Ballotpedia:How we categorize bias
Ballotpedia's mission is to inform people about politics by providing accurate and objective information about politics at all levels of government. Only by remaining neutral can we earn the trust of voters across the country and across the political spectrum. We believe that voters should have the information they need to cast votes aligned with their beliefs.
To this end, we have made it our goal to become experts in the field of journalistic neutrality. Throughout the course of our research, we have identified and labeled 25 unique types of bias.[1] In practice, these types of bias may overlap. But by dividing bias into distinct categories and creating a common language around them, we have found it easier to discuss and prevent neutrality concerns in our coverage.
Types of bias
As of January 2022, Ballotpedia had identified 25 unique types of bias. Note that this list evolves over time as we discover new forms of bias. For more information about our bias documentation, email us at editor@ballotpedia.org. The following list is in alphabetical order:
- Changes to the status quo
- Cherry-picking
- Coatrack bias
- Editorializing
- Explicit recommendation
- Factive verbs
- Failure to capture competing narratives
- Ideological labeling
- Inappropriate descriptors
- Labeling bias
- Loaded language
- Overreach in partisan affiliation
- Parroting the news
- Photo selection bias
- Placement bias
- Prediction bias
- Protagonist selection
- Scare quotes
- Source selection bias
- Speech tribe terminology
- Star-maker bias
- Story selection bias
- Unclear attribution
- Unsubstantiated news
- Weasel words
Changes to the status quo
Ballotpedia must be careful when discussing changes to the status quo to ensure that we haven’t inserted ourselves into a debate by highlighting proposals that weren’t being prominently discussed before we wrote about them. This can make it appear as though we are attempting to make the proposed change more popular or legitimate, and thus, taking sides.
Cherry-picking
Cherry-picking occurs when a writer, intentionally or unintentionally, selects evidence or facts that confirm one point of view while ignoring other evidence or facts that present a case against that point of view.
Coatrack bias
Coatrack bias occurs when a subtopic of an article becomes the main focus of the page. The ostensible topic (the coat rack) is hidden by less important subtopics (coats). This scenario may introduce bias if the subtopics provide a misleading association with the main topic or if they associate the main topic with an undue positive or negative connotation.
Editorializing
Editorializing consists of language that indicates the writer’s views on a subject while communicating news or facts.
Explicit recommendation
Explicit recommendation is the most obvious form of bias. It occurs when a writer directly encourages the reader to take some action, such as voting for a particular candidate, supporting a particular ballot measure, or opposing a particular policy.
Factive verbs
Factive verbs are verbs of attribution that presuppose the truth of the claim they try to make. For instance, a person can state or write something that isn't true. But a person cannot note, affirm, or realize an untrue statement. Consequently, using those words implies that the statement is true. Factive verbs are acceptable to use when discussing verifiable facts, but they should be avoided when discussing opinions.
Failure to capture competing narratives
For any contested issue, there are always at least two competing narratives. If we fail to provide a description of both narratives, then we are biased by way of omission. If we don’t acknowledge a side of the story, then readers will naturally assume that we have, in fact, taken a different side.
Ideological labeling
Ideological labels such as conservative, liberal, and progressive must be used with care to avoid bias. In some cases, there may be disagreement about which label to use in describing a specific individual or group. In these cases, we should note the disagreement and put the associated labels into context. We also avoid giving labels based on association. For example, it is not necessarily true that a reporter working for a conservative news outlet must also be conservative.
Inappropriate descriptors
Many adverbs and adjectives have a tendency to introduce bias, wordiness, and unclear writing. For example, it is not Ballotpedia's place to say whether a piece of legislation is significant or whether an incumbent overwhelmingly defeated her opponent. Instead, we give the facts and let readers draw their own conclusions.
Labeling bias
Labeling bias can take place when a writer uses a label—such as expert or extremist—that carries a positive or negative connotation. It can also occur if individuals are labeled unevenly, such as if individuals from one political party are given more extreme-sounding labels than individuals from the opposing party.
Loaded language
Loaded language carries an emotional connotation—positive or negative. A loaded term communicates something beyond its literal meaning; it communicates how the author feels about the subject in question. This may be perceived as trying to sway the reader’s opinion through appeal to emotion.
Overreach in partisan affiliation
We use party labels (Democrat, Republican, Green, Libertarian, etc.) to refer to candidates who run for election with a specific party or officials who are affiliated with a party. In cases where a person's partisan affiliation is less direct, we need to give additional context to avoid bias. Click here for more information.
Parroting the news
Ballotpedia faces a unique challenge because we do not have reporters on the ground collecting facts. Instead, we get our information from existing government and media sources. This means that if the sources we use contain bias, we could unintentionally mirror that bias in our articles. To avoid parroting the news, Ballotpedia cites sources that are widely trusted across the political spectrum, and we diligently research and lay out competing narratives in any debate.
Photo selection bias
Just as words can carry a positive or negative connotation (see loaded language), photos can also provoke positive or negative responses. For example, unflattering photos of a candidate may make that person look silly or menacing.
Placement bias
Placement bias occurs when a writer inappropriately highlights or obscures information through where they place it in the article. For instance, if a fact that is relevant to a current election is placed near the bottom of the page and far from the section about that election, it could appear as though Ballotpedia is hiding that fact from readers.
Prediction bias
Prediction bias occurs when the use of predictive wording, such as, “if event X happens, then Y will result,” implies that one person or political party has more momentum or energy than another.
Protagonist selection
Protagonist selection occurs when the tone and direction of our writing indirectly imply to readers that there is a protagonist in the story, usually at the expense of other candidates or officials.
Scare quotes
Quotation marks may be used to distance a writer from a certain term, as if to say, “These are someone else's words, not mine.” However, using quotes around short phrases or around one word can appear sarcastic or skeptical. These are called scare quotes. When the quotation marks could be replaced by the phrase so-called, they are likely scare quotes and should be removed.
Source selection bias
Source selection bias occurs when a writer or publication relies disproportionately on sources that align with one particular party or ideology. This can cause bias in various ways:
- If readers look at an article's sources and see that they predominantly consist of Democratic- or Republican-leaning sources, they will assume that Ballotpedia aligns with that particular ideology.
- The disproportionate use of Democratic- or Republican-leaning sources can lead us to cover specific talking points from one party without covering counterpoints from the opposing side. This can give the appearance of cherry-picking.
- Relying more heavily on sources from one party can lead us to cover stories that are discussed more by those individuals while failing to cover stories discussed by their political opponents. This can lead to story selection bias.
Speech tribe terminology
Speech tribe terms are words or phrases used predominantly by one particular group of people. People from different parties or ideologies often use different words or phrases to talk about the same thing. For example, one group may use the term illegal aliens while an opposing group may use the term undocumented immigrants. Ballotpedia avoids speech tribe terms in order to avoid being associated with any specific group.
Star-maker bias
Star-maker bias occurs when coverage on Ballotpedia repeats talking points found in media sources about the importance of a particular candidate or influencer rather than remaining grounded in the facts. For example, a candidate who is not a frontrunner in a race may receive undue attention from the media if he or she has a unique story to tell. We at Ballotpedia must recognize this undue attention and avoid parroting it in our own articles.
Story selection bias
Story selection bias occurs when a writer covers stories that favor one individual, party, or position while failing to cover stories that make that position look bad or favor the opposing side. This can occur within an article or across a series of articles.
Unclear attribution
Unclear attribution occurs when a writer quotes a selection of text without drawing a clear line from the quote to its source. A writer’s failure to properly attribute a quote to its source can imply that the author stands by the quoted material as fact.
Unsubstantiated news
Unsubstantiated news refers to stories that are circulated on social media or by media outlets without sufficient evidence or specificity. These stories often take the form of rumors, hearsay, allegations, or threats of a lawsuit when no legal action has taken place.
Weasel words
A weasel word makes an ambiguous claim that is not common knowledge without providing evidence. These phrases may introduce bias by alienating readers who disagree with the claim in question. Weasel word examples include many experts say and research has shown. Specificity and proper sourcing of evidence should be used in the place of weasel words.
Footnotes
- ↑ Some of these terms were adapted from Wikipedia's neutral point of view policy
