Journalists traditionally have a one-way relationship with sources. Whether it is a CEO speaking on the record or a nervous whistleblower with a confidential tip, our approach tends to be the same: We take the information, publish the story and deal later with whatever the source thinks of what we wrote.

This practice makes a lot of sense in investigative reporting when we interview powerful people for a story that may put them in a bad light. Giving them advance word of what we plan to write gives them opportunity to hide the evidence — or even get out ahead of us by denouncing our story before it’s published.

But the vast majority of our interviews are not adversarial situations. People we interview are often helping us out. Our relationship with them is a cooperative one. Being human, we sometimes misinterpret what they tell us. Our interviewees, also being human, can overstate a case or leave out an important qualification.

Certainly there is nothing wrong with calling sources back to clarify a point we’re unclear on. We also might read back to a chemist how we plan to describe a complicated chain reaction she told us about.

But what about doing more than that? Might we read back quotes we’re not doubtful about, just to make sure one more time that they’re correct? In the interest of avoiding a correction later, what about letting the source review background information she gave us — how many meals the project distributes per day, what year the group traveled to Hong Kong?

Some reporters have been known to send an entire story to a cooperative source before publication — not only to check the facts, but for a reality check on the whole thrust of the piece.

Other than a quick call for an essential accuracy check, letting sources review content in any more detail is fraught with potential danger. Our right to quote material from sources as we heard it, in the fashion we want, is a precious one. Yet some newsrooms allow pre-publication review to varying degrees.

The E.W. Scripps Company doesn’t let its staff send sources full stories, but its ethics guidelines say “we do often read quotes back to a source prior to publication for purposes of accuracy and fairness.”

The BuzzFeed News standards guide says, “Sending a note to the subject that includes allegations or a description of what will be published is a reporting tool that also acts as a safeguard for the reporter.”

The Denver Post ethics policy bans sending content back to sources, except when a senior editor approves sharing passages from stories in the interest of accuracy.

Many ethics codes make no reference at all to pre-publication review. For that reason, and because staffers may have worked under different rules at different places, the issue deserves thorough discussion and clear guidelines in each newsroom.

Everyone must understand what policies exist, the thinking behind them and what exceptions might be possible. Since there is no way to monitor what journalists are sharing with their sources, reporters are more likely to follow the policy if they had a chance to participate in making it.

What should be non-negotiable in such a policy? My view is that reporters should never agree in advance to show an interviewee either his quotes or the completed story. (Not everyone agrees; The Washington Post allows some leeway for quotes.) Showing any content to a source should always be our choice, especially in adversary situations. If we make promises to one interviewee, others will expect the same treatment.

If we decide some degree pre-publication review should be allowed, here are questions to consider in fleshing out a policy:

When would we do it?

There should be a specific reason for checking material with a source. Are we genuinely concerned that we might have gotten the quote wrong? Does a background fact the source gave us seem questionable? Are we concerned that something in the story could endanger the source?

We should check what a source gave us for a reason, not as a general CYA crutch.

How much of the story will we show?

Sending a full story is highly risky. It makes it look like we have doubts about the whole thrust of our reporting. If we’re that unsure about what the story should be saying, we haven’t done enough reporting.

It might be possible to test out the story’s angle in the course of general conversation, perhaps when calling the source for additional information.

What are we willing to change?

We might decide to let a source change a word or two of a quote to make its meaning more precise (e.g., “jets” instead of “aircraft”). But collaborating with a source to change the whole message of a quote is rewriting history.

The Washington Post allows a source to change a quote in rare situations, but says a better option is to “permit a source to add to a quotation and then explain that sequence to readers.”

In checking facts, we might also set clear expectations about what we’re willing to do. When Tanya Mohn, a New York-based freelance journalist, feels a need to check a quote in a story, she tells the interviewee, “I am unable to make changes or add anything unless something is factually incorrect.”

Do editors need to be involved?

If an editor must give approval for a reporter to check material with sources, this should be clear. Unless such a policy is officially stated, editors cannot profess shock when it turns out a reporter has shared something with a source before it’s published.

Whatever pre-publication review policy a newsroom decides on, two factors should not influence it:

Ingratiating ourselves with sources. We should prove our value to sources through balanced and accurate reporting. Regularly providing a view of what we’re going to publish can lead sources to think we’re giving them a veto on what we publish.

Production issues. If we send a quote to a source at 5 p.m., we can’t be sure we’ll get an answer by a 5:30 deadline. If an organization decides source review is journalistically justified, production problems shouldn’t be allowed to short-circuit that. Publication of some stories may be delayed as a result of the policy.

The most important thing is that every news outlet has a clear understanding on source review, set with the involvement of as many writers as possible.

Thomas Kent (@tjrkent) is a consultant on journalistic ethics and combating disinformation. He has led ethics workshops for Poynter and teaches at Columbia University.