RSS
From my new column in Taki’s Magazine:
We live in a century of nonstop adulation over how statistical analysis of big data is changing the world. Brad Pitt, for instance, starred in a successful Hollywood movie, Moneyball, about the fast-changing realm of baseball statistics.
Last week, however, Andrew Gelman, a professor of statistics at Columbia, offered some heresy about his now fashionable field on his blog Statistical Modeling, Causal Inference, and Social Science. In a post entitled “A World without Statistics,” Gelman reflected:
“A reporter asked me for a quote regarding the importance of statistics. But, after thinking about it for a moment, I decided that statistics isn’t so important at all. A world without statistics wouldn’t be much different from the world we have now.”
How can we test that?
Read the whole thing there.
English
Arabic
Chinese (Simplified)
Dutch
French
German
Hebrew
Hindi
Italian
Japanese
Korean
Persian
Polish
Portuguese
Russian
Spanish
Swedish
Turkish
Vietnamese
I think Andrew is wrong. Statistical Process Control is more or less essential to modern industrial production.
I don't think it would even begin to be possible to create much of current electronic components without statistical methods leading the way to huge improvements.
And for Christ's sake, how much would we know even about physics without statistics? Much of quantum theory just is statistics. And how would we know that the Higg's boson exists without statistics?
Everywhere in technology statistics is crucially important -- pulling signal out of noise in cell phone reception, in radar, in weather and just about a million other things involves statistics.
I can see the argument that statistics in politics -- Gelman's specialty -- produces little of value. Maybe it subtracts value. Certainly if we look at the average politician of today -- the product of a million polls and focus groups -- who can pretend they look and behave better than politicians even of 50 years ago?
No more insurance industry, statistical mechanics, IQ testing, and it would be a lot harder to argue with a leftist. As Charles Murray says, you can lie with statistics, but it is a lot easier to lie without them.
A world without hate facts and noticing wouldn’t be much different.
No evidence based medicine…
One area that certainly has been improved by applying statistics to ever-increasing amounts of data is investing.
Back in the day, people naively believed that a big-time investor or mutual fund manager that outperformed the Dow or S&P 500 must be doing so through skillful stock picking. But using statistics, computers and data, academics began to uncover that, outside of a few anomalies, the mutual fund manager’s skill was actually the fund tapping into various “risk factors,” such as small, value, momentum, etc, which weren’t found in the Dow or S&P 500.
(For example, value (cheap based on various measurements such as P/B, P/CF, etc.) stocks have historically outperformed growth (expensive) stocks by around 2 percentage points a year. Of course, the value premium doesn’t show up every year and often disappears or goes negative for multiple years, but, sooner or later, at least historically, it shows up and improves performance.)
In essence, some manager boasting about his outperforming the market through his idiosyncratic skill was really just buying value stocks.
By uncovering those risk factors, academics were able to show that paying these superstar investors huge fees (average managed mutual fund charges ~1.5% a year compared to 0.1% to 0.4% for index funds) was a waste. Investors could cheaply access these risk factors though low-cost index funds or ETFs. Heck, even Buffet’s performance largely been explained.
For your average (well, upper middle class) Joe, those savings easily could add up to hundreds of thousands of dollars over 25 years. It’s not flying cars, but it ain’t bad either.
Somehow the realization that “A World Without Statistics Wouldn’t be Much Different” is not particularly comforting.
Exactly. A better question than “would the world be different w/o statistics” is “would the world be better w/o statistics.” One consequence of getting rid of statistics would be getting rid of the pseudo-scientific plagues of randomized controlled trials, pyramids of evidence, and evidence-based medicine.
Rutherford said “If your experiment needs statistics, you ought to have done a better experiment.”
Great column, Steve
The author fabricated the quotation or Prof. Whodat is jerking your chain.
I agree with you about this.
I don’t think it would even begin to be possible to create much of current electronic components without statistical methods leading the way to huge improvements.
And for Christ’s sake, how much would we know even about physics without statistics? Much of quantum theory just is statistics. And how would we know that the Higg’s boson exists without statistics?
Everywhere in technology statistics is crucially important — pulling signal out of noise in cell phone reception, in radar, in weather and just about a million other things involves statistics.
I can see the argument that statistics in politics — Gelman’s specialty — produces little of value. Maybe it subtracts value. Certainly if we look at the average politician of today — the product of a million polls and focus groups — who can pretend they look and behave better than politicians even of 50 years ago?
Do people even know what statistics are?
Hint, it’s a branch of mathematics, and much more than just columns of numbers in a newspaper article justifying a point of view. You might as well ask about a world without mathematics.
Yes, statistics and the theory of probability are very interesting branches of mathematics with many fascinating and counterintuitive results. It’s also interesting that it took so long to develop. There seems to have been little notion, at least in the West, of a quantitative notion of probability before Pascal and Fermat. Some years ago Springer published a volume of their correspondence in which one can see the fundamental ideas of probability emerging. Fermat, who was generally not much interested in his contemporaries, was fascinated by Pascal and attempted to lure him down to Toulouse but without success. The two never met in person.
Best of the Web Today had a good CBC headline to complement your piece:
http://www.cbc.ca/news/business/plane-crashes-are-more-survivable-than-ever-1.2722089
Basically, the most common of events are commonly understood without experts
Statistics are often erroneously practiced. In fact, one could say that part of science today is largely broken. Many scientific studies use the wrong statistical tools. I’ve reach a conclusion that any scientific study should have a theoretical statistician involved in the set-up of the study, and provide the statistical results. Scientist are prone to using plug and play computer statistical programs. Unfortunately, many times, even the assumed distributions are erroneous for the application. And most egregious, the researchers don’t stick to the original sampling plan. They keep sampling until they get the statistical result that they want.
And here is an article that points out other wide-spread statistical mis-behaviors.
http://www.economist.com/news/briefing/21588057-scientists-think-science-self-correcting-alarming-degree-it-not-trouble
To answer the question, of course statistics has made a huge difference in the world, as other commenters have already said and gave examples. As someone else mentioned, statistical quality control is a huge one. Determining the efficacy of medicines is another. Risk analysis and mitigation, Cost estimating, Safety analysis, on and on.
My old stats prof said that we really don’t know how many people were out of work during the Depression, we just know that a lot of people were out of work during the Depression. He started his career as a statistician during the Depression.
Gelman’s right. We had a Depression. It was bad. It doesn’t matter how many people were out of work, it was very depressing.
One of your commenters at takimag made the good point that a few simple but rarely grasped statistical ideas cut through a bunch of leftist pieties. The bad news, though, is that statistical analysis rarely if ever changes minds on hot political controversies.
Gelman could be fascinating in that he does not know how much he does not know but he tries to measure how close he is to either not knowing what he does not know or how close he is to knowing something. On the other hand I have read his blog and it is usually boring because he always seems to discover that his previous liberal hunches were correct. I mean, how statistically likely is it that, say, a Mets fan will delve into baseball stats and always discover that the Mets should have won many more World Series than they did? Anyway this blog post leads me to believe he might be stepping up his game, as did, say, at his age, Saint Saens, Roth, Waugh, St Dismas, Eric Rohmer, and Mamet. Maybe I will check his blog in a year to see if he has decided that honesty and truth are friends.
You boys just try making through day without considering, consciously or unconsciously, the female vital statistic 36-24-26.
Now let me go, I have to adjust my Higgs Boson. No peeking.
Epidemology
http://en.wikipedia.org/wiki/Epidemiology
John Snow is famous for his investigations into the causes of the 19th century cholera epidemics, and is also known as the father of (modern) epidemiology.[13][14] He began with noticing the significantly higher death rates in two areas supplied by Southwark Company. His identification of the Broad Street pump as the cause of the Soho epidemic is considered the classic example of epidemiology. Snow used chlorine in an attempt to clean the water and removed the handle; this ended the outbreak. This has been perceived as a major event in the history of public health and regarded as the founding event of the science of epidemiology, having helped shape public health policies around the world.[15][16] However, Snow’s research and preventive measures to avoid further outbreaks were not fully accepted or put into practice until after his death.
see also
http://en.wikipedia.org/wiki/John_Snow_(physician)#Cholera
Steve, in response to a comment you made on Gelman’s blog,
The Greeks were serious about animal breeding.
The Greeks probably did scientific agriculture. Almost all Greek science is lost. All we have left is what the Romans understood; in the case of agriculture, what Varro understood. But he didn’t understand much. He says that he read 50 books of Greek agriculture and it was mostly “philosophy” so he threw it out.
”One of your commenters at takimag made the good point that a few simple but rarely grasped statistical ideas cut through a bunch of leftist pieties.”
The same statistical ideas also demolish the notion of equal justice, and the once oft-cited phrase that America is great because even the penniless and downtrodden can become successful. It is true that poor people with access to cut-rate lawyers can and do win their cases, and that those born in poverty can and do become wealthy, but when you calculate the odds of that happening, things seem much grimmer.
The left is all over statistics when it comes to measuring inequality. They just have a blind spot when it comes to understanding what (other than an insufficiently redistributive income tax) might be the cause of it .
From your Takimag article “If they really needed to know the area under a curve, they would cut out little rectangles of graph paper with scissors.”
Years ago I read somewhere (in a math history book, I believe) that Galileo calculated areas under a curve by making traces of the curve on, say a hundred, sheets of paper, then cutting all the sheets along the curve, then measuring the weight of all the parts of the sheets corresponding to the area under the curve together on a scale. If he calibrated his scale and accurately weighed the paper before and after cutting it, he could calculate areas accurately. Sort of a balance scale used as analog computer.
Does anyone know anything more about this? What is the source for this story? I’ve never seen anything else about it, though I haven’t really looked, and it seems curious. I’m sure I saw a claim that Galileo actually did this and that it wasn’t just an exercise or example.
Years ago I read somewhere (in a math history book, I believe) that Galileo calculated areas under a curve by making traces of the curve on, say a hundred, sheets of paper, then cutting all the sheets along the curve, then measuring the weight of all the parts of the sheets corresponding to the area under the curve together on a scale. If he calibrated his scale and accurately weighed the paper before and after cutting it, he could calculate areas accurately. Sort of a balance scale used as analog computer.
Does anyone know anything more about this? What is the source for this story? I've never seen anything else about it, though I haven't really looked, and it seems curious. I'm sure I saw a claim that Galileo actually did this and that it wasn't just an exercise or example.Replies: @Steve Sailer
The video I linked to at the word scissors shows weighing your cut paper to calculate the area under the curve. But graph paper makes that less necessary.
The Statistical Revolution in baseball really hasn’t made as much difference as many people think.
Look, if you asked a bunch of old-timers who know nothing of WAR, OPS, ERA+ or WHIP “Name the three greatest hitters of all time,” they’d probably say “Babe Ruth, Ted Williams and Lou Gehrig.” And even the modern number-crunchers would have to say, “Yeah, you’re pretty much right.”
The numbers have made a difference, but only at the margins. The sabremetricians have never proven that Willie Mays actually sucked or that Mario Mendoza was actually a Hall of Famer. At best, they’ve proven that SOME players who were widely regarded as great were merely very good, or that some players widely perceived as very good were actually great.
Of course, there is a handful of stars in MLB (currently, a guy like Mike Trout), and if a team like the Yankees or Boston can acquire enough of them, then they will be competitive. They won't necessarily win the World Series, which has a significant element of luck - but they will likely be in the playoffs, which is damned difficult.
For the rest, success or failure is determined by 'replacement-level' players. Few teams have the resources to sign expensive talent, and thus of the (what are there now, 30) teams in the league who are competing for the second-tier players, being able to see that Benito Santiago is actually a poor player, and Scott Hatteburg a decent one, can make all the difference.
“The video I linked to at the word scissors shows weighing your cut paper to calculate the area under the curve.”
Did that link get cut from the article somewhere along the line? I can’t find it, but then I’ve been blind more than once in this life…
I’d also still be interested to know if Galileo is known for sure to have used this technique, if he originated it, or if it has an earlier history.
I suppose you could call this statistics (state-istics, a la epidemiology):
“Groundbreaking research maps cultural history by mapping mobility of notable intellectuals”, ScienceDaily, July 31, 2014:
“New research from Northeastern University has mapped the intellectual migration network in North America and Europe over a 2,000-year span. The team of network scientists used the birth and death locations of more than 150,000 intellectuals to map their mobility patterns in order to identify the major cultural centers on the two continents over two millennia…
…the distance between the birth and death locations of notable individuals has not increased much over the span of eight centuries… “The observed rapid changes offer a fascinating view of the transience of intellectual supremacy.”
…broke new ground in terms of its data-driven approach to understanding cultural history. …
…relied on large data sets, including the curated General Artist Lexicon…”
"Groundbreaking research maps cultural history by mapping mobility of notable intellectuals", ScienceDaily, July 31, 2014:
"New research from Northeastern University has mapped the intellectual migration network in North America and Europe over a 2,000-year span. The team of network scientists used the birth and death locations of more than 150,000 intellectuals to map their mobility patterns in order to identify the major cultural centers on the two continents over two millennia...
...the distance between the birth and death locations of notable individuals has not increased much over the span of eight centuries... "The observed rapid changes offer a fascinating view of the transience of intellectual supremacy."
...broke new ground in terms of its data-driven approach to understanding cultural history. ...
...relied on large data sets, including the curated General Artist Lexicon..."Replies: @Steve Sailer
The 1920s American novelists like Hemingway and Fitzgerald and the 1930s New Yorker journalists like Thurber and Liebling all spent time after WWI in Paris. The house style of the New Yorker was a distant descendant of 17th Century French prose style.
Look, if you asked a bunch of old-timers who know nothing of WAR, OPS, ERA+ or WHIP "Name the three greatest hitters of all time," they'd probably say "Babe Ruth, Ted Williams and Lou Gehrig." And even the modern number-crunchers would have to say, "Yeah, you're pretty much right."
The numbers have made a difference, but only at the margins. The sabremetricians have never proven that Willie Mays actually sucked or that Mario Mendoza was actually a Hall of Famer. At best, they've proven that SOME players who were widely regarded as great were merely very good, or that some players widely perceived as very good were actually great.Replies: @DWB
But the point of using so-called “sabremetrics” in baseball is not really to prove that Ted Williams was a better player than Mario Mendoza. Rather, the point is about operating at the margins.
Of course, there is a handful of stars in MLB (currently, a guy like Mike Trout), and if a team like the Yankees or Boston can acquire enough of them, then they will be competitive. They won’t necessarily win the World Series, which has a significant element of luck – but they will likely be in the playoffs, which is damned difficult.
For the rest, success or failure is determined by ‘replacement-level’ players. Few teams have the resources to sign expensive talent, and thus of the (what are there now, 30) teams in the league who are competing for the second-tier players, being able to see that Benito Santiago is actually a poor player, and Scott Hatteburg a decent one, can make all the difference.