Analysis

With the widespread availability of the news due to the rise of digital media, there has been an increase in biased and inflammatory language in news headlines. What’s more is that this bias impacts the polarity of the headline. News nowadays appears to portray a heavily negative view of America, with very little positive sentiment to be found. This in fact works in their favor — negativity asserts more power over positivity in the brain; people are more likely to remember an negative headline over a positive one.

Let’s take a look at some of the data…

Has the media gotten more negative?

We tracked three metrics for sentiment: positivity, neutrality, and negativity.

Ignoring the gap in the CNN data, we can observe an upwards trend in media negativity between 2012 and 2021.

There are some particular areas of note where negativity spikes in several publications, such as during the 2015-2016 US presidential election cycle, as well as the first few months of a new presidents term. CNN and The New York Post both sharply increasing around mid-2015. The New York Times maintained a relatively low negativity, until a month into Donald Trump’s presidency. From there, the trend continues climbing until mid-2020, with a spike after Biden’s inauguration.

And these spikes make sense. After all, the 2016 election was arguably one of the most contentious elections in American history

Comparing all three sentiments

For the most part, the number of positive articles remains at a constant 5-10% or so! To put that into perspective, only 1 in 10 articles are considered "good news".

As negativity increases, neutrality decreases. We see that negativity peaks at around 40% for CNN, FOX, and the New York Post. That’s roughly 1 in every 2.5 articles.

As a side note, the AI model is really good at detecting explicit negativity (visibly negative sentiment). For implicit negativity, which appears neutral at first but contains negative undertones, it often will return a false-neutral. This means that the actual percentage of negative articles may very well be higher than what is shown here.

Implications

Your brain is biased towards negativity; you’re more likely to remember a negative headline than a positive one. The media employs this for obvious reasons — it’s much easier to get people to click and read through a negative article.

The media, undoubtedly, is an important part of our daily lives. News has the power to shape our opinions and views on the world, and thus, such trends should be acknowledged.

With a shroud of negative news around us, we tend to see a more distorted, pessimistic view of the world, which only hampers our outlook on our lives.

However, the perils of negative media expand beyond the feelings of individuals. When negative media is paired with biased media, it has the potential to create a toxic, divisive environment for our whole society. It creates an inferno of hate and polarization, fueled by incendiary headlines.

Bias skyrockets

Implicit vs Explicit Bias

To preface this section: Media bias can be an incredibly complex, nuanced topic, especially when working with Artificial Intelligence.

Take this headline for example:

What Biden's latest outlandish immigration action tells us about the next one

It’s quite easy to tell that it is biased; the headline expresses clear distaste for Biden’s immigration policies without providing any information about the policies themselves. It’s sole purpose is to perpetuate and reaffirm a negative view, one that may likely already be shared by the audience of FOX.

Headlines like this one, the journalistic equivalents to playground insults, are everywhere. But others aim to grasp the readers attention more subtly.

Clarence Thomas Should Not Get Away With It

Linguistically, this headline portrays a simple, neutral message. After all, it's just stating that Clarence Thomas should not get away with... something. But __what__ is he getting away with?

This is an example of implicit bias. The statement contains an opinionated political view, but requires extra context to be truly understood.

This is where AI models struggle. It’s easy to detect explicit bias, but implicit bias is much harder to detect.

If you’d like to try it out for yourself, you can run your own headlines through the detector used to gather the data.

Bias over time

With that out of the way, let’s take a look at how bias has risen over the years.

Much like negativity, we see media bias has been consistently increasing by great proportions. Given that such media molds and sways public opinion, it is truly concerning to observe such a derrailment in the reporting of facts.

Once more, we see that bias spiked heavily during the 2015-2016 election cycle, indicating that period of time was a clear catalyst in the rise of political bias.

Bias and Negativity go hand in hand

Let’s take a look at the bias of each news organization over time, plotted alongside it’s negativity.

These charts demonstrate a similar incline in both data points; the rises and drops in bias and negativity appear to be correlated.

It’s important to note that correlation does not imply causation. Negative media and biased media may very well be two separate issues, both increasing at the same time, but nevertheless, they indicate the rise in hatred and polarization within the media.

Keywords

Throughout the data set, certain topics and keywords were mentioned extremely frequently. These keywords were generally controversial in nature, and their frequency was correlated with the observed bias and negativity trends.

Trump

Seeing as there were huge spikes in bias and negativity during election cycles, we decided to plot the amount of times a headline mentioned either Donald Trump, Joe Biden, or Hillary Clinton.

Mentions of Trump skyrocketed after he announced his bid for presidency, reaching a rolling average of 5 to 18 times per day depending on the publication. If you look carefully after that, you will notice a small blip in his relevancy following the January 6 riots. It takes a steep drop after that point.

The fact that such a big event immediately lost relevancy, highlights that the media is quick to move on from events after they get stale.

Russia and China

Russia and China are the United States’ largest competitors on the economic and military front, and fueds between these countries are common.

Between the trade war in China, and an ever-lasting race for more power between the US and Russia, it’s no surprise that these countries are mentioned a lot in the news. Here’s what that looks like:

While there is a lot more to unpack between these data points, like the previous examples, we can see big spikes in their mentions during Trump’s presidency, along with smaller spikes during the election cycles. Additionally, there is a large spike in the mentions of Russia after it invaded Crimea in 2014.

Republicans and Democrats

Keeping in tune with the theme of politics, let’s look at a chart of the words “Republican” and “Democrat” over time.

The keywords appear to be in a rough equilibrium, with the exception of a few spikes in 2018 and 2019.

The media’s favorite word

Person A SLAMS! Person B for doing a totally normal thing!

Look who just got BLASTED! by congress!

Words like “slams” and “blasts” are used to make headlines more enticing as they indicate a certain emotional sentiment about an event, and are very commonly utilized in headlines.

Interestingly enough, we see that the New York Times and the BBC refrain from using this kind of language, potentially indicating a different style of reporting.

A few key observations:

• FOX News and the New York Post use this language all the time
• CNN loved to use “slam” during the 2016 election cycle.
• FOX and the New York Post used “slam” and “blast” a lot during the 2020 election season

Conclusion

In examining the data, it’s evident that media bias and negativity are intricately intertwined, shaping public perception in profound ways. Our analysis reveals a concerning rise in biased language and negative sentiment, particularly during significant events like election cycles. This pattern, as showcased in the data, not only influences individual attitudes but also exacerbates societal divisions.

Moreover, our research sheds light on the interplay between biased headlines and their impact on political polarization. The spikes in negativity coincide with increased mentions of political figures like Trump, Biden, and Clinton. This correlation highlights the media’s role in fueling political animosity, painting a picture of a society deeply affected by the narratives presented to them.

Crucially, media consumers are not passive recipients but active participants in this landscape. By critically evaluating news sources and recognizing implicit bias, we as a people have the ability to counter the divisive effects of sensationalized reporting, but polarization keeps us divided. The data underscores the urgency for media literacy initiatives, empowering people to discern between objective reporting and skewed narratives.

In essence, our analysis serves as a loud call for a more discerning, critically engaged society. Utilizing insights from the data, we can advocate for responsible journalism, demand transparency from media outlets, and promote unbiased reporting. By challenging the status quo and embracing a culture of informed skepticism, we can foster a media environment where accuracy, integrity, and fairness prevail — ultimately bridging gaps, fostering understanding, and uniting a polarized nation.

I've padded the bottom of this page for readability, thank you to https://cpu.land/ for the idea!