Comment by legitster | Hacker News Reader

Comment by legitster | Hacker News Reader

Parent Full thread

> Rho and her colleagues focused on Black drivers because this group is stopped by the police at higher rates and are more likely to be handcuffed, searched, and arrested than any other racial group.

Okay, so this doesn't seem like a particularly useful study because they didn't actually include a control group.

> Rho says in planning this study, they had initially set out to look at patterns related to traffic stop escalation for white drivers too, but realized that it happened so infrequently for white drivers that there just weren't sufficient numbers to even include them in the analysis.

Okay, I'm sorry is this for real? The author just completely threw out the control? How is this not malpractice?

So it could very well be that white drivers were issued the same set of commands at the same rate (for all we know). But we will never know because the author intentionally withheld the analysis that would have established the connection between the phrases used and race.

The author just completely threw out the control?

There was no control. Reread what you quoted: there wasn't a statistically valid number of incidents involving white people to make anything of it. I don't even know why I'm paraphrasing, what you quoted was crystal clear to me.

> in the one-month period covered by this study, the city's police officers did 588 stops of Black drivers and only 262 stops of white drivers.

You left out a variable: black people make up what percentage of that population of that city? If it's, I dunno, the approximately 15% that I would guess it to be, then that's a significantly large number of stops in comparison to white folk.

It's all fun and games until Bayes comes to play. They didn't state the city, but simply said a "Medium Diversity City." Indianapolis is roughly medium at 30% black population although the country average is 13.6%. So the probability of a car containing a black person is roughly 1:3. This means that if you put a random car in front of an officer they are 6 times more likely to stop a black person in this study. Then after that there is such a huge difference in how the stops are handled.

The TableS1 in the data for the study is telling. It's no wonder they don't want to collect statistics on how policing is handled. It's downright damning.

https://www.pnas.org/doi/suppl/10.1073/pnas.2216162120/suppl...

> Over 15% of Black drivers experienced an escalated outcome such as a search, handcuffing, or arrest, while less than 1% of white drivers experienced one of those outcomes.

So I think they had 2 "escalated" cases with white drivers, which I would agree is too small to do any statistics with.

Defund NPR.

NPR loves them studies.

"We wanted to use white people as control for arrest stats but the police don't arrest white people" seems like a result you would also want to publish.

> in the one-month period covered by this study, the city's police officers did 588 stops of Black drivers and only 262 stops of white drivers.

So keep in mind they had recordings from 262 traffic stops of white drivers.

If they had the data that said "white drivers" were 80% less likely to hear the escalatory phrases and so they didn't escalate - boom! That's an open and shut case.

But them using the excuse of a lack of escalations to justify the complete absence of this data is just way too problematic.

They're being pretty clear about the limits on the data they had, the decision they made because of it, and why they made that decision.

What exactly are you alleging here? Malpractice is a pretty strong word but also... vague. PNAS is fairly prestigious and there would be substantial reputational risk of malfeasance with this data. To me that carries a lot more weight than someone on the internet throwing unclear accusations around idk.

The actual study in question is pretty limited - I don't have any problem with the data published to the PNAS report: https://www.pnas.org/doi/10.1073/pnas.2216162120

But the decision to limit the study exclusively based on race was a post facto decision that doesn't make a lot of sense given the context of what they were studying.

Imagine if they published a study called "Escalated police stops of Right Handers are linguistically and psychologically distinct in their earliest moments" and then you found out that they left out the data on left handers, you would also be very suspicious.

Malpractice might be a bit of a strong word, but I find it distinctively annoying when researchers intentionally play around with their sample sizes.

Reply on news.ycombinator.com