Last time we discussed a data set, in many ways very similar to income/wealth distributions, but simple and small enough for an unskilled person to analyse with a standard spreadsheet package.
The data are distributed very unequally. A few unusually high values had a huge influence in the statistics of the sample:
"1% of the sample determines 41% of the sample's mean.
There's more. … From 3,252 'likes', for example, the range would collapse to 480 with the removal of the "1%": a 85% fall!"One would need to consider very seriously the presence of a number of outliers: from the top six (in a very conservative "guesstimate") to the top 89 (in a much more structured and stringent estimate).
That data set could offer an interesting exercise for advanced high-school/early undergraduate students of statistics/econometrics. In particular, it would allow for ties with topical issues, like the Piketty/associates literature on income/wealth size distribution:
"In words, 25% of the sample gets between 1 and 2 ‘likes'; the next quarter gets between 2 and 4; the third gets between 4 and 8; and the last between 8 and 3,253 … Welcome to inequality."It, however, has another interesting characteristic. I never explained what that data were supposed to mean. What are those "likes" and data points? I mentioned a survey, how was it conducted? Was that a random sample? What were the questions included?
Today's the day!
The most efficient and fairest way to answer most of those questions is to point to the website publishing those results. It explains the methodology followed and describes the idiosyncratic voting method adopted (more on that soon).
Greatest 20th Century Economists
|Right click to open a larger version in a separate tab.|
Above is just an overview of the results. I didn't edit their list in any way: it contains heaps of names I didn't recognise and some typos. Occasionally people I didn't know were economists apparently are (or were) economists after all; likewise, people I thought died in the 19th century, were alive and kicking well into the the 20th, it seems. Two authors, like Jekyll and Hyde, share a single identity. Some names may be repeated.
Speaking of which, for brevity, I mention only better-known names, mostly Keynesians of one sect or another. In 2006, when that survey was conducted, Steve Keen was the only among the modern and living post Keynesians to make that list (I couldn't find any of the leading MMTers, nor Thomas Palley or Marc Lavoie, for that matter). Incidentally, Keen got 5 votes, placing him squarely within the 3rd quartile. He lost to the Mahatma Gandhi, though, who got 6 votes. Oh well.
There is, however, a graphical representation of that data set, less focused on numbers, but more colorful and surprisingly enlightening in an allegorical kind of way. It doesn't come from statistics, but from art:
Greatest 20th Century Economists
|Ascension of Christ, by Guariento D'Arpo. [A]|
Personally, I like that one, but you can choose: there are many such graphical representations.
So, how should one interpret those results?
That is a frightfully hard question -- and probably irrelevant. The publishers themselves did not answer that question. In itself, that is telling.
The idea is that subscribers to that website determine their own lists of five economists: it's not a random sample of voters, in other words, voting on a fixed set of choices (the open character of the list goes a long way into explaining the abundant presence of little-known economists).
Then it's time for voters to rank economists in their lists. Voters proceed like this: if, in the voter's estimation, Economist A "most added to our understanding of economic phenomena" then she gets 5 votes. If Economist E added least, he gets 1. Economists B, C, and D -- who added intermediate contributions -- get intermediate numbers of votes: 2, 3, or 4.
It's an odd scheme, isn't it?
Suppose now two voters elaborate their lists independently, to the best of their knowledge, as honestly as possible, with no strategical thought whatsoever. Unbeknown to them, however, in their unanimous assessments Economist E added the least to our knowledge. Each voter, therefore, separately gives Economist E 1 vote, thereby assuring Economist E gets 2 votes.
In an extreme situation, given that 1,249 people cast their votes, Economist E (who in their unanimous opinion added least to our knowledge of economic phenomena) could well make it into the "1%" (or Heaven).
Take that, econo-plutocrats (or econo-supernatural-beings)!
It turns out, one doesn't need fat fingers, an oddly-designed voting mechanism on a non-random sample seem enough to explain those outliers.
What it does not explain is how the whole thing escaped criticism for over 10 years, awaiting for a commie worker, untrained in statistics. It's not that the results themselves don't look weird, or that it takes a rocket scientist to perceive them design flaws, or that critics didn't take notice of the survey: Tyler Cowen did know about the survey and its results. So did his readers. Cowen couldn't go any deeper than: "Given the source, expect a left-wing, anti-neoclassical perspective". So, he did notice the "non-random sample" thing. That's not being very perceptive: that's the standard criticism to that kind of surveys. But one can still reasonably excuse Cowen: he's got bigger fish to fry.
Somehow, that doesn't sound as reasonable for his readers. Trust me, fellows, you can do better. No need to appeal to cheap shots. No need to pretend you know better. You can actually demonstrate that.
The same applies to those across the ideological street: just because it flatters your preconceptions or it's embarrassing to the other guys, it doesn't mean it ain't nonsense or that you need to keep quiet for "political reasons". There's a limit to being team-players.
It's not just economics that's in trouble, it's the whole bourgeois intelligentsia.
Anyway, given that there's no proposed official interpretation, one is left to guess the interpretation. Two seem possible.
One is that only the ordering matters: 2 votes do not indicate double the contribution of 1, just a larger (or more "important", or more "crucial", or more "liked") contribution. It's, in other words, just a ranking, one with rather humorous properties.
Another is that voting figures represent a measure, however approximate, of the magnitude of the contribution those economists objectively made to "our understanding of economic phenomena". It has cardinal and ordinal properties, much like like political election results, or the marks students get from their teachers are supposed to objectively reflect their performance, or wages in a free market are supposed to objectively reflect costs and productivity at one hand and incomes and tastes at the other. Given that the publishers went to the trouble of publishing the voting figures, I tend to prefer it.
In that case, I just hope economists' salaries do reflect their contributions. But that's me. It's up to you. :-)
[A] "Ascension of Christ", by Guariento D'Arpo. Circa 1344. Source: Wikimedia. Image in the public domain.