Should I avoid all biased sources?

No. Every source has a purpose. Use a biased source for the claim it can support, name the producer in your sentence, verify facts in a second source, and do not let an advocacy brief stand in for a systematic review or an agency table.

Is a peer-reviewed journal unbiased?

No. Peer review is a screening process, not a neutrality certificate. Check the funding statement and conflicts, look for a protocol or a systematic review when the finding is surprising, and remember Lundh et al. (2017): industry-sponsored drug and device studies more often report favorable results than other-sponsored studies.

How do I tell political bias from propaganda?

Political bias is a declared or detectable point of view plus a process you can check. Propaganda blurs news and opinion, hides the producer, and will not correct a false claim. Run the CRAAP Purpose questions, then leave the page and see who funds the organization.

Can I cite the Heritage Foundation or CBPP in a research paper?

Yes, when the assignment allows grey literature and you name the organization as what it is: a mission-driven policy institute. Cite a dated report you opened — Sherk (2016) or CBPP’s 16 April 2026 primer — not “Heritage says” with no title. Pair it with an agency score such as CBO publication 55410 when the claim is a contested fact, and say they are not the same instrument.

Can I cite an opinion column?

Yes, as evidence of someone’s public argument. Label it as opinion in your prose. It should not be the only source for a factual claim that has a news report, a dataset, or a journal article behind it.

What if every source on my topic looks biased?

That is common on live policy questions. Include producers with different jobs (advocacy, agency, journal), say so in the paper, and move the weight onto checkable numbers. If there is no nonpartisan estimator, write that gap instead of inventing a neutral citation.

Are media-bias charts reliable enough to cite?

Treat them as starting maps with their own methods, not as official ratings. They disagree, they are not peer-reviewed findings about a newsroom, and they fail the same Purpose questions as any other site. Cite the article you read, not the chart’s color.

Is this the same as APA bias-free language?

No. APA 7 Chapter 5 is about inclusive wording in your prose (age, disability, race, gender). This page is about evaluating other people’s sources. You can write in bias-free language and still need to name a think tank’s mission.

Where do I start if I only have ten minutes?

Open the About or funding page, ask the five CRAAP Purpose questions, search the organization name in a new tab, and verify one number in a second source. If you cannot name the producer, do not cite the page.

How to Identify Bias in Sources: CRAAP, Cochrane, and Think Tanks [2026]

How to identify bias in sources: quick answer

Do not reject a source because it has a point of view. Name who produced it and what job the page is doing, check funding and conflicts, verify one factual claim in a second source, and pair it with a document that has a different job. A conservative think-tank brief, a poverty-policy primer, and a Congressional Budget Office score can all be citable for the claim each one actually supports. They are not interchangeable evidence.

Three sentences that keep bias visible

Name the producer. “According to the Heritage Foundation, which states that it promotes conservative public policies…”
Separate fact from frame. Use a number you can open in a table, a methods note, or a scored option. Do not import the adjective that came with it.
Pair the jobs. A mission-driven brief is evidence of that organization’s argument. A Congressional Budget Office cost estimate is evidence of a scored budgetary or employment effect. A Cochrane review is evidence of a synthesis. Do not let one stand in for the others.

What this page is (and is not)

CitationEasy already has a page for is this source credible enough to cite? (credible sources), a page for run the CRAAP questions then leave the page (evaluate sources), a page for is this claim true? (fact-check sources), and a page for journal versus magazine (scholarly vs popular). This URL answers a narrower question: how do you recognize bias, tell it apart from unreliability, and write with sources that have a declared mission? It is not a media-bias chart, and it is not APA’s bias-free language chapter — that chapter is about the words you write, not about whether a source has a point of view.

Two official meanings of “bias”

Instructors use one word for two different problems. Information-literacy handouts mean purpose and point of view. Methodologists mean systematic error. Mixing them up is how a student throws out a usable advocacy brief, or treats a peer-reviewed trial as neutral because the PDF looks official. You need both readings.

Official definitions (keep these English)

Cochrane Handbook, Chapter 7 (version 6.5, 2024; chapter last updated August 2022)

We define bias as a systematic error, or deviation from the truth, in results.

ACRL Framework for Information Literacy for Higher Education (2016)

Experts understand the need to determine the validity of the information created by different authorities and to acknowledge biases that privilege some sources of authority over others, especially in terms of others’ worldviews, gender, sexual orientation, and cultural orientations.

The Cochrane sentence is about whether a study over- or under-estimates an effect. The ACRL sentence is about whose voices get treated as authoritative. A well-run randomized trial can still be industry-sponsored. A primary source can be indispensable and one-sided. Credibility is a judgment about fit; bias is part of that judgment, not a substitute for it. Cochrane Handbook version 6.5.1 (2025–2026) corrected later chapters; it did not rewrite Chapter 7.

Bias is not imprecision, and “at risk of bias” is not “this study is false”

Students often treat a small sample, a wide confidence interval, or an industry funder as the same red flag. Cochrane Chapter 7 separates those problems. A small trial can be precise-poor and still low-risk in its design. A large trial can be precise and still high-risk if the design was built to lean. The Handbook also refuses the cartoon of calling a paper “biased” as a fact about the universe:

Official Cochrane wording on bias versus imprecision

Chapter 7, §7.1.1

Bias should not be confused with imprecision. Bias refers to systematic error, meaning that multiple replications of the same study would reach the wrong answer on average.

Chapter 7, §7.1.1

it is more appropriate to consider whether a result is at risk of bias rather than claiming with certainty that it is biased.

That is the student-paper version too. Write “this result is at risk of sponsorship bias” or “this brief argues a conservative policy case,” not “this PDF is fake.” For medical source types after you have named the risk, use the medical citation guide.

Quick tests: CRAAP Purpose and SIFT

Librarian Sarah Blakeslee’s CRAAP handout (Meriam Library, 17 September 2010) treats Purpose as “the reason the information exists.” Those questions are the fastest bias screen that still comes from a library, not a rating company:

Official Purpose questions (Meriam Library)

Official gloss

Purpose: The reason the information exists.

What is the purpose of the information? Is it to inform, teach, sell, entertain or persuade?
Do the authors/sponsors make their intentions or purpose clear?
Is the information fact, opinion or propaganda?
Does the point of view appear objective and impartial?
Are there political, ideological, cultural, religious, institutional or personal biases?

A checklist you run only inside the page will miss a professional-looking front group. Mike Caulfield’s SIFT moves — Stop, Investigate the source, Find better coverage, Trace claims to the original context — take you off the page. His Web Literacy for Student Fact-Checkers is also the place that separates slant from process:

Mike Caulfield, Web Literacy for Student Fact-Checkers

Note that reputation and process might be apart from issues of bias: the New York Times is thought by many to have a center-left bias, the Wall Street Journal a center-right bias, and USA Today a centrist bias. Yet fact-checkers of all political stripes are happy to be able to track a fact down to one of these publications since they have reputations for a high degree of accuracy, and issue corrections when they get facts wrong.

That is the rule this page keeps coming back to. Bias is not the same as a missing corrections policy. A newsroom with a slant and a corrections desk is a different object from a page that never names an author and never fixes a number. For the Stanford evidence that students stay on the page and get trapped by lookalike organizations, use the credibility guide.

Same topic, three different jobs: a $15 minimum wage

Think tanks and agencies usually tell you their job on the About page. Read that page before you treat the PDF as a neutral brief. The Heritage Foundation mission is “to formulate and promote conservative public policies based on the principles of free enterprise, limited government, individual freedom, traditional American values, and a strong national defense.” The Center on Budget and Policy Priorities describes itself as “a nonpartisan research and policy institute” and says: “We combine rigorous research and analysis, strategic communications, and effective advocacy.” The Congressional Budget Office overview states the agency’s statutory job in two sentences you can quote:

Official CBO wording (Introduction to CBO / About overview)

CBO About / Introduction to CBO

CBO conducts objective, impartial analysis and hires employees without regard to political affiliation. The agency does not make policy recommendations.

An Introduction to the Congressional Budget Office

CBO makes no policy recommendations because choices about public policy inevitably involve value judgments that the agency does not and should not make.

Those jobs produce different documents even when the headline topic is the same. An older version of this page invented “Smith, 2025,” “Johnson, 2025,” and a fake “CBO, 2025” paper to stage a debate. Do not do that. Open titled reports. And do not flatten three different instruments into one fake head-to-head: James Sherk’s Heritage brief (26 July 2016) models a $15 floor by 2021 under the Pay Workers a Living Wage Act; CBO publication 55410 (8 July 2019) scores $10, $12, or $15 by 2025; CBPP’s Policy Basics page (16 April 2026) is a primer on the current $7.25 federal floor and state laws, not a score of either bill.

Three documents you can actually open

Heritage brief (advocacy research). James Sherk, “Raising Minimum Starting Wages to $15 per Hour Would Eliminate Seven Million Jobs” (26 July 2016). The brief argues that a $15 federal floor by 2021 would eliminate “approximately seven million full-time-equivalent (FTE) jobs.” Cite it as Heritage’s argument. Do not call it nonpartisan, and do not treat 2016’s 2021 horizon as CBO’s 2025 options.
CBPP primer (research plus advocacy). Center on Budget and Policy Priorities, “The Minimum Wage” (Policy Basics, 16 April 2026). The page states that the federal minimum remains $7.25, that as of March 2026 thirty states and the District of Columbia have higher floors (nineteen at $15.00 or above), and that BLS counted 843,000 workers paid at or below the federal minimum in 2024. Cite the BLS number you can open; name CBPP as the producer that also does advocacy.
CBO report (scored options, no recommendation). Congressional Budget Office, The Effects on Employment and Family Income of Increasing the Federal Minimum Wage (publication 55410, 8 July 2019). CBO examines $10, $12, or $15 per hour by 2025 and states two principal effects: most low-wage workers would see higher earnings, and some would become jobless. Still check the assumptions in the report you opened.

Worked citations (APA 7) for those three documents

Heritage (Sherk 2016)

Sherk, J. (2016, July 26). Raising minimum starting wages to $15 per hour would eliminate seven million jobs. The Heritage Foundation. https://www.heritage.org/jobs-and-labor/report/raising-minimum-starting-wages-15-hour-would-eliminate-seven-million-jobs

CBPP (Policy Basics, 16 April 2026)

Center on Budget and Policy Priorities. (2026, April 16). The minimum wage. https://www.cbpp.org/research/policy-basics-the-minimum-wage

CBO (publication 55410, 8 July 2019)

Congressional Budget Office. (2019, July 8). The effects on employment and family income of increasing the federal minimum wage. https://www.cbo.gov/publication/55410

Prose that names the job (use a document you opened)

According to the Heritage Foundation, which states that it promotes conservative public policies, a $15 federal starting wage by 2021 would eliminate approximately seven million FTE jobs (Sherk, 2016). The Congressional Budget Office, which states that it makes no policy recommendations, examined $10, $12, and $15 options by 2025 and reported that a higher federal floor would raise earnings for most low-wage workers and make some of them jobless (Congressional Budget Office, 2019). CBPP’s April 2026 primer, from an institute that also describes its work as advocacy, reports the current $7.25 federal floor and the BLS 2024 count of workers paid at or below that floor.

Commercial and industry bias

Financial interest is the bias students under-weight because the PDF still has a methods section. Andreas Lundh, Joel Lexchin, Barbara Mintzes, Jeppe Bennekou Schroll, and Lisa Bero’s Cochrane methodology review “Industry sponsorship and research outcome” (2017, Article MR000033) compared drug and device studies sponsored by industry with studies that had other sponsors. Across 75 included papers, industry-sponsored studies more often had favorable efficacy results (risk ratio 1.27, 95% confidence interval 1.17 to 1.37) and more often had favorable conclusions (risk ratio 1.34, 95% confidence interval 1.19 to 1.51). The authors concluded that the industry association “cannot be explained by standard ‘Risk of bias’ assessments.” The review is open at DOI 10.1002/14651858.MR000033.pub3; the PubMed abstract is the text opened for those numbers.

Worked citation (APA 7) for a real sponsorship review

Reference list

Lundh, A., Lexchin, J., Mintzes, B., Schroll, J. B., & Bero, L. (2017). Industry sponsorship and research outcome. Cochrane Database of Systematic Reviews, 2017(2), Article MR000033. https://doi.org/10.1002/14651858.MR000033.pub3

In prose

A Cochrane methodology review of 75 papers found that industry-sponsored drug and device studies more often reported favorable efficacy results than studies with other sponsors (Lundh et al., 2017). That is a finding about sponsorship, not a license to discard every industry trial or to treat every non-industry trial as clean.

Disclosure is how readers judge that risk. The ICMJE Recommendations (the January 2026 text, opened on the ACP mirror of icmje.org) put the conflict in two sentences:

ICMJE Recommendations, conflicts of interest

The potential for conflict of interest and bias exists when professional judgment concerning a primary interest (such as patients' welfare or the validity of research) may be influenced by a secondary interest (such as financial gain).

Perceptions of conflict of interest are as important as actual conflicts of interest.

A disclosed sponsor does not make the paper unusable. An undisclosed relationship, a sponsor that controlled the analysis, or a paper that never says who paid is a different problem. Read the funding statement and the conflict form, then decide what claim the paper can carry.

Publication bias, non-reporting bias, and the file drawer

Some bias never appears on the page you found, because the contradictory study was never published. Cochrane’s current Handbook still uses the classic definition in Chapter 7 and, in the 2024 update to Chapter 13 (last updated August 2024), names the ROB-ME tool for assessing missing evidence in a meta-analysis:

Official publication-bias and ROB-ME wording

Cochrane Handbook, Chapter 7, Table of reporting biases

The publication or non-publication of research findings, depending on the nature and direction of the results.

Cochrane Handbook, Chapter 13 (version 6.5, 2024)

ROB-ME is the first structured approach for assessing the risk of bias that arises when entire studies, or particular results within studies, are missing from a meta-analysis because of the P value, magnitude or direction of the study results (Page et al 2023).

John P. A. Ioannidis, PLoS Medicine (30 August 2005)

Moreover, for many current scientific fields, claimed research findings may often be simply accurate measures of the prevailing bias.

Worked citation (APA 7) for Ioannidis 2005

Ioannidis, J. P. A. (2005). Why most published research findings are false. PLoS Medicine, 2(8), Article e124. https://doi.org/10.1371/journal.pmed.0020124

Ioannidis’s essay is a real, open-access, peer-reviewed argument with a 2022 correction. It is a model of power, bias, and how many teams chase the same question — not a census of every paper, and not a license to ignore journals. Chapter 13’s worked case is more concrete: when unreported results were added to a systematic review of oseltamivir (Tamiflu), the drug was not shown to reduce hospital admissions and the harm profile changed (Jefferson et al. 2014, as summarized in the Handbook). The practical move for a student paper is smaller: look for a registered protocol, a review that searched trial registries, or a replication, instead of treating the first significant abstract as the literature. Cochrane, the Campbell Collaboration in the social sciences, and the Annual Review series are places that synthesis already happened.

Confirmation, selection, and cultural bias

These are the biases that live in how the evidence was assembled, not only in the About page. They are still real when the author has a Ph.D. and the journal is refereed.

Confirmation bias. Searching only for studies that already agree with the thesis. Caulfield’s textbook warning is the practical fix: decide whose expertise you would respect before you search, or the results will flatter you.
Selection / sampling bias. A survey of one campus presented as “adults”; a case series of successes with the failures left out; a convenient sample treated as a population.
Cultural and language bias. English-indexed journals over-represent some countries and methods. The ACRL frame above is the literacy version of that problem: some authorities get privileged. Include work from other regions when the question is not U.S.-only.
Methodology bias. Leading survey items, an underpowered sample, an outcome chosen after the results arrived. Cochrane’s Chapter 7 is the place that separates this risk of bias from mere smallness (imprecision).

How to use a biased source without laundering it

A biased source is often the right source for a perspective claim. Historical primary sources are biased by definition — they are someone’s record — and they are still required in history. An advocacy brief is the right source for “what this organization argues.” A grey-literature agency report is often the only public method for a statistic. The failure is treating any of those as the last word on a contested fact.

Use it / do not launder it

Accept for perspective. “Heritage argues that…,” “CBPP contends that…,” “In a letter you actually opened, …” Name the producer in the sentence, not only in the parentheses.
Accept for a checkable fact. A table, a statute, a methods note, a DOI you opened. Quote the number, not the slogan.
Balance with a different job. Pair an advocacy brief with an agency score, a systematic review, or a paper from another school of thought. Pairing two blogs that already agree is not balance. Pairing Sherk (2016) with CBO (2019) is closer, and you still have to say they scored different calendars.
Do not launder. Do not write “research shows” when the only source is a funded brief. Do not move a think-tank claim into your voice and cite a journal you did not read.
Do not invent the other side. If you cannot open a real counterpart, say so, or narrow the claim.

When a position is required

Academic writing is not a demand that you have no thesis. An argument paper takes a position, names the counterargument, and shows the evidence. Neutrality is the wrong goal when the literature itself has a consensus — climate attribution, vaccine effectiveness, or a Cochrane review of a drug. False balance (treating a fringe claim as equal to a body of trials) is a bias of your page, not a virtue. The ACRL frame “Scholarship as Conversation” is the better picture: you enter a conversation that already has terms, not a blank scale with two equal pans.

News: process versus slant

News sources have slants. They also have, or lack, a process. Caulfield’s distinction above is the one to keep. A news report that separates news from opinion, names sources, and runs corrections can still be used for a dated public fact. An item that never marks opinion, never corrects, and exists to recruit is a different genre. Fact-checking is the claim-level method; this section is only the bias screen.

Process signals, not a political rating
Signal of a usable news process	Signal that slant has become unreliability
News and opinion are labelled as such	Opinion is presented as a news brief
A public corrections policy you can find	Errors sit uncorrected or silently edited
Named sources you can follow, or a stated reason for anonymity	A single anonymous source carrying the whole claim
More than one involved party is quoted on a contested fact	The other side appears only as a caricature
An editor or standards page you can open	No accountable masthead

Commercial media-bias charts (Ad Fontes, AllSides, Media Bias/Fact Check, NewsGuard) are themselves sources with a method and a business model. They can be a starting map. They are not official manuals, they disagree with one another, and they should get the same Purpose questions you just ran on the article. Do not cite a chart as if it were a peer-reviewed finding about a newspaper.

Academic publishing is not bias-free

Peer review is a process, not a purity stamp. The same literature that teaches you to trust journals also documents how they fail:

Funding and industry bias. Lundh et al. (2017) is the Cochrane synthesis; the ICMJE sentences are the disclosure rule.
Publication and citation bias. Positive results appear more often and get cited more often. Cochrane Chapter 7 lists time-lag, language, citation, duplicate-publication, and outcome-reporting bias as cousins of the file drawer. Chapter 13’s ROB-ME tool is how reviewers now score missing evidence in a meta-analysis.
Language and prestige bias. English-language and high-impact venues crowd out other work. That is an ACRL “who counts as an authority” problem as much as a statistics problem.
Your own citation list. Preferentially citing supportive studies is how a literature review becomes a brief. The literature-review guide is the place that fights that habit.

Your own bias is part of the method

The source is not the only biased object in the room. The searches you run, the first PDF you trust, and the counterexample you never open are part of the record. Common names for that problem:

Biases that show up in the search box

Confirmation. Querying only the phrase that already matches the thesis.
Anchoring. Treating the first article as the frame for every later one.
Availability. Over-weighting the study you can remember or that your library loaded instantly.
Authority / in-group. Trusting a famous lab, a familiar newspaper, or “people like us” without a lateral search.

Write the claim you want to be true, then search for the best source that would falsify it.
Decide the expert venue (a statistical agency, a Cochrane review, a methods journal) before you Google the slogan.
Read the methods or the About page before you harvest a sentence.
Ask someone who disagrees with you to mark the places you treated a brief as a finding.

When bias becomes unreliability

Bias becomes a reason to reject the source for this claim when it stops being a perspective and starts being a broken process. That is the same three-way decision as the credibility page, restated for this problem:

Reject for this claim when

Facts are fabricated, or a number you opened does not match the number in the prose.
Contrary evidence is suppressed and the methods cannot be checked.
A conflict of interest that would matter is not disclosed.
The design was built to guarantee the desired result (unblinded outcome, incomparable doses, outcome switched after looking).
Opposing views appear only as straw men, or the page is there to sell, recruit, or inflame.
You cannot tell who is accountable.

Frequently asked questions

Should I avoid all biased sources?: No. Every source has a purpose. Use a biased source for the claim it can support, name the producer in your sentence, verify facts in a second source, and do not let an advocacy brief stand in for a systematic review or an agency table.
Is a peer-reviewed journal unbiased?: No. Peer review is a screening process, not a neutrality certificate. Check the funding statement and conflicts, look for a protocol or a systematic review when the finding is surprising, and remember Lundh et al. (2017): industry-sponsored drug and device studies more often report favorable results than other-sponsored studies.
How do I tell political bias from propaganda?: Political bias is a declared or detectable point of view plus a process you can check. Propaganda blurs news and opinion, hides the producer, and will not correct a false claim. Run the CRAAP Purpose questions, then leave the page and see who funds the organization.
Can I cite the Heritage Foundation or CBPP in a research paper?: Yes, when the assignment allows grey literature and you name the organization as what it is: a mission-driven policy institute. Cite a dated report you opened — Sherk (2016) or CBPP’s 16 April 2026 primer — not “Heritage says” with no title. Pair it with an agency score such as CBO publication 55410 when the claim is a contested fact, and say they are not the same instrument.
Can I cite an opinion column?: Yes, as evidence of someone’s public argument. Label it as opinion in your prose. It should not be the only source for a factual claim that has a news report, a dataset, or a journal article behind it.
What if every source on my topic looks biased?: That is common on live policy questions. Include producers with different jobs (advocacy, agency, journal), say so in the paper, and move the weight onto checkable numbers. If there is no nonpartisan estimator, write that gap instead of inventing a neutral citation.
How do I cite a source while noting its bias?: The reference list entry does not change. The sentence does: “According to the Heritage Foundation, which states that it promotes conservative public policies, …” Then give the normal APA, MLA, or Chicago citation for the report you opened.
Are media-bias charts reliable enough to cite?: Treat them as starting maps with their own methods, not as official ratings. They disagree, they are not peer-reviewed findings about a newsroom, and they fail the same Purpose questions as any other site. Cite the article you read, not the chart’s color.
Is this the same as APA bias-free language?: No. APA 7 Chapter 5 is about inclusive wording in your prose (age, disability, race, gender). This page is about evaluating other people’s sources. You can write in bias-free language and still need to name a think tank’s mission.
Where do I start if I only have ten minutes?: Open the About or funding page, ask the five CRAAP Purpose questions, search the organization name in a new tab, and verify one number in a second source. If you cannot name the producer, do not cite the page.

Finish the evaluation, then cite what you opened

Bias analysis does not format the reference. Once you know what job the source is doing, cite the report, article, or table you actually read — Sherk (2016), CBPP (2026), CBO (2019), Lundh et al. (2017) — not a slogan. The generator will not decide whether the source was usable.

Open the citation generator Run the CRAAP evaluation Decide if the source is credible