Pawel Sobkowicz writes:
How many zombies do you know?’ Using indirect survey methods to measure alien attacks and outbreaks of the undead, Arxiv preprint arXiv:1003.6087, 2010
I hope you would find interesting the following paper, recently posted on arXiv:
Aliens on Earth. Are reports of close encounters correct?, arXiv:1203.6805
This is soooooo much better than getting links to bad graphs or to papers on sex ratios!
I saw an analysis recently that I didn’t like. I won’t go into the details, but basically it was a dose-response inference, where a continuous exposure was binned into three broad categories (terciles of the data) and the probability of an adverse event was computed for each tercile. The effect and the sample size was large enough that the terciles were statistically-significantly different from each other in probability of adverse event, with the probabilities increasing from low to mid to high exposure, as one would predict.
I didn’t like this analysis because it is equivalent to fitting a step function. There is a tendency for people to interpret the (arbitrary) tercile boundaries as being meaningful thresholds even though the underlying dose-response relation has to be continuous. I’d prefer to start with a linear model and then add nonlinearity from there with a spline or whatever.
At this point I stepped back and thought: Hey, the divide-into-three analysis does not literally assume a step function. It doesn’t assume anything at all; it’s just a data summary! People discretize input variables all the time! So why am I complaining?
I justify my complaints on two levels. First on the grounds of interpretation: my applied colleagues really were interpreting the three-category model in terms of thresholds. The three categories were: “0 to A”, “A to B”, and “B to infinity”. And somebody really was saying something about the effect of exposure A or exposure B. Which just ain’t right.
My second issue is statistical efficiency. You can say that the categorical-input model is nothing but a summary, an estimate of averages—but by binning like this, you lose statistical efficiency. And you become the slave to “statistical significance”; there’s the temptation to butcher your analysis and throw away tons of information, just so you can get a single clean, statistically significant result.
P.S. The more categories you have, the less of a concern it is to discretize. And sometimes your data come in discrete form (see here, for example).
In a single day, New York City obtained two data analysis/statistics/machine learning organizations:
New York already has Facebook’s engineering unit, Twitter’s East Coast headquarters, and Google’s second-largest engineering office.
The data community here is on an upswing, and it might be one of the best places to be if you’re into applied statistics, machine learning or data analysis.
Post by Aleks Jakulin.
P.S. (from Andrew): The formerly-Yahoo-now-Microsoft researchers have a more-or-less formal connection to Columbia, through the Applied Statistics Center, where some of them will be organizing occasional mini-conferences and workshops!
US National Academy of Sciences elects 84 new members
(Please click through and read the whole thing.)
What we did in our Stan meeting yesterday:
Some discussion of revision of the Nuts paper, some conversations about parameterizations of categorical-data models, plans for the R interface, blah blah blah.
But also, I had two exciting new ideas!
Google Translate for code
Wouldn’t it be great if Google Translate could work on computer languages? I suggested this and somebody said that it might be a problem because code isn’t always translatable. But that doesn’t worry so much. Google Translate for human languages isn’t perfect either but it’s a useful guide. If I want to write a message to someone in French or Spanish or Dutch, I wouldn’t just write it in English and run it through Translate. What I do is try my best to write it in the desired language, but I can try out some tricky words or phrases in the translator. Or, if I start by translating, I go back and forth to make sure it all makes sense.
An R help-list bot
We were talking about how to build a Stan community that will be helpful to a diverse range of users without taking up too much of our time, and that’s when I came up with a brilliant idea. Let’s take a successful existing help group—for example, the R-help mailing list—then make a database of the helpful bits of advice of a distinguished and frequent contributor to the list. The bot would be easy: whatever the question is that comes in, just send back a random tip. I have a feeling that advice such as “PLEASE do, and not send HTML” and “My guess is that this is a Mac-specific question (e.g. you are using the R.app GUI), so please consider if this is the appropriate list” and “The posting guide was not followed” and “Please use the R-devel list to comment on current development versions” would work pretty well for almost any question (maybe after some global sub of Stan for R).
It’s sort of like when we were kids and had this book, Bennett Cerf’s Book of Riddles—it was a gray with a picture of a big red rock-eater on the cover. We started out by just reading through the riddles one at a time but we had more fun after inventing a game where we’d open the book and read a riddle, then open again at random to give the answer. For example:
Q: What’s big, red and eats rocks?
A: Two cats stuck in a tree!
My friend Seth, whom I know from Berkeley (we taught a course together on left-handedness), has a blog on topics ranging from thoughtful discussions of scientific evidence, to experiences with his unconventional weight-loss scheme, offbeat self-experimentation, and advocacy of fringe scientific theories, leavened with occasional dollops of cynicism and political extremism. I agree with Seth on some things but not others. (Here’s Seth’s reason for not attempting a clinical trial of his diet.)
Recently I was disturbed (but, I’m sorry to say, not surprised) to see Seth post the following:
Predictions of climate models versus reality. I [Seth] have only seen careful prediction-vs-reality comparisons made by AGW [anthropogenic global warming] skeptics. Those who believe humans are dangerously warming the planet appear to be silent on this subject.
In response, Phil commented:
Funny, on the day you [Seth] made your post saying that you haven’t seen comparisons between models and predictions except by skeptics, the top entry on RealClimate, the single most prominent global-warming-related blog that is not run by skeptics, was “Evaluating a 1981 temperature projection.”
Pretty amazing, huh? On its face it would seem surprising to claim that the majority of leading climate scientists don’t do “careful prediction-vs-reality comparisons,” and indeed on the very day of Seth’s post, there is such a comparison right there on the first place you might look for what the climate scientists are doing!
How did Seth miss it?
A clue comes from the sources on which Seth relies. His link above points to the webpage of the Von Mises Institute, a political advocacy organization. Other Seth links to global warming stories have come from an climate change skeptic blog, a right-leaning politics blog, another climate change skeptic blog, another advocacy organization, one more climate change skeptic blog, a letter to the Wall Street Journal, yet another climate change skeptic blog, an op-ed in Forbes magazine, a lecture by science writer and political activist Matt Ridley, another blog that seems to specialize in climate change skepticism, conservative political columnist Jeff Jacoby, a Wall Street Journal op-ed, still one more climate change skeptic blog, a conservative religious magazine, . . . ummm, you get the idea.
If these are your sources, you can get a distorted view of the opinions and arguments of “those who believe humans are dangerously warming the planet”! Any one or two or three of the above sources might be informative, but it doesn’t make sense to only look there. (Quick: glance again at the list of sources in the above paragraph.)
Continue reading ‘Selection bias, or, How you can think the experts don’t check their models, if you simply don’t look at what the experts actually are doing’ »
Upon reading this, Susan remarked, “Don’t you think it’s interesting that a guy who promotes smoking has a last name of ‘Huff’? Reminds me of the Dennis/Dentist studies.”
Good point.
P.S. As discussed in the linked thread, the great statistician R. A. Fisher was notorious for minimizing the risks of smoking. How does this connect to Fisher’s name, one might ask?
Thomas Basbøll [yes, I've learned how to smoothly do this using alt-o] gives some writing advice:
What gives a text presence is our commitment to asserting facts. We have to face the possibility that we may be wrong about them resolutely, and we do this by writing about them as though we are right.
This and an earlier remark by Basbøll are closely related in my mind to predictive model checking and to Bayesian statistics: we make strong assumptions and then engage the data and the assumptions in a dialogue: assumptions + data -> inference, and we can then compare the inference to the data which can reveal problems with our model (or problems with the data, but that’s really problems with the model too, in this case problems with the model for the data).
I like the idea that a condition for a story to be useful is that we put some belief into it. (One doesn’t put belief into a joke.) And also the converse, that thnking hard about a story and believing it can be the precondition to ultimately rejecting it because its implications don’t make sense. It’s like in chess: the way to refute a move is to consider making the move (which is as irrevocable in a chess context as believing a story is, in the context of basing a social-scientific theory on it.)
I’m also reminded of the advice from Pólya (or somebody like that) about solving math problems. If the question is, “Is statement A true?”, you can try to prove A or find a counterexample. But it’s hard to do both at the same time! Better to take a guess and go from there: if you try and try to prove it and fail, this may give insight into where to find a counterexample (in those folds of the problem that make A so hard to prove); conversely, if you can’t find a counterexample no matter how hard you look, you can try to systematize that search, thus perhaps leading to a proof that no counterexample exists.
Via Yalda Afshar, a 2005 paper by Hans-Hermann Dubben and Hans-Peter Beck-Bornholdt:
Publication bias is a well known phenomenon in clinical literature, in which positive results have a better chance of being published, are published earlier, and are published in journals with higher impact factors. Conclusions exclusively based on published studies, therefore, can be misleading. Selective under-reporting of research might be more widespread and more likely to have adverse consequences for patients than publication of deliberately falsified data. We investigated whether there is preferential publication of positive papers on publication bias.
They conclude, “We found no evidence of publication bias in reports on publication bias.” But of course that’s the sort of finding regarding publication bias of findings on publication bias that you’d expect would get published. What we really need is a careful meta-analysis to estimate the level of publication bias in studies of publication bias of studies of publication bias.
Daniel Murrell writes:
I see you have a blog post about turing chess . . . I’ve seen another reference to it but am unable to find a definitive source. Do you know of a source where I could find out about the history of the idea?
My reply:
You mean the run-around-the-house thing? I don’t know where it comes from. It’s a well known story, if you google Turing chess run around the house you can find lots of references but I don’t know the definitive source. I can blog and see if anything comes up!
I’ve never actually played the game. I’ll try it outdoors sometime, perhaps. When I last posted on the topic, we had a fun discussion, revealing that the rules are not as clear as one might think. It makes me wonder if anyone’s thought hard about it and come up with a good set of “official rules.”
Any thoughts?
This post is by Phil.
Psychologists perform experiments on Canadian undergraduate psychology students and draws conclusions that (they believe) apply to humans in general; they publish in Science. A drug company decides to embark on additional trials that will cost tens of millions of dollars based on the results of a careful double-blind study….whose patients are all volunteers from two hospitals. A movie studio holds 9 screenings of a new movie for volunteer viewers and, based on their survey responses, decides to spend another $8 million to re-shoot the ending. A researcher interested in the effect of ventilation on worker performance conducts a months-long study in which ventilation levels are varied and worker performance is monitored…in a single building.
In almost all fields of research, most studies are based on convenience samples, or on random samples from a larger population that is itself a convenience sample. The paragraph above gives just a few examples. The benefits of carefully conducted randomized trials are well known, but so are the costs and impediments. Lucky people studying some natural phenomena like solar output or earthquakes can deal with complete datasets, but for most data analysts and applied statisticians the fact that your data are not a random sample from your population of interest is so commonplace that it usually goes without saying. This does not mean that it goes without thinking, of course: most researchers, and all good ones, think about the extent to which their results might or might not be applicable to a wider population and try to frame their conclusions accordingly. But most or all researchers are willing to extrapolate their results to wider populations to some degree. The movie studio reshoots their ending not because they want to please their 9 test audiences, but because they think that the response of those 9 test audiences tells them something about millions of other viewers, even though those 9 audiences were not selected according to a careful, randomized sampling scheme.
If you think everything I’ve said so far is so obvious as to be boring, so did I, but I was proven wrong. Read on.
Continue reading ‘We go to war with the data we have, not the data we want’ »
Cassie Murdoch points to a report from a corporate survey:
Sixty-two percent of U.S. employees say it’s not likely they or a family member will be diagnosed with a serious illness like cancer, a survey indicates.
The Aflac WorkForces Report, a survey of nearly 1,900 benefits decision-makers and more than 6,100 U.S. workers, also indicated 55 percent said they were not very or not at all likely to be diagnosed with a chronic illness, such as heart disease or diabetes.
Here are some actual statistics:
The American Cancer Society, Cancer Facts & Figures 2012, said 1-in-3 women and 1-in-2 men will be diagnosed with cancer at some point in their lives, and the National Safety Council, Injury Facts 2011 edition, says more than 38.9 million injuries occur in a year requiring medical treatment.
The American Heart Association, Heart Disease & Stroke Statistics 2012, said 1-in-6 U.S. deaths were caused by coronary heart disease, Tillman said.
And some details on the survey:
The survey conducted in January and February by Research Now. The first 3,151 worker interviews were nationally representative, while the remaining 3,000 interviews were conducted among the Top 30 designated market areas.
Did these people really say they that neither they nor a family member will have a serious illness? Is this for real? What were they thinking? I’m used to seeing wacky survey findings, but this one is ridiculous.
Adrien Friggeri has a lovely visualization of US Senators movement between clusters:
You have to click the image and play with it to appreciate it. The methodology isn’t yet published – but I can see how this could be very illuminating. The dynamic clustering aspect hasn’t been researched much – one of the notable pieces is the Blei and Lafferty dynamic topic model of Science.
I did a static analysis of the US Senate back in 2005 with Wray Buntine and coauthors. Some additional visualizations and the source code are here. We did a dynamic analysis of US Supreme Court on this blog but there’s also a paper.
My knowledge on this topic is out of date, however. Who has been doing good work in this area? I’ll organize the links.
[added 4/29/12, via Edo Airoldi]: Visualizing the Evolution of Community Structures in
Dynamic Social Networks by Khairi Reda et al (2011) [PDF].
[added 4/29/12, via Allen Riddell] Joint Analysis of Time-Evolving Binary Matrices and Associated Documents by Eric Wang et al (2010) [PDF] [Video]
David Hogg writes:
My (now deceased) collaborator and guru in all things inference, Sam Roweis, used to emphasize to me that we should evaluate models in the data space — not the parameter space — because models are always effectively “effective” and not really, fundamentally true. Or, in other words, models should be compared in the space of their predictions, not in the space of their parameters (the parameters didn’t really “exist” at all for Sam). In that spirit, when we estimate the effectiveness of a MCMC method or tuning — by autocorrelation time or ESJD or anything else — shouldn’t we be looking at the changes in the model predictions over time, rather than the changes in the parameters over time? That is, the autocorrelation time should be the autocorrelation time in what the model (at the walker position) predicts for the data, and the ESJD should be the expected squared jump distance in what the model predicts for the data? This might resolve the concern I expressed a few months ago to you that the ESJD is not affine-invariant, and etc. Thoughts?
Hogg continues with an example:
Imagine you have a three-planet model for some radial velocity data. In the naivest implementation, you have a three-factorial exact degeneracy from swapping planets, but the modes are very well separated in parameter space: Your autocorrelation time in the parameters is essentially infinite (because you will never switch from one permutation of the planets to another, realistically), but in the predictions the autocorrelation time is finite and fine.
My reply:
It depends on the context. Sometimes we have a redundant parameterization in which the individual parameters are not identified, but predictions are well-identified. For a simple example, suppose you have a model, y ~ N (a+b, 1), with a uniform prior distribution on (a,b). Then your data don’t tell you anything about a or b, but you can get good inference for a+b and good predictions for new data from the same model. On the other hand, if you want to make a prediction for new data z ~ N(a,1), you’re out of luck.
More generally, one problem I have with the hard-line predictivist stance—the idea that models and parameters are mere fictions whereas predictions are real—is that models and parameters can be thought of as bridges between the data of yesterday and the data of tomorrow. Consider the speed of light. It’s not just part of a prediction for some particular measurement. It’s also a universal constant. For a more humble example, consider our discussion of physiologically-based pharmacokinetics models in Section 4.3 of my article with Bois and Jiang. In a Bayesian model, good parameterization can be important, as it is typically through the parameters that we put in prior information. In many ways, the parameterization represents a key source of prior information.
Remember How to Lie With Statistics? It turns out that the author worked for the cigarette companies. John Mashey points to this, from Robert Proctor’s book, “Golden Holocaust: Origins of the Cigarette Catastrophe and the Case for Abolition”:
Darrell Huff, author of the wildly popular (and aptly named) How to Lie With Statistics, was paid to testify before Congress in the 1950s and then again in the 1960s, with the assigned task of ridiculing any notion of a cigarette-disease link. On March 22, 1965, Huff testified at hearings on cigarette labeling and advertising, accusing the recent Surgeon General’s report of myriad failures and “fallacies.” Huff peppered his attack with with amusing asides and anecdotes, lampooning spurious correlations like that between the size of Dutch families and the number of storks nesting on rooftops–which proves not that storks bring babies but rather that people with large families tend to have larger houses (which therefore attract more storks).
This was all a surprise to me, and I suspect to other statisticians as well. For example, Huff’s activities with the cigarette companies are not mentioned on his Wikipedia page (as of 17 Apr 2012), nor are they mentioned in an article on Huff by probabilist J. Michael Steele from 2005.
“How to Lie with Smoking Statistics”
Darrell Huff is best known for How to Lie with Statistics but he wrote or cowrote several other books, including Pictures by Pete (1944), The Dog that Came True (1946), How to Take a Chance (1959), Score: The Strategy of Taking Tests (1961), Cycles in Your Life (1964), How to be the Parent of a Successful Creative Child (1968), Twenty Careers of Tomorrow (1945), and How to Lower Your Food Bills (1963).
It appears that in the late 1960s he was also working on a book called “How to Lie with Smoking Statistics,” which the publisher saw “high likelihood of proceeding into print.”
In November 1965, a letter was sent to Huff as follows:
Here‘s a letter from 1967 where Huff asks the tobacco dudes for another $1500 to keep writing.
And here‘s a letter from mid-1968 from Huff’s publisher, Macmillan:
But publication “as soon as possible” never seems to have occurred.
What happened? In one of these documents, William Kloepfer, vice president for public relations for the Tobacco Institute, wrote of the manuscript, “Frankly, this mass of verbiage needs drastic editing before it will directly address itself to the needs of our industry.”
After glancing at a couple of sections from the draft, I gotta say that William Kloepfer had a point: “mass of verbiage” is a pretty good description! Huff’s book chapter reads like a bad sitcom where the writers were too lazy to put together enough material and they just milk the same couple of jokes over and over again.
In retrospect, I think Huff really dodged a bullet on this one. If “How to Lie with Smoking Statistics” had come out, I expect it would’ve destroyed his reputation—remember, we’re talking 1969 here, that’s five years after the Surgeon General’s report—and taken a big bite out of the later sales and reputation of his 1954 bestseller.
How sincere was Huff? Did he tank his book for strategic reasons?
I wanted to call this post, How to Lie With “How to Lie With Statistics,” but to be fair I have no reason to believe that Huff was lying or intentionally deceiving in his testimony. He may well have simply been misleading himself in analogizing research on the effects of smoking to silly things like studies of storks and babies. And if he was sincere in his views, I can hardly fault him for collecting some money for his efforts.
On the other hand, this document makes me think that Huff may have seen his role as producing talking points in support of a predetermined conclusion. I guess we’ll never know if he really wanted to publish How to Lie with Smoking Statistics. Maybe he intentionally sabotaged it because he sensed it would ruin his reputation, whereas it was possible for him to keep the consulting and testimony under the radar.
Sociologist Fabio Rojas reports on “a conversation I [Rojas] have had a few times with statisticians”:
Rojas: “What does your research tell us about a sample of, say, a few hundred cases?”
Statistician: “That’s not important. My result works as n–> 00.”
Rojas: “Sure, that’s a fine mathematical result, but I have to estimate the model with, like, totally finite data. I need inference, not limits. Maybe the estimate doesn’t work out so well for small n.”
Statistician: “Sure, but if you have a few million cases, it’ll work in the limit.”
Rojas: “Whoa. Have you ever collected, like, real world network data? A million cases is hard to get.”
The conversation continues in this frustrating vein. Rojas writes:
This illustrates a fundamental issue in statistics (and other sciences). One you formalize a model and work mathematically, you are tempted to focus on what is mathematically interesting instead of the underlying problem motivating the science. . . .
We have the same issue in statistics. “Statistics” can mean “the mathematics of distributions and other functions arising in statistical models.” Or it can mean the traditional problems of statistics like inference, measurement, model estimation, sampling, data collection/management, forecasting, and description. The problem for a guy like me (a social scientist with real data) is that the label “statistician” often denotes someone who is actually a mathematician who happens to be interested in distributions. . . . What I really want is a nuts and bolts person to help me solve problems.
My first reaction—actually, my main reaction—is that Rojas hangs out with the wrong sort of statistician. Following the links, I see that Rojas works at Indiana University, which features a large statistics department. I suspect he had the misfortune to encounter “a mathematician who happens to be interested in distributions” and he didn’t realize he could shop around among the many statisticians in that department who work on applied social research.
On the other hand, it’s a bad sign that Rojas reports having this conversation multiple times. I thought that statisticians nowadays know they’re supposed to be helpful on real problems. That “n -> infinity” thing seems so old-fashioned! I’d like to believe that Rojas was just having some bad luck, but maybe there’s more of this bad stuff going on than I realized. Or maybe it was just a communication problem?
It’s hard for me to imagine a statistician in 2012 telling a sociologist, “if you have a few million cases, it’ll work in the limit,” except as a joke, as an ironic comment on the limitations of some of our theory. But perhaps that just reflects the poverty of my imagination.
Andre de Boer writes:
In my profession as a risk manager I encountered this graph:
I can’t figure out what kind of regression this is, would you be so kind to enlighten me?
The points represent (maturity,yield) of bonds.
My reply: That’s a fun problem, reverse-engineering a curve fit! My first guess is lowess, although it seems too flat and asympoty on the right side of the graph to be lowess. Maybe a Gaussian process? Looks too smooth to be a spline. I guess I’ll go with my original guess, on the theory that lowess is the most accessible smoother out there, and if someone fit something much more complicated they’d make more of a big deal about it. On the other hand, if the curve is an automatic output of some software (Excel? Stata?) then it could be just about anything.
Does anyone have any ideas?
Peter Woit reports on the sympathy that well-known physicist Freeman Dyson has with crackpot theorists. The interesting part is that Dyson has positive feelings for these cranks, even while believing that their theories are completely wrong:
In my [Dyson's] career as a scientist, I twice had the good fortune to be a personal friend of a famous dissident. One dissident, Sir Arthur Eddington, was an insider like Thomson and Tait. The other, Immanuel Velikovsky, was an outsider like Carter. Both of them were tragic figures, intellectually brilliant and morally courageous, with the same fatal flaw as Carter. Both of them were possessed by fantasies that people with ordinary common sense could recognize as nonsense. I made it clear to both that I did not believe their fantasies, but I admired them as human beings and as imaginative artists. I admired them most of all for their stubborn refusal to remain silent. With the whole world against them, they remained true to their beliefs. I could not pretend to agree with them, but I could give them my moral support.
Dyson first writs about Eddington. I agree with Peter Woit that “this sympathy for a great physicist who headed down a wrong path in his later years is easy to understand, but the case of Velikovsky is less so. Velikovsky was a well-known author of crackpot best-sellers starting in the 1950s . . . and a neighbor of Dyson’s in Princeton.”
Woit quotes what Dyson “wrote as a proposed blurb for Velikovsky in 1977″:
First, as a scientist, I [Dyson] disagree profoundly with many of the statements in your books. Second, as your friend, I disagree even more profoundly with those scientists who have tried to silence your voice. To me, you are no reincarnation of Copernicus or Galileo. You are a prophet in the tradition of William Blake, a man reviled and ridiculed by his contemporaries but now recognized as one of the greatest of English poets. A hundred and seventy years ago, Blake wrote: “The Enquiry in England is not whether a Man has Talents and Genius, but whether he is Passive and Polite and a Virtuous Ass and obedient to Noblemen’s Opinions in Art and Science. If he is, he is a Good Man. If not, he must be starved.” So you stand in good company. Blake, a buffoon to his enemies and an embarrassment to his friends, saw Earth and Heaven more clearly than any of them. Your poetic visions are as large as his and as deeply rooted in human experience. I am proud to be numbered among your friends.
Now back in 2012, Dyson writes:
Science is a creative interaction of observation with imagination. “Physics at the Fringe” is what happens when imagination loses touch with observation. Imagination by itself can still enlarge our vision when observation fails. The mythologies of Carter and Velikovsky fail to be science, but they are works of art and high imagining. As William Blake told us long ago, “You never know what is enough unless you know what is more than enough.”
I’m with Woit (I think) here. I don’t see the appeal of bad science. I don’t think such voices should be silenced (as Dyson puts it), but it’s probably a good thing to keep these theories off the nonfiction shelves. I see the appeal of poetry and literature and philosophy and all sorts of things that aren’t science, and I recognize that poets, philosophers, etc., can motivate themselves by all sorts of wacked-out theories. Look at Philip K. Dick. His visions are part of who he was, and I wouldn’t trade Valis for anything, but without the art the visions aren’t so exciting. My impression is that the problem with crackpot scientific theories is not that they are not beautiful but that they lead to no scientific progress or understanding. To put it another way: as a poet, Velikovsky does not have much to offer. It is only if his theories point toward scientific understanding that they have value. In contrast, Blake was an artists whose visions are appealing without any necessity for them to correspond to scientific reality.
To me, a good analogy would be with the fascinating “outsider art” done by schizophrenics, where an entire canvas is covered with tiny scribbles relating to the nature of the universe. These artworks can be just amazing and I don’t think anyone should try to silence their voices. But I don’t think it does anybody any favors to call this science.
I think my view is shared by most scientists. Dyson gets some attention here partly from his eminence and partly because of his contrary views. That’s fine—he’s done enough good work that he’s earned the right to have his ideas broadcast—but it all seems a bit odd to me. Whatever people think of William Blake’s scientific ideas now, he is admired as a poet and artist. Velikovsky is more of a historical footnote in the annals of past bestsellers, a pop-culture artifact who belongs with the Chariots of the Gods guy, the Jupiter Effect guy, the Bible Code guy, the people who made the Search for Noah’s Ark movie, etc etc. I don’t think anyone will be reading his books for the pleasure of his prose.
This is somebody’s dream job, I’m sure . . .
ESPN is looking for a statistician to join the HR department as a Research Analyst. The job will consist of analytical research and producing statistics about the people that work at ESPN. Topics of interest will include productivity, efficiency, and retention of employees, among other items. In addition to data mining and producing reports, we also field surveys and analyze results.
The position is located at the headquarters in Bristol, Connecticut, the same campus where nearly all ESPN shows are produced. ESPN is a Disney company, so discounts and free admission to Disney parks are available for employees. Flexible work arrangements are available, along with working in the New York City office part-time if desired.
The role is a relatively new function and will have a high impact very quickly on helping the business function. Statistical software, text books, and any other resource needed to get the job done will be provided.
The link for the application is below. Any interested candidates with questions can contact Michael.J.Springer@espn.com or the recruiter Amy.McManus@espn.com.
Econometrician and statistician Dale Poirier writes:
24 years ago (1988, Journal of Economics Perspectives) I [Poirier] noted cognitive dissonance among some economists who treat the agents in their theoretical framework as Bayesians, but then analyze the data (even in the same paper!) as a frequentist. Recently, I have found similar cases in cognitive science. I suspect other disciplines exhibit such behavior. Do you know of any examples in political science?
My reply:
I don’t know of any such examples in political science. Game theoretic models are popular in poli sci, but I haven’t seen much in the way of models of Bayesian decision making.
Here are two references (not in political science) that might be helpful.
1. I have argued that the utility model (popular in economics and political science as a way of providing “microfoundations” for analyses of aggregate behavior) is actually more of a bit of folk-psychology that should not be taken seriously. To me, it is silly that many economists and political scientists give this model such prominence. Utility theory can be a helpful normative model in many situations, but I don’t think it should be anything close to foundational as
2. Are you familiar with the work of Josh Tenenbaum? He is a cognitive scientist at MIT who has been working on Bayesian models for human reasoning and also Bayesian methods for fitting such models given data from psychological experiments. See here and here.
Getting back to Poirier’s original point, it could make complete sense to me to use non-Bayesian inference to learn about Bayesian agents, as long as you believe that (a) people’s behavior can be reasonably approximated by Bayesian decision rules, and (b) from a normative standpoint, non-Bayesian inference is to be preferred. It seems that many economists believe both a and b, so I don’t necessarily see any cognitive dissonance in using non-Bayesian statistical inference while modeling behavior as Bayesian.
The funny thing is, I believe not-A and not-B, so my preference would be to use Bayesian inference for non-Bayesian models of behavior.
This (by Frans Hofmeester) is excellent.
What really makes it work, I think, is that it goes slowly enough. 2 minutes and 45 seconds is enough time for me, as a viewer, to feel like I’m living through each stage of development. If the video were sped up to go from 0 to 12 in only 30 seconds, that would be cool in its own way but would give up the sense of local stability that is characteristic of development.
As a statistician I am particularly worried about the rhetorical power of anecdotes (even though I use them in my own reasoning; see discussion below). But much can be learned from a true anecdote. The rough edges—the places where the anecdote doesn’t fit your thesis—these are where you learn.
We have recently had a discussion (here and here) of Karl Weick, a prominent scholar of business management who plagiarized a story and then went on to draw different lessons from the pilfered anecdote in several different publications published over many years.
Setting aside an issues of plagiarism and rulebreaking, I argue that, by hiding the source of the story and changing its form, Weick and his management-science audience are losing their ability to get anything out of it beyond empty confirmation.
A full discussion follows.
1. The lost Hungarian soldiers
Thomas Basbøll (who has the unusual (to me) job of “writing consultant” at the Copenhagen Business School) has been writing in different places about the a story that has been making the rounds over the past few decades among organizational sociologists and management consultants. The story started with a discovery that of plagiarism by the eminent scholar Karl Weick but then moved toward a more general exploration of storytelling and belief. (I learned about this example via an email from Basbøll (who had become aware of my interest in plagiarism); it turns out we also have a common interest in the bases of scientific and scholarly ideas.)
From Basbøll’s latest and most historical telling (linked from here, via Basbøll’s blog), supplemented by Wikipedia, I summarize what happened in time order (which is somewhat ahistorical in that it does not represent the order in which Basbøll, and perhaps Weick, learned about these events):
Continue reading ‘“Any old map will do” meets “God is in every leaf of every tree”’ »
Pointing to some horrible graphs, Kaiser writes, “The Earth Institute needs a graphics adviser.”
I agree. The graphs are corporate standard, neither pretty or innovative enough to qualify as infographics, not informational enough to be good statistical data displays.
Some examples include the above exploding pie chart, which, as Kaiser notes, is not merely ugly and ridiculously difficult to read (given that it is conveying only nine data points) but also invites suspicion of its numbers, and pages and pages of graphs that could be better compressed into a compact displays (see pages 25-65 of the report). Yes, this is all better than tables of numbers, but I don’t see that much thought went into displaying patterns of information or telling a story. It’s more graph-as-data-dump.
To be fair, the report does have some a clean scatterplot (on page 65). But, overall, the graphs are not well-integrated with the messages in the text.
I feel a little bit bad about this, because I’m involved with the Earth Institute. I should be their graphics adviser! I’m actually surprised they didn’t ask me for advice on this. I gave a talk to the Earth Institute postdocs a year or so ago, so they should know I like graphs!
P.S. Nathan Yau reproduces the report’s Figure 11, which is just as bad as the exploding pie chart shown above.
