a nerve.

“I find it disgusting that an article like this paints a broad picture that CT cops are racist when that is the furthest thing from the truth,” one reader wrote.

“Way to play the race card and drive a bigger wedge between the police and the community they serve!” wrote another.

And an out-of-state police detective minced no words. “Selling out for the ratings totally ruins your credibility as a real investigative reporter and makes you just another unethical, greedy scumbag ruining a great country,” he told me by email. “Oh, and good luck with your next meeting with one of my brothers or sisters on the highway. I’m sure they will remember your name and try to impress you with how fast they can write up a white guy to stay within the bounds of political correctness.”

Attacks like that come with the job. But there were also more-temperate comments aimed at the Courant’s statistical analysis. A few correctly identified limitations in the data; most were based on misunderstandings of what we did or how we did it. We think it’s worth addressing all of those criticisms to help readers understand what was, and was not, in the data.

What follows are the most common comments we heard following Sunday’s story:

The numbers are higher because minorities commit more violations.

“Between the whites, blacks, and Hispanics, which group is committing the majority of the infractions? The stats stated mean absolutely nothing. Just another attempt to play the race card.”

“If the Blacks and Hispanic drivers didn’t run stop signs, go through red traffic lights and have burned out headlights & taillights, they wouldn’t get stopped. What do the Blacks and Hispanic drivers want the Police to do, look the other way? This is a poor analysis by the Hartford Courant!”

 Analyses of racial-profiling frequently look to compare the rates at which white and minority drivers are pulled over by police. But as we explained high in the story, that is not the analysis we performed. Rather, we examined all stops and asked essentially these questions: For every 100 white drivers pulled over because police determined they violated a certain motor-vehicle law, how many ultimately received a ticket or summons? For every 100 black drivers pulled over for the same violation, how many were cited? And for every 100 Hispanic drivers pulled over for that violation, how many were cited?

For that analysis, the proportion of tickets among the three groups is irrelevant. Whether white drivers comprised 90 percent of those who are stopped or 5 percent of those who are stopped, the number of citations per 100 drivers would not change.

 The numbers are higher because the stops were made in areas where there are more minorities.

“I think while there may be some disparity, we have to look at the communities where the minority population is large. This would account for more frequent police interaction and skew the statistics.”

“There certainly is a logical explanation for this so called ‘disparity.’ What are the demographics of the areas these tickets are being written in? Are there more Hispanics and blacks than whites in the north and south end of Hartford. Look at the majority population in the areas of the offenses. This story is racially biased…against the whites!”

As above, the racial makeup of the area where the stops occurred is also statistically irrelevant to the analysis conducted by the Courant. Regardless of whether the stops were made in cities with significant black and Hispanic populations or made in nearly all-white towns, we looked only at the percentage of stopped drivers who were cited. That analysis is independent of the demographics where the stops occurred.

 The numbers are too small to be relevant.

“You fail to cite the total number of people stopped. In one section, you say 52% of Hispanic motorists got a citation. Well that statement is a little unfair if you fail to state the total number of Hispanics stopped. If 2 Hispanics were stopped and one received a citation that would make 50%!”

The story did in fact describe the minimum number of stops we required before a statistic was cited, and if the numbers were too small to be statistically meaningful, we didn’t use them. In the example this commenter used, 52 percent of Hispanics stopped for an equipment violation in Westport received a citation, compared to 31 percent of white motorists. That was based on stops involving 244 Hispanics and 607 whites. At those numbers, the disparity is statistically significant at the 99.999 percent level.

 You confused correlation with causation; it’s not race and ethnicity, it’s some other factor that simply correlates to those demographics.

“There are innocent explanations, that may or may not be true. Cops could ticket younger drivers intentionally. Are Hispanics, on average, younger than blacks and white? And are blacks, on average, younger than whites?”

Data do not establish causation, but statisticians work to isolate factors to support or refute a hypothesis. Could the traffic-stop data really just indicate that younger motorists are more likely to get ticketed? Black and Hispanic drivers stopped by police were, on average, younger than whites, and it’s a factor – but it doesn’t explain the numbers. The Courant did not perform a rigorous multi-variable regression analysis, but we did run some tests to equalize the data by both age and gender to see how those variables affected the racial and ethnic disparity. They had some impact, but not much. As an example, without adjusting for age and gender, blacks were 58 percent more likely than whites to receive a citation when stopped for an equipment violation. Adjusting for age and gender reduced the figure from 58 percent more likely than whites, to 51 percent more likely. Similarly, the raw data showed Hispanic drivers were 160 percent more likely to be cited than whites for an equipment violation. Adjusting for age and gender, Hispanics were still 136 percent more likely to be cited.

But those who wrote in suggested other possible correlations.

“It’s well-known that a driver being polite is more likely to result in just a warning. But if the drivers were rude and indignant, sure they’re going to get a ticket,” one wrote. “Are blacks and Hispanics ruder during traffic stops than whites? Very possibly so. It’s just a cultural thing.”

Other posters objected to that generalization. From a statistical standpoint, it’s a factor the data can’t address.

 The data doesn’t capture other factors that can affect whether a ticket is issued.

“The article does not delve into the relationship between the reason a car is initially stopped and the type of infraction cited in a ticket. If a driver is initially stopped for an equipment failure (such as a broken tail lamp) and is issued a ticket for failure to have a valid insurance certificate, that could explain some of the disparity. Blacks and Hispanics may drive uninsured cars at a higher rate than whites.”

“There are many factors that go into whether someone gets a ticket or not. Often people’s driving history is also taken into account but the article fails to cite that.”

 Those are legitimate concerns with the data. The form used by police does not record whether more serious violations were found after a car was stopped, nor does it indicate a driver’s past violations — both factors that could reasonably affect a police officer’s decision whether or not to issue a citation.

Are black and Hispanic drivers substantially more likely than whites to have multiple violations or poor driver histories? Is that the primary motivating factor controlling an officer’s decision whether or not to open the ticket book? That will require more study – and more data.

 

Douglas Fuchs, the police chief in Redding and president of the Connecticut Police Chiefs Association, said he has already reached out the African-American Affairs Commission to work on promoting sensitivity training in police departments. But he does not believe officers in Connecticut engage in discrimination, and he doesn’t think the racial-profiling form departments have used for more than a decade accurately reflects what is happening on the state’s roadways.

“I do also think that the raw data does not truly indicate the motivation for the stop in the first place,” he wrote to me this week, “and had the data been continuously analyzed that the ‘form’ (which is now computerized in most PDs) might have been tweaked to better reflect what was actually taking place.”

But that analysis didn’t happen. That’s why we decided to acquire the data ourselves and let our readers know what we found.

If there are changes or developments on the topic, we’ll let you know. In the meantime, we want to continue hearing from you. So share your thoughts on the comment board for this post, or send an email to thescoop@courant.com or use our online contact form here.

The Courant is using Facebook comments on stories. To comment on courant.com articles, sign into Facebook and enter your comment in the field below. Comments will appear in your Facebook News Feed unless you choose otherwise. To report spam or abuse, click the X next to the comment. For guidelines on commenting, click here.

8 thoughts on “Race + Data Journalism = Angry Readers

  1. RustedRoot

    The “out of state detective’s” diatribe proves the point that racial profiling is present. Just what we need, another irrational cop. “Doth protest too much, detective?”

    Reply
  2. Pingback: Courant takes heat for report on police traffic stops « Chalmers@Lincoln

  3. Pikavippi-pankkitunnuksilla.fi

    I have my own blog and might bid on projects to blog for others. I believe there are some guidelines out there around what is allowed or not regarding images posted on public blogs (such as copyright rules, licensing, or trademarks). Is there anywhere to go to find industry guidelines for blog post images?.

    Reply
  4. Magali Romito

    This could be the appropriate blog for hopes to check out this topic. You already know a great deal its almost tricky to argue with you (not that When i would want…HaHa). You definitely put a fresh spin using a topic thats been discussing for many years. Wonderful stuff, just fantastic!

    Reply
  5. Yasmin Mannheimer

    Usually I don’t read article on blogs, however I would like to say that this write-up very pressured me to check out and do so! Your writing style has been surprised me. Thank you, quite great post.

    Reply

Leave a Reply

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