Hacker News Reader

Comment by rayiner | Hacker News Reader

> In the end, only two men of color make it to the jury, and I am one of them. The other is Latino. There are two Latina women, one African-American woman, and one Asian woman. The remaining six jurors are white.

Thats basically the racial composition of the U.S. Indeed, people of color are over-represented in that jury.

View on HN

Juries are drawn from local populations, whose demographics can differ radically from the national average. We can't know whether or not people were under-represented or over-represented without knowing where this took place.

So the jury was 6/12=50% white, 3/12=25% Latino, 2/12=17% Black and 1/12=8% Asian.

The US racial demographics are White: 64%, Hispanic: 16%, Black: 12%, Asian: 5% [1]

Looks like Blacks and Latinos were overrepresented on that jury.

[1] https://en.wikipedia.org/wiki/Demographics_of_the_United_Sta...

Perhaps the author's city's racial composition is very different from that of US's national average, hence his implicit complaint.

Since the author wrote anonymously (and understandably so), there's no way for us to know either way.

That very well could be the case. But even if we knew the demographics of the community the sample size is too small to draw any conclusions.

really? There are places in Texas or California, for example, where the population is over 50% latino. In that case having only 1 of 12 Latino jurors would be sign of something, right?

A single jury is still too small of a sample size to be statistically significant. If you have more than two groups in the population (races in this case) then you need at least 5 expected cases of each in order for analysis of the actual versus expected proportions to be meaningful, which you can't get with 12 jurors. The closest you can get to it is if there are 3 groups evenly divided in the population, which gives you 4 expected cases each. If you had only 2 groups in the population that were each 50% then you would have 6 expected cases each. However, when analyzing sample proportions of 2 groups you need at least 10 cases each.

Those minimum case numbers are needed in order for the central limit theorem to be applicable in the statistical analysis.

Huh, interesting, thanks for the explanation. Is there a nice summary/term I can search for that explains your needed sample side for things to be statistically significant?

Comparing proportions is the general topic. You are typically comparing a sample proportion (like the demographics of a jury) to a population proportion (like the demographics of a community from which jurors are drawn).

Juries are composed of 12 people. Saying there are discrepancies between the jury pool and the US of less than 8 percentage points is somewhat silly -- making even the smallest change would not reduce the size of those errors.

In the particular case, this jury has two missing whites and a full extra "hispanic", so swapping the hispanic out for a white would improve everything. The other missing white is filled in for by a partially-black partially-Asian guy.

Indeed. One jury is far too small a sample size to draw any conclusions (too obvious to even state but I did anyway).

When comparing proportions (in this case racial makeup of juries versus racial makeup of the population) you want a sample size large enough that you have at least 5 expected cases for each group. In this case the expected cases are the population proportion (p) times the number of jurors (n). So you would need n * p >= 5 and since Asians are the lowest proportion at 5% you would need data on at least 100 jurors. Then you could do a Chi-square goodness-of-fit test to see if something is really going on.

If it seems like I'm being geeky it's only because I just recently worked on a problem exactly like this, determining if jury demographics matched population demographics. Turns out in many cases there IS something going on, but then the reasons for it aren't always what you think. One reason is that in some communities there is a shortage of eligible jurors due to high rates of felony convictions.

On the other hand, we're supposed to be judged by a jury of our peers so, demographics aside, it's disturbing how often a person of color is convicted by a primarily white jury. We're seeing nationwide testimony, by how certain isolated events are receiving reactions that indicate that they aren't isolated, that police misconduct towards minorities is far more common that we white people would like to believe. And yet this story shows just how easy it would be for the minority jury members (both in terms of race and dissenting opinion) to cave in and allow someone to be convicted based almost entirely on the presupposition that the police did nothing wrong.

> On the other hand, we're supposed to be judged by a jury of our peers so, demographics aside, it's disturbing how often a person of color is convicted by a primarily white jury.

What the hell? Are you saying that people of color are automatically peers with each other because of their race?

peer: a person of the same age, status, or ability as another specified person.

>peer: a person of the same age, status, or ability as another specified person.

If that is your defination of a peer then it is likely this person was not judged by their peers...

a "peer" to the government is any citizen living in the jurisdiction of the court forming the jury....

A jury of your peers does not mean a jury of your race.

The United States is a white majority country. Of course the majority of jurors are white, even when the defendant is black.

> 2/12=17% Black

There was one black woman and no black men [1], so that should be 1/12=8% Black. I don't see anything in the article that lets us figure out what category the author belongs in. He's a male non-black non-latino who describes himself as a person of color, which leaves open a lot of possibilities.

[1] "Every black man had a story: police harassment, spurious arrests, intimidation. They were all eliminated"

Oh, I misunderstood that he was a black male. It's a mystery to me what race a non-black non-latino person of color is.

The sample size is too damn small!

Added: I guess I'm not used to the term "person of color." It means "a person who is not white or of European parentage." Many people other than blacks and latinos could identify as people of color.

No, over-represented with respect to the U.S. population which is irrelevant. Chinese people were criminally underrepresented with respect to the population of the world. It depends on the local demographics which we don't have access to. What we do have access to is the fact that there were many, many more people of color in the selection pool who were removed because of racial discrimination.

Latinos aren't men of color?

>two men of color make it to the jury, and I am one of them. The other is Latino.

Perhaps you need to read this again

The author could still be Latino as well.

There were two men of color, one black and one latino. All the other people of color were women.

There were no black men: "Every black man had a story: police harassment, spurious arrests, intimidation. They were all eliminated".

I think he meant all the others were eliminated.

The author himself is a black man.

> two men of color make it to the jury, and I am one of them. The other is Latino.

That statement you quote does not imply that the author is black. All it says is that he's not white and not Latino. That leaves at least Asian, Indian (either definition), and Pacific Islander open.

Reply on news.ycombinator.com