Hacker News Reader

Comment by dre85 | Hacker News Reader

"Remarkable" things happen when kids grow up in projects, such as earning 500 bucks more ANNUALLY. Maybe the authors managed to p-hack their way to some statistical significance with this number, but I would think that the life significance of an extra 500$ a year in the U.S. is not something astounding.

View on HN

The poverty level in United States is $11,880. At the top end, $500 is 4.2% and plenty of people are making less than $10,000 a year. At that rate, I think $500 is starting to look statistically significant.

https://aspe.hhs.gov/poverty-guidelines

Let's not conflate the very different concepts of "statistical significance" and magnitude of the effect.

An effect with a small magnitude can be highly statistically significant, and vice versa.

If you don't understand the very important difference, then you probably flunked out of freshman stats.

Sadly, I suspect a significant number of people who didn't flunk freshman stats have trouble with this. The terminology is simply poor.

No need to throw shade, BTW.

Well, considering that median household income for the US is $52k/yr, $500/yr is ~1% increase in household income per person, so a ~1-2% effect per household, which isn't nothing, even if not astounding.

Of course, that's if it were households which earned a median income that received the benefit. Given the correlation between poverty growing up and poverty later in life, it's likely more like a 5% increase per household on the low end of the spectrum, which I would call a fairly strong effect.

Thinking that an extra $500 a year in America couldn't significantly impact someone's life says more about you than the study.

I was more so going for the point that it doesn't propel you into a different socio-economic class. 500$ is not necessarily a higher education or a better job, it's more like an extra few overtime shifts.

There is no "p-hacking", you can see the original study here [1].

And $500 seems like allot to privileged people but even shopping at a bottom level chain such as Ross, you could get 50 pairs of pants for that price. Or a loaf of cheap bread almost every day.

Even for someone just above the poverty line it represents a 5% raise and for the 40 million below that, even more.

If data don't match ideology, it is not the data that is wrong.

[1]http://www.nber.org/papers/w22721?utm_campaign=ntw&utm_mediu...

So, you discount the research because your intuition says otherwise. I personally find that remarkable.

I don't think it's discounting the research because of intuition. It's discounting the research because, while there is a statistically significant difference, the commenter above is arguing that it is not practically significant.

They also found decreased rates of incarceration, which is pretty significant.

Statistical significance is a measure of certainty that the effect exists, not about how big it is.

To add to this -- whenever one does analysis based on very broad correlations, and where there is much potential for confounding variables that can never be perfectly controlled for, it is very easy to cut the data in a way that results in a 5-20% effect either way. That is why there are studies that show pretty much everything from broccoli to butter both causes and prevents cancer -- http://www.vox.com/2015/3/23/8264355/research-study-hype. So when looking at correlations, you need to ask, is this a "broccoli causes cancer result" (a small effect, easily caused by confounding variables and/or p-value hacking) or is this a "cigarettes cause lung cancer result" (a giant effect that cannot be explained any other way). The study in this article is a "broccoli prevents cancer" type of result. It is possible there is a small effect from the intervention, it is also possible there is no effect. The only information we can actually take away from the study is that there is not an obvious large effect.

The article says the earnings are $500 more future yearly income _per one incremental year in public housing_. For someone in public housing for a number of years, this is significant.

It's good that you used "remarkable" as a bridge for your analysis, but can we also edit this title to be less click-baity (I'm sad to see Washington Post start steeping as low as Buzzfeed here too)?

Suggested edit:

"The remarkable thing that happens to poor kids when you help parents with rent"

"Children of parents who receive public housing assistance associated with having better life outcomes"

...but I would think that the life significance of an extra 500$ a year in the U.S. is not something astounding.

Based on what data?

Reply on news.ycombinator.com