By Stephen Young
By Stephen Young
By Stephen Young
By Jim Schutze
By Rachel Watts
By Lauren Drewes Daniels
"That's what I told them. I said, listen, to me, that's the story. We get 20 percent African-Americans on our jury panel, and lo and behold, we have 20 percent African-Americans on that particular jury. I think once they reached that conclusion, then they had to start manipulating the thing a little bit to make a story out of it."
This is where we get into the statistical model and the really convoluted arguments. The News says in its stories that the juries come out the same ratios as the jury pools more or less by accident. The paper reports that prosecutors reject blacks at twice the rate they reject whites, but the ratios come out the same because defense attorneys reject whites at three times the rate they reject blacks.
See if you get this: One form of discrimination balances the other one out. But what's against the law is discriminating against black people, and the prosecutors are still wicked because they're still doing it, no matter where the ratios wind up.
So hold on here. I know this is turning into a head-buster already, which is part of the problem. But I'm afraid I have to toss in another wrinkle. Frederick C. Moss, an associate professor in the Dedman School of Law at SMU, explained to me that the Batson doctrine has evolved to apply to defense lawyers as well as prosecutors and to all jurors of all races.
"When the first Batson came out, everybody thought it was just going to apply in effect to prosecutors," Moss said. "But when subsequent cases came down, they made it clear that they weren't just protecting the 6th Amendment rights of criminal defendants. What they really were protecting was the right of jurors to sit as jurors and not be thrown off of juries because of their race or ethnicity.
"Transforming it to the rights of the individual juror expanded the Batson protections to all jurors in every case. So now it outlaws discriminatory strikes by defense attorneys in criminal cases as well as attorneys in civil cases."
So...bear with me here...according to the News' own findings, defense lawyers are striking whites at three times the rate they strike blacks. So the News could have stuck a headline on this sucker saying: "WHITE PEOPLE UNFAIRLY KNOCKED OFF JURIES BY RACIST DEFENSE LAWYERS."
Time for the martini mixer. The News argues in its stories that its logistic regression analysis measured all of the factors involved in jury strikes and found that a person's racial status as an African-American was the most important one.
But before the stories were published, Hill hired his own experts and asked them to look at the News' findings. Eric Fritsch, an assistant professor of criminal justice at the University of North Texas, told me that in order to know anything about the News' findings, he needed to examine the inner workings of the statistical model.
The News refused. The newspaper told Fritsch that the model itself and all supporting data were confidential information on a par with the contents of a reporter's notebook. And they never turn over notebooks.
Fritsch says that's not even remotely kosher. He told me he is the editor of a journal. People submit articles to him all the time with findings based on this kind of model. He always asks for back-up information in order to know if their model is any good.
If the model's no good, the rest of the series of articles is garbage. The News presented a number of comparisons that seemed to show racism, especially where blacks were struck from juries more often than whites who answered the same question the same way. But you don't know if it's racism until you know what other factors were in play. Maybe they answered the question about punishment the same way, but more of the blacks who answered that question also had previous bad experiences with law enforcement.
The only thing that weaves all of that together properly is the model. And you can't tell if the model is any good if you can't see the typical measurements statisticians use to judge effectiveness. That's what Fritsch asked to see.
In particular, he asked for the "hit ratios"--a measurement of how often the model accurately predicts outcomes. The News wouldn't even tell him that.
"In my field, I say, 'I need to know what the hit ratio is,' and anybody could blurt that off the top of their head," Fritsch said.
"It's not something that should be hidden. If the model is hanging around 50 percent, it's not any better than flipping a coin."
I asked him what would happen if somebody submitted an article to his journal but declined to provide details about the model used.
"They'd get a rejection letter," he said. "Quickly. It would get the red rejection stamp."
I talked about all of this with people at the Morning News, on and off the record. At first they defended their refusal to provide Hill with details on the model, because they said they and Hill had an adversarial relationship on the story. They didn't see why they should provide Hill with ammo he would use to shoot them down or even sue them.