Hacker News Reader

Comment by hervature | Hacker News Reader

How is this on the front page? There are just so many statistical methodology problems that you cannot draw any conclusions. First of all, they vast majority of the article is n=1 where they sampled one year. Then, they talk about increase over one year and come up with the mind blowing stat of "6 out of 10 Highest Murder Rate Increases are in Trump-Voting States". From a statistical view point, that is to be expected if there are no differences. Furthermore, the entire analysis is stupid because 2020 was the pandemic year. Finally, it doesn't take a genius to correlate these to wealth levels. Go to [1] and sort by median household income. What you will find is basically the same ranking as in this article. What do you believe - poverty or political preference leads to increases of crime?

[1] - https://en.wikipedia.org/wiki/List_of_U.S._states_and_territ...

View on HN

> First of all, they vast majority of the article is n=1 where they sampled one year.

Wouldn't "n" be number of murders?

> Furthermore, the entire analysis is stupid because 2020 was the pandemic year.

It was a pandemic year in every one of these states.

> Go to [1] and sort by median household income. What you will find is basically the same ranking as in this article. What do you believe - poverty or political preference leads to increases of crime?

What if the two of them are linked? If a handful of states follow the same political ideology, and also share an outsized poverty and murder rate it is worth analyzing if those same policies have an impact.

> Wouldn't "n" be number of murders?

No. I'm not disputing the murder rates. Those are probably very accurate for that year. I'm disputing the claim that red states have a murder problem based on one year of information.

> It was a pandemic year in every one of these states.

Not equally. Given the information they used, it is entirely plausible that the murder rate did not increase more in blue states because, in general, they weren't allowed to go outside unlike, say, Florida. To be clear, this is not "proof" that an ideology is better. This is a temporary measure that could lead to longer term issues.

> What if the two of them are linked?

That's what I'm saying. This article is claiming a link without actually analyzing the impact. You need a longitudinal study is my whole point.

> You need a longitudinal study is my whole point.

I agree. I think this report is just meant to provide a counter to the prevailing (accepted?) narrative. Maybe it's meant to inspire/fund the type of study you suggest?

>they vast majority of the article is n=1 where they sampled one year

For the sections on murder rates (not their change), n is the population. These sections make up the majority of the article. For the section on year-to-year changes, those are comparisons of two populations (each with n being that year's population count). There's no way sampling error would have a noticeable effect on the rates for states or major cities. Instead, those changes are caused by things a model should account for. TFA doesn't present a model, because that's not the point.

>"6 out of 10 Highest Murder Rate Increases are in Trump-Voting States". From a statistical view point, that is to be expected if there are no differences.

I think that's the point of TFA.

>What do you believe - poverty or political preference leads to increases of crime?

TFA's conclusion is the narrative of murder rates increasing because of Democratic political leaders doesn't jive with reality. The title is provocative, but and it takes an unnecessary jab with, "A more accurate conclusion from the data is that Republicans do a far better job blaming others for high murder rates than actually reducing high murder rates."

But you're really just agreeing with the article.

No, I'm not agreeing at all. The title and the first line of the article literally state that red states have a murder problem compared to blue states and use bogus stats. You don't know my stance on this issue because I have not stated it. I'm simply saying that you cannot take a stance from the information presented in this article.

Reply on news.ycombinator.com