Updated May 28, 2021 at 12:27 PM ET

The way the Minneapolis Police first described George Floyd's murder — "Man Dies After Medical Incident During Police Interaction" — didn't mention that an officer held his knee on George Floyd's neck for more than nine minutes. It did mention that Floyd physically resisted officers, a detail which former officer Derek Chauvin's defense team leaned on during the murder trial — although Chauvin was ultimately found guilty.

For decades, journalists have treated official police reports and statements as trusted primary sources. Now, some are questioning the reports' reliability and objectivity as part of a reckoning in the media spurred by George Floyd's murder.

Crucially, what a police report states — or doesn't state — impacts the narrative of an incident. Phillip Atiba Goff, co-founder and CEO of the Center for Policing Equity and a professor of African American studies and psychology at Yale University, says that's "always been a portion of what's been wrong with law enforcement."

"But what we've seen in the last seven years, since Ferguson in particular, is that folks have started to see there's a pattern in the ways in which facts are omitted," he says.

Goff spoke to NPR's All Things Considered about the human nature of bias, the limitations of asking police to police themselves and the way that introducing skepticism can lead to more accurate reporting. Listen in the audio player above, and read on for highlights of the interview.

Interview Highlights

On the kind of information normally left out of police reports

We obviously don't know. You can't have a record of the things that don't exist. But what we do know is that in city after city, there are communities that are concerned that the elements that police are responsible for, especially when there's a bad outcome, are left out because oftentimes in the worst of those situations, the only person left alive to record the incident is the person that did the dirt.

On how bias functions in police reports

I'm a social psychologist. If I'm writing down the story of what happened during this interview, there's going to be a pro-Phil bias. ... That's a human thing. And so we don't need to be sort of villainizing people who are doing a human thing. It's also the case that between the two of us, if I'm writing down the story of this interview, there's likely to be a pro-Black bias. I'm Black. I'm pro-"my group." Those are normal human functions. So when police are writing down reports, they're writing down things that are either subtly, or explicitly, pro-police.

The problem comes in when the slightly pro-Phil bias and slightly pro-Black bias I have — it becomes untethered to reality and I'm writing down out-and-out lies. And that starts to become part of a culture when you can do it and avoid any consequence. The issue has been for a long time [that] we've asked police to police themselves, and unfortunately, that doesn't always work. Because when one person does something wrong and gets covered for: that becomes two, that becomes 20, becomes 100, becomes an entire department.

On what journalists can do to more accurately cover policing

So I want to be clear: I'm not saying all police lie about everything all the time. What I'm trying to say is communities have known for some time that a police report is not the whole story and it's not the way that the community would tell the story. So at the very least, start with [saying] "police claim." That's the first thing. Contextualize it by [asking] "Who the heck is saying it?"

Just the phrase "police claim" frames the whole conversation in terms of "police are saying this," and we now have enough evidence that we're reasonably skeptical of when they are, and when they're not, telling the truth — and when they say that someone died because of [an incident] and that person obviously is nowhere around, we want some corroborating evidence. And that's got to be a part of the way that we get out of the situation we're in right now.

Copyright 2021 NPR. To see more, visit https://www.npr.org.