Ask Someone to Check Your Work · Method

Purpose

By the time you finish a piece of research, you know what every sentence is supposed to say, so you read what you meant rather than what you wrote. You also want it to be right, which makes you a poor judge of whether it is. A second reader doesn’t have either problem.

Publications that take accuracy seriously build this in. Science journalist Brooke Borel describes three models in the Knight Science Journalism Program’s science editing handbook. In the magazine model, a fact-checker who is not the writer or editor checks every fact in a story, along with its overall argument, often going back through the writer’s sources and calling people again. In the newspaper model, reporters check their own facts, with spot checks by editors. Hybrid newsrooms use the magazine model for long, complex pieces and the newspaper model for short news.

Even among professionals, a dedicated checker is not the norm. A 2018 report led by Borel for MIT’s Knight Science Journalism Program surveyed and interviewed people at science publications: 34% of outlets gave fact-checking to designated fact-checkers, 15% to copy editors, and the rest to writers and editors.

You probably don’t have a checker on staff. But you can borrow the method: give a reader your claims and your sources, and ask them to try to prove you wrong.

When you need it

You’re about to publish, submit or present something that makes factual claims.
The work names real people or organizations, or says something about them they might dispute.
It includes numbers, quotes or summaries of studies.
You’ve worked on it so long you can no longer see it fresh.
The conclusion is one you hoped to reach.

How to do it

1Prepare an annotated copy

Make a version of your draft where every factual claim points to its backup: a footnote or comment with the source, the page or timestamp, and a link (archived where possible). For interviews, give the recording file and the time. The KSJ handbook describes writers giving checkers exactly this kind of annotated draft, plus the source files. Doing it yourself is half the check: claims you can’t annotate are claims you haven’t supported.

2Choose the right reader

Pick someone careful who didn’t help write it. A colleague in your field catches errors of substance. A smart outsider catches jargon, gaps and leaps you’ve stopped noticing. If you can, choose someone likely to disagree with your conclusion. They will look harder.

3Ask them to try to break it

Don’t ask “Does this look okay?” Give them a specific job:

Check the key claims against the sources I cite. Does each source say what I say it says?
Check every name, title, date and number.
Check each quote against its original.
Tell me where the argument goes further than the evidence.
Tell me what a critic would say first.

If time is short, ask them to check the three claims the whole piece depends on.

4Check the things checkers check

Whether or not you have a helper, go line by line. The KSJ handbook lists the details that trip people up most: spellings of names and places, titles and affiliations, ages, dates, numbers and statistics, and locations. Then check the characterizations, too: words like “most,” “first,” “only,” “proved” and “rejected” are claims of their own.

5Respond to every note

Treat each comment as a question to answer, not a judgment to defend against. For each one: fix it, explain why it isn’t an error (and add the source or wording that makes that clear to the next reader), or soften the claim to match the evidence. If a note shows that a central claim doesn’t hold up, rethink the piece rather than patching the sentence.

6Decide in advance what goes back to sources

Some writers read quotes or facts back to the people they interviewed to confirm them; others never do. The 2018 KSJ report found no consensus on sharing unpublished drafts or quotes with sources: some outlets ban it, others allow it. If you do check with a source, confirm facts and quotes, not your conclusions, and don’t promise them approval.

Try it

Your task

Be the checker. The Knight Science Journalism Program’s news post “Checking in on Fact Checking in Science Journalism” (September 12, 2018, on ksj.mit.edu) summarizes the 2018 report The State of Fact-Checking in Science Journalism, and links to the report as a PDF. Check three claims in the post against the report itself:

About a third of the publications employ independent fact-checkers.
About 15% rely on copy editors for fact-checking.
On average, fact-checkers in the study earned about $30 an hour.

For each, find the passage in the report and decide: supported, partly supported, or not supported.

Show a hint

Start with the report’s executive summary, then look in the results section on outlets (“Who fact checks?”) and the section on fact-checkers’ pay.

Show the answer and the path

The report is Borel et al., The State of Fact-Checking in Science Journalism (June 2018), hosted by the Gordon and Betty Moore Foundation, which funded it.

Supported. The executive summary and the “Who fact checks?” section both say 34 percent of outlets assign fact-checking to designated fact-checkers. The post’s “independent” is a slight shift from the report’s “designated,” but the number matches.
Supported. The same passages say 15 percent assign it to copy editors.
Partly supported. The report has two different pay figures from two different groups. Of 16 editors who answered questions about pay, rates ran from $15 to $50 an hour, with an average of $27.76 and a median of $27.20. The fact-checkers the team interviewed reported $19.28 to $75 an hour, with an average of $34.27 and a median of $30. The executive summary puts the average “between $27.76 and $34.27.” So “about $30” is the fact-checkers’ median, not their average, and the post’s “$15” and “$19” come from the two different groups.

What it shows: even a careful summary by the report’s own program blurs details. The headline numbers held up; the one that combined two groups needed a second look. That’s what a checker is for.

Common mistakes

Asking a friend who’ll say it’s great. Choose someone willing to tell you it’s wrong, and tell them that’s the job.
Handing over the draft without the sources. Then the reader can only check whether it sounds right, not whether it is.
Asking too late. A check the night before a deadline can catch typos, not a flawed argument.
Arguing with every note. If a careful reader misread you, others will too. Clarify the text even when the reader was wrong.
Fixing the sentence but not the pattern. One wrong date means checking every date.
Letting a source rewrite your conclusions. Confirm facts with sources; the judgment stays yours.

Limits

A checker can only check what’s in front of them. They can’t catch a source you left out, a study you never found, or a quote you made up and didn’t footnote. The annotated copy is only as honest as you make it.

Checkers miss things too, and two people who share the same assumptions can miss the same error. As Katie Palmer, then science and health editor at Quartz, puts it in the KSJ handbook, fact-checking is “also about checking assumptions.” A reader from outside your field or your point of view is more likely to spot those.

Checking takes time, often more than writing the passage did. For a short piece, focus on the claims that would do the most harm if wrong.