Ethics Policy

“A journalism based on scientific method leaves a trail where error can be detected and truth verified.”

Philip Meyer, “Precision Journalism”

The Markup is a nonprofit newsroom that investigates how powerful institutions are using technology to change our society. We are a new kind of media organization, staffed with an unparalleled roster of quantitative journalists who pursue meaningful data-driven investigations. Our approach is scientific: We build datasets from scratch, bulletproof our reporting, and show our work. We call this The Markup Method. 

We strive to be accurate and fair in our reporting on all platforms. We aim to be transparent about our evidence and our sources, and provide supporting data and documentation whenever possible. We endeavor to provide context for the data we present. Our goal is for our work to represent our best independent judgment. Here are our policies on accuracy and corrections:

• We are truthful about the sources of our information. We seek to attribute information or images to their original source if we did not collect it ourselves. We do not tolerate plagiarism or fabrication.
• We seek to be fair. We will make a significant effort to obtain a response from people and institutions that we write about, and we will explain in the article what efforts we made to seek comment.
• When we learn that we have made a mistake, we will quickly seek to correct it in the place that the incorrect material originally appeared. We will make it easy for readers to contact us to seek corrections.
• We are transparent about the updates, corrections, clarifications, or other changes we have made to a story or a dataset after publication—even if those changes occur just a few minutes or hours after publication.

Editorial decisions at The Markup will not be influenced by donors, sources, or other outside forces. Here are our policies on impartiality:

• Donors, members of the board of directors, and others outside The Markup’s editorial process are not allowed to see our stories before publication.
• We always pay our own way when traveling for editorial purposes.
• Our journalists may not buy or sell individual securities in companies that they cover regularly.
• We do not pay for interviews.
• We believe in transparency. We will disclose within an article anything that we think might be perceived as a conflict of interest—whether it be a mention of a donor’s project or a journalist with a history of advocacy on a particular issue.

Gathering large amounts of data, including through automated means, is a centerpiece of our work. We analyze data to help us find and tell stories that will help people better understand their world. We are only interested in the truth. We don’t cherry-pick from the data we have or exaggerate the strength of our findings where they are weak or uncertain.

• We make serious efforts to protect data from unauthorized access, tampering, accidental disclosure, or other harm or actions while we are collecting and storing it. We aim to keep data safe, secure, and private.
• We make serious efforts to check and verify the data that we have collected. If we are unable to verify it, we are transparent about that.
• When we publish findings, we aim to be transparent about our assessment of their strength, reliability and limitations. We acknowledge the assumptions we have made in our analysis, including any appropriate disclaimers and (if used) weighting.
• When we publish data, we strive to be forthcoming about its source and how it was compiled, cleaned and analyzed. We aim to give some assessment of our evaluation of its overall quality and validity. We clearly and fully report the steps we have taken to preserve data integrity.
• Whenever possible, we aim to publish all the data underlying our stories so that others can read, evaluate, and reuse it. We aim to balance the value of making it public with individual privacy or safety considerations, and/or by our need to protect sources.
• We use accepted statistical analysis techniques and prefer techniques that are well-known and time-tested; we do not p-hack.
• We aim for transparency and reproducibility in our findings.
• When we publish personally identifiable or other sensitive personal information, we aim to balance the public interest against the legitimate expectation of privacy by individuals. Whenever possible, we strip out identifiers that we don’t need.
• When we publish personally identifiable data or other sensitive personal information, we strive to make it clear how individuals can request corrections of any inaccuracies, and we specify how long we will retain and publish the data and when we will delete and dispose of it.
• We aim for our tables and graphics to accurately reflect the data they contain and to be clear and simple. We generally only use actual data. In instances in which we need to project, surmise, or estimate, we label that clearly.
• If we discover postpublication that we have made mistakes in our analysis, or there have been other problems that disprove our findings or seriously call them into question, we publicly correct the record and disseminate the correction publicly.

We both investigate artificial intelligence and use artificial intelligence to investigate. When we use AI in our journalism, we use it in pursuit of the truth, and when using AI tools, we apply the same transparency, accuracy, and ethical guidelines as we do to the rest of our work.

• Our stories are hand-crafted; we do not publish AI-written stories. If we publish excerpts of text generated by AI because that is the point of the story, we will clearly label in the story what text has been written by AI. 
• We do not use AI to generate our artwork. If we publish imagery generated by AI because that is the point of the story, we will clearly label what art has been generated by AI.
• When we use AI in our published journalism, we disclose it clearly.
• The full rigor behind our Data Ethics policy applies to all data or outputs we collect or generate using AI.
• As AI evolves, we will continue to evaluate the privacy, security and ethical implications of existing and emerging tools before using them in our journalism.

The Markup accepts only on-the-record statements from employees who are speaking on behalf of a company or agency. On the record means a statement can be reported on and attributed to a specific person by name. This means: 

• We will state this on-the-record requirement upfront, in an initial conversation with or email to a source or when requesting comment from a company or agency in a story. 
• We only publish statements that are attributable to a specific person, not the company or agency as a whole. If statements cannot be attributed to a spokesperson or representative by name, The Markup will not use the statement.
• If a spokesperson asks that the conversation be “off the record” or “on background” or similar, we will state that it’s against The Markup’s policy.
• If a spokesperson sends us a written statement that they say is off the record, even though we have no such agreement with the source, we will inform them that we consider the statement to be on the record since we have no such agreement and will be attributing it to them.

Read more about why we believe this policy can lead to better and more precise responses from spokespeople.

---

Our ethics policy borrows from the policies set by other organizations including The Texas Tribune, ProPublica, The Washington Post, the American Statistical Association ethical guidelines and the Data Science Association code of professional conduct.