A common complaint of President Trump and others in the GOP is that a high percentage of media coverage of him is “negative.” The official GOP Twitter account often tweets about this, sometimes citing a statistic from a Harvard study stating that over 90% of media coverage of Trump is negative. This, the President and his allies complain, is evidence of bias. In this post I argue that “negative” coverage itself isn’t necessarily “biased,” and is often perfectly fair. However, it is often easy to confuse negativity and bias, and it is similarly easy for them to overlap within the reporting of a story. As a result, many casual media observers feel like media sources have become recently more biased against Trump because of a seeming increase in negative reporting about him.

When is negative reporting simply unbiased reporting of the facts, and when is it bias? Almost 100% of stories about Harvey Weinstein’s sexual assaults are negative, but no one says it is because newspapers are biased against Weinstein himself. Almost 100% of stories about drunk drivers are negative, but no one says it is because the local news anchors are biased against drunk drivers. We intuit that the reporting is appropriate because the sexual assaults and the drunk driving themselves are bad things. Often, when the news reports that someone did a bad thing, it’s because the thing was actually bad.

It’s especially important in our current media and political environment to distinguish between negative coverage that contains minimal (or what we will refer to as “appropriate”) bias and from that which contains overt (or what we will refer to as “inappropriate”) bias. Making such distinctions helps explain a particular phenomenon of our time, which is that of conservative/Republican journalists and writers increasingly writing critical stories about President Trump. What does it mean when an historically conservative and/or Republican writer writes a piece that is “negative” about Trump? Does it mean that the conservative/Republican is now a liberal/Democrat? I argue that the answer is no, and many such journalists/writers have argued the same themselves. Does it mean they are “biased” against Trump? Not necessarily, as I explore below.

            Recent examples of this phenomenon abound. There are the articles, from pre-election 2016 to the present, of prominent “Never Trump” conservatives, from Jennifer Rubin of the Washington Post, to Max Boot of the Washington Post, to Rick Wilson (conservative author in publications across the political spectrum) and Bill Kristol of the former Weekly Standard. Then there are the articles of conservative authors who didn’t identify with the “Never Trump” crowd, such as Jonah Goldberg and David French of the National Review, but whose Trump-negative articles have recently increased in frequency.

            To meaningfully separate the concepts of negativity and bias, we need definitions of both. For purposes of this post, I define an article or story as “negative” if it creates a bad impression of a person or organization who is the subject of the story. I’ll get to definitions of bias shortly. We need these definitions in relation to a third defined concept, which we will call “the underlying thing.”

“The underlying thing” is defined here as the particular fact, action, deed, statement, or event that happens that is the topic of the story, separate from whom (person or organization) that underlying thing is about. For example, an underlying thing of a story can be “______ wore fancy boots” or “_______decorated the White House in this way” or “______ went on trial for corruption” or “______ said this false thing,” or “________ visited the troops,” or “________ pushed for a tax cut,” or “the stock market declined 500 points” or “the stock market rose 500 points.”

Here, an underlying thing can be positive, neutral, or negative, and widely agreed upon as such, so I refer to them as “consensus-positive,” “consensus-neutral,” or “consensus-negative.” I refer to an underlying thing as “consensus-negative” if the vast majority of Americans view a particular action, deed, or statement as definitely bad or wrong. We’re already treading in highly philosophical waters here but bear with me. There are some things we (members of our American society) agree upon as generally more negative than positive (keeping in mind that all generalizations are false).

The dictionary definitions of the word “bias” are incredibly broad, and (ironically) too broad to be useful in most discussions of media bias. I define two kinds of bias here.

The first I’ll refer to as “ideological bias,” which is the type of bias present when there is significant disagreement among Americans about whether the underlying thing is good/right (positive) or bad/wrong (negative). We identify bias within ourselves or within an article when we or the journalist/writer advocate that something is definitely right or wrong, knowing that a big chunk of our fellow citizens disagree. We can also use the term “political position bias” for this concept, and it is the type of bias most people are usually talking about when discussing bias in the media. We recognize this kind of bias fairly easily, detecting it when an article discusses polarizing issues such as abortion, guns, taxes, racism, and regulations.

The second kind of bias I will refer to as “personal bias,” which is the human tendency to import judgments about a person based on various factors, such as the person’s past actions, traits, or identity. Personal bias is often summarily expressed as just “liking” or “disliking” a person; liking or disliking Trump, liking or disliking Hillary for example. Personal bias in ourselves is present when we import judgments of the person into the underlying thing at issue, and personal bias in an article is present when a journalist/author imports his or her judgments of the person whom the underlying thing is about into coverage of the underlying thing.

Let’s go back to the trusty Media Bias Chart horizontal axis to explain why we define “negativity” and “bias” (both ideological bias and personal bias) in these ways.

Right in the middle of this axis is the word “Neutral.” Though this left-to-right spectrum feels intuitive, the question “what is neutral?” reveals the complexity of this simple left-to-right scale. Can anything really be “neutral?” Or does every person, culture, and society have biases? I submit it is the latter; all humans, cultures, and societies have biases. So what does it mean that many of us recognize, conceptually, the idea that something can neutral?

I submit that we generally consider two types of underlying things to be “neutral” The first is incontrovertible facts—things that happen and things that simply are. By “things that happen,” I mean things like “it snowed in Denver today” or “______ announced a run for President.” By “things that are,” I mean things like “the Earth is round” or “there are approximately 330 million people living in the US.” (We need not address contrarians’ opinions about these matters here). On the Ad Fontes Media Veracity scale, these facts are scored as 1 on a scale of 1-5. On our scale, a rating of 1 means “easily provable, and the proof is widely accepted.”

 I suggest that other underlying things that we, in our American society, consider to be neutral, are things about which we have a widely shared morality. That is, the things that large majorities (80-95%, depending on the topic) of us consider to be definitely good/right or bad/wrong are the things that fall into the category of “neutral.” Given our wide differences across our society, there isn’t a lot that falls into this category, but broad American moral principles like “murder is bad” or “democracy is good,” or “stealing is wrong” and “freedom is good” are ones on which we agree very widely (again, all generalizations are false). We agree on them so widely that when a journalist/writer takes a position or writes a story in line with those positions, it doesn’t even register to most of us as bias. We consider those positions neutral, even though some (such as “democracy is good” and “freedom is good”) are in fact American-specific biases. These positions aren’t necessarily widely agreed upon in other countries or cultures, which is why the left-right spectrum of the Media Bias Chart is US-specific, and other left-right spectrums would need to be created for international Media Bias Chart versions.

            We can have these shared opinions that something is good (a “consensus-positive,” like “a good Samaritan helped a person in need”) something is neither good nor bad (a “consensus-neutral” like “someone wore an appropriate piece of clothing”) or that something is bad (a “consensus-negative” like “a drunk driver killed someone”)

When an underlying thing in an article is either 1) an incontrovertible fact or 2) something on which we have a widely shared morality, straightforward reporting of the fact or a opinion in line with that widely shared morality is viewed as “neutral.” Here’s a flowchart of the concepts I have been laying out up to this point:

When a journalist/writer reports that it snowed today in Denver, we don’t consider that to be biased because that is an incontrovertible fact. When a journalist/writer reports that a drunk driver killed someone, and indicates through body language or tone that it is a bad thing that the drunk driver did, we don’t consider that story to be (ideologically or personally) “biased” against the drunk driver, because according to our shared morality, the underlying thing “drunk driver killed someone” is bad, and therefore the topic is consensus-negative. This would go in the center column on the chart for bias (reflecting appropriate levels of ideological and personal bias), as would a straightforward story saying a consensus-positive underlying thing is positive, or a consensus-neutral underlying thing is neither good nor bad.

It is easy to detect and distinguish ideological (political position) bias. These are the things on which we do not have a consensus, or shared morality about. From civil rights, to immigration, to criminal justice, to the environment, we can easily identify that there is a large diversity of opinion, and therefore a divergence of morality, on these subjects.

 The second kind of bias, which is personal bias, is not hard to detect, but is hard to distinguish the propriety thereof in the presence of reporting about a consensus-negative underlying thing. Here, we should again split the definitions of personal bias into the categories of “appropriate” and “inappropriate” bias.

In articles about an underlying thing and an associated person, we can define “personal bias” in an article as judgements made by the journalist/writer about the person in view of the underlying thing. “Appropriate” personal bias is a moral judgment directly tied in time and scope to the underlying thing. “Inappropriate” personal bias is a judgment not tied in time and scope to the underlying thing, such as a judgment about the person’s unrelated personal characteristics or the person’s unrelated past actions. In essence, inappropriate personal bias is when the journalist imports whether he/she already likes or dislikes the person regardless of the present underlying thing.

 For example, in a story about a drunk driver who killed someone, moral judgments that 1) the driver made bad decisions and 2) was probably indifferent or reckless as to the impact of their actions on others would constitute appropriate personal bias. However, if the reader or journalist/writer were to make judgments about the driver because of that person’s race, gender, age, or unrelated background, those judgments easily become inappropriate. Consider the reporting of that same drunk driving incident that insinuates further nefarious motive because the driver was a male illegal immigrant from Mexico, or invites sympathy because the driver was a young, white, female honor student.

The inappropriate bias that most often occurs when the reporting is about someone who is not a famous public figure is often based on ingrained cultural prejudices and stereotypes. Another type of inappropriate personal bias exists when the reporting is about a famous public figure, particularly a political one, because all individuals, including journalists, are prone to developing strong, ingrained opinions about that person over time that inform any judgment about the future underlying things they do. Consider how entrenched a reader’s or a journalist’s opinion of a public figure, such as Barack Obama or Donald Trump, becomes after the cumulative effects of all the stories written about them daily. The body of past information, whether or not that past information itself is inappropriately biased, naturally informs people’s overall personal biases when the next story comes along. A person who has read (or written) story after story which makes him like or dislike things Obama does will eventually make him like or dislike who Obama is. A person who has read (or written) story after story which make her like or dislike things Trump does will eventually make her like or dislike who Trump is. This is human nature, and this tendency makes it hard for readers and journalists/writers alike to not import personal bias into the reading or reporting of an underlying thing.

How, then, can we distinguishwhether something is actually personally biased, or whether the underlying thing is simply a consensus-negative thing, and therefore the reporting of it is not biased in and of itself?

A first way to check for personal bias is by distinguishing how a statement about an underlying thing is framed. Does it state the underlying thing as what happened or does it state what the underlying thing says about the person? For example, if an article is reporting on a statement President Trump made today which is false, the writer can state what happened in several ways.

Example 1: “Trump said x, which is contrary to y evidence” (fact statement about what happened)

Example 2: “Trump said x, which is contrary to y evidence; therefore, he lied” (fact + analysis about what happened)

Example 3: “Trump lied” (analysis/conclusion about what happened)

Example 4: “Trump said x, which is contrary to y evidence; Trump is a liar”  (stronger analysis/conclusion about what the underlying thing says about the person)

Example 5: “Trump is a liar” (conclusion about what the underlying thing says about the person.

Much contemporary criticism of the media consists of people on Twitter yelling at the NY Times to use examples 3, 4, and 5 (“just say that that he is a liar!”) instead of example 1. Each of these statements are essentially “true,” so writers have a choice of how to convey the information. However, reading through examples 1-5, most readers will sense that statements 1-3 seem less personally biased than statements 4-5; the main difference being that 1-3 are about what happened and 4-5 are about what the underlying thing says about the person. I submit that the closer a statement is to 1 than 5, the more people are likely to trust the statement and view it as unbiased, and the closer the statement is to 5, the more people are likely to distrust the statement and view it as biased.

A second way to check yourself or an article author for personal bias, and to distinguish if something is merely consensus-negative reporting or personally biased reporting is, to do a mental exercise called “what if X did it instead?” This is a common genre of Twitter argument, and it can be effective. People on Twitter, in an attempt to show that Trump did a consensus-negative thing and is not being criticized for it by his supporters because of their inappropriate pro-Trump personal bias, will say “what would you say if Hillary did this,” or “what would you say if Obama did this?” That exercise is revealing, but it’s often hard for people to perform because people who feel very strongly one way or the other about Trump usually feel very strongly in the opposite way about Hillary and Obama.

I suggest substituting the name of a random other elected official whom you don’t feel strongly about, or maybe whose political affiliation you don’t know. Sometimes a news story will help you perform this exercise accidentally. If you heard “Senator Bob Menendez went on trial for corruption,” and you don’t know who he is or whether he is a Republican or Democrat, your reaction to your opinion of that story should be informative. You may notice that at first you think to yourself “well, that sounds bad and I think he’s guilty. I hope they get to the bottom of this,” but then you find out he is a Democrat, and you start to feel a twinge of either sympathy or superiority, you can distinguish 1) that the underlying thing is a consensus-negative fact and 2) you have some personal bias because of the person’s party. If you detect the sympathy or superiority in the article itself, you can distinguish the bias in the article.

When reading a “negative” story, the more distinctly you can parse out the underlying thing from the person the underlying thing is about, the easier it is to distinguish whether the negativity stems from the reporting of a consensus-negative fact itself or from inappropriate personal bias.

A recent example of consensus-negative reporting that is minimally (appropriately) biased would be a straightforward story about Trump’s Helsinki Summit with Vladimir Putin. Most political observers, left, center, and right, were aghast that Trump publicly stated that he believed Putin and not the US intelligence community about Russian election interference. This was widely viewed as negative—so widely that it could be considered consensus-negative. I submit this is because if the underlying thing was “[powerful US elected official] sides with [authoritarian leader of adversary country] over the consensus of the US intelligence community,” that would be viewed negatively by most Americans regardless of who the powerful US elected official was.

Stories saying that this was a bad move on the part of Trump were not necessarily biased against Trump; many of his most ardent historical supporters (those having a strong pro-Trump personal bias) agreed this was bad. One notable exception was Sean Hannity’s reporting of that story, which downplayed and defended Trump’s actions, and made his pro-Trump personal bias highly apparent.

            The reason it is hard to distinguish between whether a negative story is negative because of inappropriate personal bias or because it is simply reporting about something consensus-negative is because both kinds of stories about underlying things are usually about people about whom we all have strong, pre-formed, biased opinions. It becomes hard to separate the underlying thing from our opinion of the person. Take for example, a relatively consensus-neutral underlying thing, like “______ wore fancy boots.” If the story is “Michelle Obama wore Balenciaga boots,” this instantly takes on different meaning for people who like Michelle Obama and those who don’t. Her fans will import their strong, pre-formed opinions onto the consensus-neutral thing, and think “wow, so stylish! You go ahead and rock those boots!” Her detractors will import their strong, pre-formed opinions and say “ugh, so out-of-touch and snobby.” Note that if those very statements are written by the authors of these articles, that’s a dead giveaway of personal bias; they took a consensus-neutral thing and imported bias based on the person.

Take another relatively consensus-neutral underlying thing, such as “_____decorated the White House in this way.” If the article is “Melania Trump decorated the White House with a forest of bright red trees,” people will import their strong, pre-formed opinions of Melania Trump onto the underlying thing of the tress decoration in the same way. If you are saying in your head “ugh, what an out-of-touch snob” or “wow, so avant-garde!” you are biased. If the article author is saying it, the article is biased.

The importation of personal bias to an underlying thing becomes magnified if the underlying thing isn’t a consensus-neutral, but a consensus-negative. This happens all the time with Trump because he does a lot of consensus-negative things. For example, some underlying things most Americans would consider to be consensus-negative (regardless of who did them) include

1) knowingly lying

2) knowingly lying repeatedly

3) not showing remorse when proven to have lied repeatedly

4) showing admiration for or praising authoritarian leaders of adversary countries

5) insulting one’s political opponents by name-calling and

6) saying that it is ok to just grab women by their genitals, particularly if the grabber is rich, among other examples.

The media extensively reports on these consensus-negative underlying things. The reporting of these things, just because they are about Trump, does not necessarily mean the stories are themselves personally biased. The journalists can be, and often are, personally biased against Trump. But that does not mean that every story those journalists report is personally biased.

In most reputable mainstream news sources, consensus-negative stories about Trump are typically reported in at least one instance in an unbiased way that just presents the facts. For example, if a New York Times or Washington Post story reports just the straight facts of the Michael Cohen plea story, those facts are that Michael Cohen pleaded guilty to violating campaign finance laws at the direction of President Trump. That is a negative story about Trump. It makes Trump look bad. But it is not biased.

However, the same publications, and other publications, will take that same set of facts and write their opinion and analysis takes on it, which will often add layers of personal bias in other ways. Journalists have bias, and strong pre-formed opinions of Trump, so it is easy for these journalists (especially ones who do not like Trump) to take these consensus-negative facts and pile on.

It is useful to compare stories about consensus-negative topics about Trump by conservative authors from both those who have developed a strong anti-Trump bias and those willing to give Trump the benefit of the doubt or highlight Trump’s accomplishments (i.e., those who have not shown a particular anti-Trump personal bias). From the former category, see conservative Rick Wilson’s article here and here—each are laden with personal bias on top of the facts presented. From the latter category (much less personal bias) see this article and this one in the conservative National Review, by David French, which are analyses in which the author lays out the consensus-negative facts, and as a result, comes to a conclusion that cannot be anything but negative. A similar example from conservative columnist Jonah Goldberg lays out why an underlying thing Trump did is in fact consensus-negative by using the “if Obama did it” argument.

This phenomenon, of conservatives/Republicans writing negative Trump stories is often confusing to people who are looking to those authors for confirmation bias or some political argument in support of their “side.” However, the satisfaction they may have received in the past from those authors are arguments about something about which Americans have a diverging morality, like excessive use of force by police or abortion rights. Those moral positions are not any sort of neutral; they are left and right. When a right-leaning journalist/writer comes to the same conclusions as left-leaning journalists/authors (i.e., Trump violating campaign finance laws is trouble and a big deal, or Trump siding with Putin in Helsinki was bad), that’s an indication that the topic itself is consensus-negative.

It is often unusual and confusing (to a reader) for a right-leaning writer to agree with a left-leaning writer, the right-leaning reader will often accuse the right-leaning writer of being either biased, more liberal now, or a Democrat. It’s fascinating to watch reliably conservative authors be accused of these things on Twitter. It’s not that, all of a sudden, they changed their conservative views on taxes, immigration, abortion, and guns; it’s merely that they reported that Trump did a thing that most American political observers view as objectively bad; a consensus-negative thing. Former Tea-Party Republican congressman Joe Walsh has become one of President Trump’s most vocal critics on Twitter, and regularly gets accused of being “a Democrat.” He argues back that he is still ardently committed to conservative principles, and that he opposes Trump’s actions for principled conservative and common-sense reasons. I argue that he criticizes Trump for consensus-negative underlying things, and because they are consensus-negative, he happens to agree with left-learning figures on those things.

For our purposes at Ad Fontes Media, and for your purposes as a media-literate citizen, it is critical to distinguish what makes the story you are reading “negative.” Is it because the article is reporting a consensus-negative thing? Or is it because the author is injecting inappropriate personal bias? Is it a combination of both? Distinguishing the difference gives you reliable anchors in an environment where many actors in politics and media seek to gain from confusion and discord. Ask yourself what the underlying thing is and whether it is good or bad, based on your own good judgment. Don’t throw your hands up and say, “I don’t know what to believe anymore.” You’re smart. You have the tools to figure it out.