Hacker News Reader

Comment by rgbrenner | Hacker News Reader

Did you filter retired people? Adjust for the disabled? It doesn't sound like you did any of these things (if you did, you should mention it).

Did you adjust for students? Doesn't it seem odd to you that in the BLS data the 'less than 5000' group is second only to the '70000 or more' group in the amount spent on education? And that more of them have college until you reach the 40000+ groups.

Then you draw conclusions from your faulty data.

Also, where are your references for this. You put footnote numbers in the article, but didn't include the actual references.

This is what I came up with in 5 minutes. What are your qualifications?

[Edit: just to emphasize how odd the BLS data is without adjustments, it says that the AVERAGE person in the 'less than 5000' income group spends almost $1500 on education... more than double any other income group except the '70k or above' group)

View on HN

You put footnote numbers in the article, but didn't include the actual references.

Weird, huh? Anyway, since I couldn't readily find substantiation for that 24% number, and the added dubiosity coming from the fact that he doesn't even mention whether it's adjusted for students, disabled, etc (as is standard practice in econometrics), I didn't even bother with the other charts or arguments he was trying to make.

I couldn't readily find substantiation for that 24% number

That's because you didn't read the first paragraph of the source I cited.

The point is, if you're going to make not only highly charged, but basically, incredibly insulting assertions about a very large group of people (that also happens to be much weaker than yourself) -- especially assertions that are readily seen to be based on flawed selection biases and other sloppy reasoning -- then you had better make at least the numerical parts of your claims instantly verifiable from the sources you cite.

It's just as in academic research -- yeah, maybe it's buried in the source somewhere, but no one has time to parse paragraphs, and divide numbers for you -- especially when you're already trying their patience by making such structurally weak arguments in the first place.

Pointing fingers (and, implicitly, accusing people of being lazy or unable to read) to distract from the fact that don't seem to have an answer for their main criticisms of your "argument" doesn't help, either. It's just grasping at straws.

10.4 million / 43.6 million ~= 24%

It's not adjusted even for children, though.

It wasn't adjusted to exclude traditional non-working categories of any kind at all. That's what made is such an obviously useless figure, right there in the first sentence.

Sure, but some of those non-working categories (in particular, students and retirees) could arguably go into the workforce if sufficiently motivated, and might be assumed to be a small portion of the population. Children seems more unequivocal - and thus deserving of calling out specially (while remembering that this is in addition to those other categories).

I did none of that. Are you asserting that a significant chunk of the poor are merely the retired, disabled and students? If so, it sounds like you are arguing that poverty is a much smaller problem than is normally believed.

The footnote referred to the previous link, fixed it now. My "qualifications" are irrelevant - my argument is not coming from authority.

No[see edit], but they -- retired, disabled, students -- are significantly more likely to be in that income group, and all of those groups fit the pattern you've observed -- they spend more than they earn.

And I asked about your qualifications because you didn't do some very basic things to the data before drawing your conclusions.

[Edit: typically in studies like yours, YOU would quantify the number of retired, disabled, etc groups that may affect the results, and justify why you did or did not exclude those groups. I don't know what portion of the under 5k income group those people represent. You should know, and should have data on why this does not affect your results.)

If you have data sources (or proper econometrics papers) that slice things more narrowly, I'd love to see them. If they are as basic as you suggest, possibly you have a link showing how to do it? If I'm so clearly wrong, I'd love to see a refutation.

(Incidentally, ultimately the right thing to do would be to simply look at the law and compute income-taxes+transfers+in kind as a function of earned income. But I don't know how to do that.)

I'm not an economist.. I imagine there are papers with better data, but I would have to spend some time trying to find them.

I did come by this article which deals with the same question: http://gregmankiw.blogspot.com/2006/09/measuring-poverty-inc... http://economistsview.typepad.com/economistsview/2007/06/fam...

If income volatility has increased, it could explain why the spending-income gap is so wide, and why they (poor) spend more on education (retraining to re-enter the workforce).

Maybe try asking EconTalk (podcast of economics topics).. Sounds like something he would find interesting, and he often takes questions and makes episodes out of them. http://www.econlib.org/library/contact.html

Those are interesting, but don't really address the same question. A quick google search also turns this up:

http://johnhcochrane.blogspot.co.uk/2012/11/taxes-and-cliffs...

The numbers are more or less in agreement with mine, albeit more accurate (since they look at actual volume of taxes/transfers) and computing slightly different numbers (disposable income vs income rather than consumption vs income). They also show quite a few income cliffs.

I hope you'll forgive me for not linking to a much better 2012 blog post from my less interesting 2011 blog post. ;)

In any argument, there is an implicit "... and I'm not leaving anything out that will substantively change this analysis" which is accepted (or not) significantly on the basis of authority by anyone without deep knowledge in the area being discussed.

I doubt most students - even if they are making little money themselves - are accounted as being below the poverty line when they are not providing for themselves. It would be tremendously misleading.

I expect the bulk of the 76% of the "43.6 million people [who] lived at or below the official poverty level" are children. Children are explicitly included in the number: "Although the poor were primarily children and adults who had not participated in the labor force during the year [...]". While this does not say that they were primarily children it certainly does say that children were included. So presenting the source's "(children + adults who did not work for any reason)/total = 24%" as "adults who chose not to work / total = 24%" is flat out wrong.

> Did you filter retired people

Especially now that the baby boomers are hitting retirement, it seems like controlling for this is important.

Reply on news.ycombinator.com