Police-involved killings have made national headlines over the past few years. Each time they do, a wave of anti-police sentiment sweeps the nation.
Using more than 1.2 million tweets, we show where in the U.S. people approve or disapprove of law enforcement the most. We also take a look at how Americans’ feelings about the police fluctuated in the nearly two years since Ferguson, Missouri, resident Michael Brown was shot to death by Officer Darren Wilson.
When we consider the sentiment of police-related tweets, it’s important to note that negative-sentiment tweets and a negative sentiment toward the police aren’t necessarily mutually inclusive. The same is true for positive tweets.
Overall – for most months from June 2014 to January 2016 – police-related tweets expressed positive opinions:
“Got pulled over thinking I was going to get a ticket. Turns out the cop just wanted to talk bc he was bored.. (sic) What’s good officer john? Haha.”
“I need to shut up about the police until I’m in the position to really make a change.”
But when reports of police-involved fatalities flared up, Twitter users tended to express rage and hostility. (Note: These messages could be skewed negatively toward either the police or the victim.)
The first big negative-sentiment dip for police-related tweets came in August 2014 when Michael Brown was shot by Officer Darren Wilson in Ferguson. Residents protested, claiming Wilson had no reason to shoot the unarmed Brown. Wilson said he acted in self-defense when Brown tried to take his pistol. This came after years of police bias against the city’s black residents, detailed in a report on the shooting by the Department of Justice.
Unsurprisingly, tweets from that month were mostly negative. Some of the most retweeted and favorited tweets had messages such as, “That’s sad that 18 year old (sic) got shot by the police.”
But it’s not just police officers’ shooting of unarmed men that trigger a dip in sentiment expressed by police-related tweets. And the negative-sentiment posts aren’t always directed at the police.
New York City resident Ismaaiyl Brinsley killed officers Wenjian Liu and Rafael Ramos in December 2014. He said he did it to get revenge on New York City police after NYPD officers killed Staten Island resident Eric Garner earlier that year in July. Liu and Ramos had no known significant connection to the Garner incident.
Negative-sounding tweets from that time referenced either the death of Garner or officers Liu and Ramos.
One popular tweet read: “My heart is broken over the two cops that were killed in Brooklyn. Really hurts me to know that people believe this is ‘getting even’. (sic)”
Another prominent tweet had a different message: “Not shocked but angry that the cop that (sic) strangled Eric Garner to death for selling cigarettes on the street will not be indicted.”
Twitter users’ sentiments in police-related tweets nosedived two more times since then.
The first hint of discord came in April 2015 – the same time Baltimore resident Freddie Gray died of a spinal cord injury after city police charged him with carrying an illegal switchblade and arrested him. Many suspected officers used excessive force during and after the arrest, causing Gray to slip into a coma before dying shortly afterward.
More conflicted tweets also came through in July 2015, after Texas state trooper Brian Encinia pulled over Sandra Bland. What started as a traffic stop became a heated argument, leading to Encinia forcing Bland out of her car at gunpoint and arresting her. The incident, which was recorded by the video camera of the police car Encinia drove, went public. Three days after Bland’s arrest, she was found dead in her jail cell.
In the same month in Ohio, University of Cincinnati officer Ray Tensing fatally shot resident Samuel DuBose during a traffic stop. Tensing pulled over DuBose because his car had no front license plate. During the stop,Tensing opened the car door, but DuBose slammed it shut then started his car, according to Tensing’s body camera footage. At that point, Tensing reached into the car to try turning off the ignition, he stated in a university report. Two seconds later, Tensing drew his pistol and shot DuBose in the head. The shooting was entirely preventable, the university’s report stated.
Interestingly, the more police-related tweets there are, the more negative the sentiment expressed by those tweets.
That’s because the number of police-related tweets spiked after police-involved killings. Since the death of DuBose, there have been no major officer-related fatalities, meaning tweets about the police have been less negative.
There’s definitely a geographic trend when it comes to tweets about police in America. The Mississippi Delta seems to radiate bad feelings in police-related tweets. Tweets from Louisiana, Mississippi, and Arkansas expressed some of the most negativity. The rest of the Deep South, Maryland, and West Virginia also held negative sentiments.
Meanwhile, in the colder, less populous parts of the country, Twitter users seem to note positive feelings in their police-related tweets. Wyoming and North Dakota top the list.
We also looked at cities with at least 1,000 police-related tweets. Twitter users in Richmond had the most positive sentiment when they talked about law enforcement matters, while four Texas cities also made the top 10 in this category: McKinney, Plano, Lubbock, and College Station.
On the other end of the spectrum, we see a similar pattern we saw at the state level. The cities where police-related tweets are most negative are in the South: Irving, Texas; Montgomery, Alabama; Jackson, Mississippi; Fayetteville, North Carolina; and Memphis, Tennessee. Ferguson is also near the bottom with tweets from Chandler, Arizona, which express more negativity.
Maryland, Louisiana, Texas and Ohio tweeted about the police the most in our analysis. Maryland and Ohio were the sites of the alleged officer-involved killings of Freddie Gray and Samuel DuBose, while Texas had the death of Sandra Bland.
Unlike the map showing Twitter users’ attitudes in police-related tweets, there seems to be no geographic pattern regarding the number of tweets about law enforcement.
But on the opposite end, the Western, Mountain, and Plains states from Montana to New Mexico tweeted less often about the police than the rest of the country.
When it comes to law enforcement, men tweet about police almost twice as much as women. Among tweets made by users who declared their gender in their Twitter profiles, nearly two-thirds were male.
Finally, here’s a look at how Twitter users in America’s 10 biggest cities feel about law enforcement. Chicago, Houston, and Phoenix tweets express the most negative opinions about police-related topics, while Dallas, San Antonio, and San Jose expressed the most positive.
While tweets about police aren’t usually negative, negative-sentiment tweets flare up after incidents where the death of a black person is perceived by many as the police’s fault. Places that tweeted most about police had a death that was blamed on police. Places with fewer tweets about police had no deaths blamed on police that made national headlines; tweets there tend to have relatively positive sentiment.
We scraped and analyzed 1,295,071 Twitter posts containing one of these words or phrases: “cop,” “cops,” “officer,” “po po,” “police,” “policeman,” “policemen,” “policewoman,” “policewomen,” and “popo.” For states and cities, we analyzed 435,872 geotagged tweets and used their coordinates to determine their locations.
We calculated the number of police-related tweets per 100,000 residents in each state by combining the number of police-related tweets in our data with the population of those states as reported by the U.S. Census’ 2014 American Community Survey.
We determined how positive or negative a tweet was by running it through Text-Processing.com’s API, which is built on the Python Natural Language Toolkit. If the API rated the text less than zero, it was deemed to have negative sentiment. If the rating was zero, it was neutral in tone. If it was more than zero, the message was deemed to have positive sentiment.
U.S. Census Bureau, 2014 American Community Survey
“Darren Wilson on why he shot Michael Brown” - New York Post http://nypost.com/2014/11/25/officer-darren-wilson-says-michael-brown-taunted-him/
“The 2014 protests over Michael Brown, explained” - Vox. http://www.vox.com/cards/mike-brown-protests-ferguson-missouri/justice-department-investigation-ferguson-police-racism
“Many identities of New York officers’ killer in a life of wrong turns” - The New York Times http://www.nytimes.com/2015/01/03/nyregion/ismaaiyl-brinsleys-many-identities-fueled-life-of-wrong-turns.html
“Six Baltimore officers suspended over police-van death of Freddie Gray” - The Guardian http://www.theguardian.com/us-news/2015/apr/20/baltimore-officers-suspended-death-freddie-gray
“Sandra Bland traffic stop” - Texas Department of Public Safety YouTube channel https://youtu.be/CaW09Ymr2BA?t=10m30s
“Assessing the legality of Sandra Bland’s arrest” - The New York Times http://www.nytimes.com/interactive/2015/07/20/us/sandra-bland-arrest-death-videos-maps.html
“Review and Investigation of Officer Raymond M. Tensing’s Use of Deadly Force on July 19, 2015: University of Cincinnati Police Department”, pp 6, 22-30 - University of Cincinnati http://www.uc.edu/content/dam/uc/safety-reform/documents/Kroll%20Report%20of%20Investigation%208.31.2015.pdf