Results matching “R”

Gianluca Baio sends along this article (coauthored with Marta Blangiardo):

As someone who relies strongly on survey research, it's good for me to be reminded that some surveys are useful, some are useless, but one thing they almost all have in common is . . . they waste the respondents' time.

I thought of this after receiving the following email, which I shall reproduce here. My own comments appear after.

Really we need the data on babies born 30 years ago, but this is still pretty stunning:

Aaron Swartz writes the following, as a lead-in to an argument in favor of vegetarianism:

Burgess on Kipling

This is my last entry derived from Anthony Burgess's book reviews, and it'll be short. His review of Angus Wilson's "The Strange Ride of Rudyard Kipling: His Life and Works" is a wonderfully balanced little thing. Nothing incredibly deep--like most items in the collection, the review is only two pages long--but I give it credit for being a rare piece of Kipling criticism I've seen that (a) seriously engages with the politics, without (b) congratulating itself on bravely going against the fashions of the politically incorrect chattering classes by celebrating Kipling's magnificent achievement blah blah blah. Instead, Burgess shows respect for Kipling's work and puts it in historical, biographical, and literary context.

Burgess concludes that Wilson's book "reminds us, in John Gross's words, that Kipling 'remains a haunting, unsettling presence, with whom we still have to come to terms.' Still." Well put, and generous of Burgess to end his review with another's quote.

Other critics, most notably George Orwell and Edmund Wilson, have also written interestingly on Kipling,

My article with Daniel and Yair has recently appeared in The Forum:

We use multilevel modeling to estimate support for health-care reform by age, income, and state. Opposition to reform is concentrated among higher-income voters and those over 65. Attitudes do not vary much by state. Unfortunately, our poll data only go to 2004, but we suspect that much can be learned from the relative positions of different demographic groups and different states, despite swings in national opinion. We speculate on the political implications of these findings.

The article features some pretty graphs that originally appeared on the blog.

It's in a special issue on health care politics that has several interesting articles, among which I'd like to single out this one by Bob Shapiro and Lawrence Jacobs entitled, "Simulating Representation: Elite Mobilization and Political Power in Health Care Reform":

The public's core policy preferences have, for some time, favored expanding access to health insurance, regulating private insurers to ensure reliable coverage, and increasing certain taxes to pay for these programs. Yet the intensely divisive debate over reform generated several notable gaps between proposed policies and public opinion for two reasons.

First, Democratic policymakers and their supporters pushed for certain specific means for pursuing these broad policy goals--namely, mandates on individuals to obtain health insurance coverage and the imposition of an excise tax on high-end health insurance plans--that the public opposed. Second, core public support for reform flipped into majority opposition in reaction to carefully crafted messages aimed at frightening Americans and especially by partisan polarization that cued Republican voters into opposition while they unnerved independents.

The result, say Shapiro and Jacobs, "suggests a critical change in American democracy, originating in transformations at the elite level and involving, specifically, increased incentives to attempt to move the public in the direction of policy goals favored by elites policies and to rally their partisan base, rather than to respond to public wishes." They've written a fascinating and important paper.

Imputing count data

Guy asks:

I am analyzing an original survey of farmers in Uganda. I am hoping to use a battery of welfare proxy variables to create a single welfare index using PCA. I have quick question which I hope you can find time to address:

How do you recommend treating count data? (for example # of rooms, # of chickens, # of cows, # of radios)? In my dataset these variables are highly skewed with many responses at zero (which makes taking the natural log problematic). In the case of # of cows or chickens several obs have values in the hundreds.

My response: Here's what we do in our mi package in R. We split a variable into two parts: an indicator for whether it is positive, and the positive part. That is, y = u*v. Then u is binary and can be modeled using logisitc regression, and v can be modeled on the log scale. At the end you can round to the nearest integer if you want to avoid fractional values.

See here (if you care).

P.S. Kaus writes that, when he was on William Bennett's radio show, "Bennett immediately zeroed in on a key political mystery: Are African-American voters on board with the Democrats' recent amnesty-for-illegal-immigrants program?" I wonder if Kaus asked Bennett about this quote:

But I [Bennett] do know that it's true that if you wanted to reduce crime, you could -- if that were your sole purpose, you could abort every black baby in this country, and your crime rate would go down.

As Brendan Nyhan notes, Bennett wasn't actually suggesting that black babies be aborted--in fact, Bennett said, "That would be an impossible, ridiculous, and morally reprehensible thing to do, but your crime rate would go down." Bennett definitely sounds like the go-to guy for a savvy discussion of the black vote!

Andrew Mack writes:

There was a brief commentary from the Benetech folk on the Human Security Report Project's, "The Shrinking Costs of War" report on your blog in January.

But the report has since generated a lot of public controversy. Since the report--like the current discussion in your blog on Mike Spagat's new paper on Iraq--deals with controversies generated by survey-based excess death estimates, we thought your readers might be interested.

Our responses to the debate were posted on our website last week. "Shrinking Costs" had discussed the dramatic decline in death tolls from wartime violence since the end of World War II --and its causes. We also argued that deaths from war-exacerbated disease and malnutrition had declined. (The exec. summary is here.)

One of the most striking findings was that mortality rates (we used under-five mortality data) decline during most wars. Indeed our latest research indicates that of the total number of years that countries were involved in warfare between 1970 and 2008, the child mortality rate increases in only 5% of them. Les Roberts has strongly challenged these findings.

Somebody named David writes:

I [David] thought you might be interested or have an opinion on the paper referenced below. I am largely skeptical on the techniques presented and thought you might have some insight because you work with datasets more similar to those in 'social science' than myself.

Dana and Dawes. The superiority of simple alternatives to regression for social science predictions. Journal of Educational and Behavioral Statistics (2004) vol. 29 (3) pp. 317.

My reply: I read the abstract (available online) and it seemed reasonable to me. They prefer simple averages or weights based on correlations rather than regressions. From a Bayesian perspective, what they're saying is that least-squares regression and similar methods are noisy, and they can do better via massive simplification.

I've been a big fan of Robyn Dawes ever since reading his article in the classic Kahneman, Slovic, and Tversky volume. I have no idea how much Dawes knows about modern Bayesian statistics (that is, multilevel models), but if he does, I assume he'd support a partial-pooling approach that makes use of data information in determining weights while keeping stability in the estimates.

To put it another way, least squares regression won't help you make maps like these, but simple averaging won't either. At some point you have to step things up to the next level.

I just reviewed the book Bursts, by Albert-László Barabási, for Physics Today. But I had a lot more to say that couldn't fit into the magazine's 800-word limit. Here I'll reproduce what I sent to Physics Today, followed by my additional thoughts.

The back cover of Bursts book promises "a revolutionary new theory showing how we can predict human behavior." I wasn't fully convinced on that score, but the book does offer a well-written and thought-provoking window into author Albert-László Barabási's research in power laws and network theory.

Power laws--the mathematical pattern that little things are common and large things are rare--have been observed in many different domains, including incomes (as noted by economist Vilfredo Pareto in the nineteenth century), word frequencies (as noted by linguist George Zipf), city sizes, earthquakes, and virtually anything else that can be measured. In the mid-twentieth century, the mathematician Benoit Mandelbrot devoted an influential career to the study of self-similarity, deriving power laws for phenomena ranging from taxonomies (the distribution of the lengths of index entries) to geographical measurements. (I was surprised to encounter neither Zipf, Mandelbrot, nor Herbert Simon in the present book, but perhaps an excessive discussion of sources would have impeded the book's narrative flow.)

Mandlebrot made a convincing case that nature is best described, not by triangles, squares, and circles, but by fractals--patterns that reveal increasing complexity when they are studies at finer levels. The shapes familiar to us from high-school geometry lose all interest when studied close up, whereas fractals--and real-life objects such mountains, trees, and even galaxies--are full of structure at many different levels. (Recall the movie Powers of Ten, which I assume nearly all readers of this magazine have seen at least once in their lives.)

A similar distinction between regularity and fractality holds in the social world, with designed structures such as bus schedules having a smooth order, and actual distributions of bus waiting times (say) having a complex pattern of randomness.

Trained as a physicist, Albert-László Barabási has worked for several years on mathematical models for the emergence of power laws in complex systems such as the Internet. In his latest book, Barabási describes many aspects of power laws, including a computer simulation of busy responses that went like this:

a) I [Barbasi] selected the highest-priority task and removed it from the list, mimicking the real habit I have when I execute a task.
b) I replaced the executed task with a new one, randomly assigning it a priority, mimicking the fact that I do not know the importance of the next task that lands on my list.

The resulting simulation reproduced the power-law distribution that he and others have observed to characterize the waiting time between responses to emails, web visits, and other data. But this is more than a cute model to explain a heretofore-mysterious stylized fact. As with Albert Einstein's theory of Brownian motion, such latent-variable models suggest new directions of research, in this case moving from a static analysis of waiting time distributions to a dynamic study of the decisions that underlie the stochastic process.

For an application of this idea, Barabási discusses the "Harry Potter" phenomenon, in which hospital admissions in Britain were found to drop dramatically upon the release of each installment of the cult favorite children's book. A similar pattern happened in relation to Boston's professional baseball team: emergency-room visits in the city dropped when the Red Sox had winning days.

In addition to this sort of detail, Barabási makes some larger points, some of which are persuasive to me and some of which are not. He distinguishes between traditional models of randomness--Poisson and Gaussian distributions--which are based on statistically independent events, and bursty processes, which arise from feedback processes that at times suppress and at other times amplify variation. (A familiar example, not discussed in the book, is the system of financial instruments which shifted risk around for years before eventually blowing up.)

Barabási characterizes bursty processes as predictable; at one point he discusses the burstiness of people's physical locations (we spend most of our time at home, school or work, or in between, but occasionally go on long trips). From here, he takes a leap--which I couldn't follow at all--to conjecture a more general order within human behavior and historical events, in his opinion calling into question Karl Popper's argument that human history is inherently unpredictable. The book also features a long excursion into Hungarian history, but the connection of this narrative to the scientific themes was unclear to me.

Despite my skepticism of the book's larger claims, I found many of the stories in Bursts to be interesting and (paradoxically) unpredictable, and it offers an inside view on some fascinating research. I particularly liked how Barabási takes his models seriously enough that, when they fail, he learns even more from their refutation. I suspect there are quite a few more bursts to come in this particular research programme.

And now for my further thoughts, first on the structure of the book, then on some of its specific claims.

Structure

The book Bursts falls into what might be called the auto-Gladwell genre of expositions of a researcher's own work, told through stories and personal anecdotes in a magazine-article style, but ultimately focused on the underlying big idea. Auto-Gladwell is huge nowadays; other examples in the exact same subfield as Barabási's include Six Degrees (a book written by Duncan Watts, who was my Columbia colleague at the time, but which I actually first encountered it through a (serious) review in the Onion, of all places), Steven Strogatz's Sync, last year's Connected by Nicholas Christakis and James Fowler's, as well as, of course, Barabási's own Linked, published in 2002. These guys have collaborated with each other In different combinations, forming their own social network.

In keeping with the Gladwellian imperative, Bursts jumps around from story to story, often dropping the reader right into the middle of a narrative with little sense of how it connects to the big story. This makes for an interesting reading experience, but ultimately I'd be happier to see each story presented separately and to its conclusion. About half the book tells the story of a peasant rebellion in sixteenth-century Hungary (along with associated political maneuvering), but it's broken up into a dozen chapters spread throughout the book, and I had to keep flipping back and forth to follow what was going on. (Also, as I noted above, I didn't really see the connection between the story and Barabási's scientific material.)

Similarly, Barabási begins with, concludes with, and occasionally mentions a friend of his, an artist with a Muslim-sounding name who keeps being hassled by U.S. customs officials. It's an interesting story but does not benefit from being presented in bits and pieces. In other places, interesting ideas come up and are never resolved in the book. For example, chapter 15 features the story of a unified field theory proposed in 1919 (in Barabási's words, "the two forces [gravity and electromagnetism] could be brought together if we assume that our world is not three- but five-dimensional"; apparently it anticipated string theory by several decades) and published on Albert Einstein's recommendation in 1921. This is used as an example to introduce an analysis of Einstein's correspondence, but it's not clear to me exactly how the story relates to the larger themes of the book.

Specifics

As noted in the review above, I thought Barabási's dynamic model of priority-setting was fascinating, and I would've liked to hear more details--ideally, something more specific than stories and statements that people are 90% predictable, but less terse than what's in the papers that he and his collaborators have published in journals such as Science. On one hand, I can hardly blame the author for trying to make his book accessible to general audiences; still, I kept wanting more detail, to fill in the gaps and understand exactly how the mathematical models and statistical analyses fit into the stories and the larger claims.

My impression was that the book was making two scientific claims:

1. Bursty phenomena can often be explained by dynamic models of priority-setting.

2. In some fundamental way, bursty processes can be thought of as predictable and not so random. in particular, human behavior and even human history can perhaps be much more predictable than we thought?

How convincing in Barabási on these two claims? On the first claim, somewhat. His model is compelling to me, at least in the examples he focuses on. I think the presentation would've been stronger had he discussed the variety of different mathematical models that researchers have developed to explain power laws. Is Barabási's model better or more plausible than the others? Or perhaps all these different models have important common features that could be emphasized? I'd like to know.

Of Barabási's second claim, I'm not convinced at all. It seems like a big leap to go from being able to predict people's locations (mostly they're at work from 9-5 and home at other times) and forecasting human history as one might forecast the weather.

Also, I think Barabási is slightly misstating Karl Popper's position from The Poverty of Historicism. Barabási quotes Popper as saying that social patterns don't have the regularity of natural sciences--we can't predict revolutions like we can predict eclipses because societies, unlike planets, don't move in regular orbits. And, indeed, we still have difficulty predicting natural pheonomena such as earthquakes that do not occur on regular schedules.

But Popper was saying more than that. Popper had two other arguments against the predictability of social patterns. First, there is the feedback mechanism: if a prediction is made publicly, it can induce behavior that is intended to block or hasten the prediction. This sort of issue arises in economic policy nowadays. Second, social progress depends on the progress of science, and scientific progress is inherently unpredictable. (When the National Science Foundation gives our research group $200,000 to work on a project, there's no guarantee of success. In fact, they don't give money to projects with guaranteed success; that sort of thing isn't really scientific research at all.)

I agree with Barabási that questions of the predictability of individual and social behavior are ultimately empirical. Persuasive as Popper's arguments may be (and as relevant as they may have been when combatting mid-twentieth century Communist ideology), it might still be that modern scientists will be able to do it. But I think it's only fair to present Popper's full argument.

Finally, a few small points.

- On page 142 there is a discussion of Albert Einstein's letters: "His sudden fame had drastic consequences for his correspondence. In 1919, he received 252 letters and wrote 239, his life still in its subcritical phase . . . By 1920 Einstein had moved into the supercritical regime, and he never recovered. The peak came in 1953, two years before his death, when he received 832 letters and responded to 476 of them." This can't be right. Einstein must have been receiving zillions of letters in 1953.

- On page 194, it says, "It is tempting to see life as a crusade against randomness, a yearning for a safe, ordered existence." This echoes the famous idea from Schroedinger's What is Life, of living things as entropy pumps, islands of low-entropy systems within a larger world governed by the second law of thermodynamics.

- On page 195, Barabási refers to Chaoming Song, "a bright postdoctoral research associate who joined my lab in the spring of 2008." I hope that all his postdocs are bright!

- On page 199, he writes, "when it comes to the predicability of our actions, to our surprise power laws are replaced by Gaussians." This confused me. The distribution of waiting times can't be Gaussian, right? It would help to have some detail on exactly what is being measured here. I understand that, for accessibility reasons, the book has no graphs, but still it would be good to have a bit more information here.

As a New Yorker I think I'm obliged to pass on the occasional Jersey joke (most recently, this one, which annoyingly continues to attract spam comments). I'll let the above title be my comment on this entry from Tyler Cowen entitled, "Which Americans are 'best off'?":

If you consult human development indices the answer is Asians living in New Jersey. The standard is:
The index factors in life expectancy at birth, educational degree attainment among adults 25-years or older, school enrollment for people at least three years old and median annual gross personal earnings.

More generally, these sorts of rankings and ndexes seem to be cheap ways of grabbing headlines. This has always irritated me but really maybe I should go with the flow and invent a few of these indexes myself.

Advice to help the rich get richer

Tyler Cowen reviews a recent book, "Lifecyle Investing," by Ian Ayres and Barry Nalebuff, two professors of management at Yale. The book recommends that young adults take out loans to buy stocks and then hold these stocks for many years to prepare for retirement.

What I'm wondering is: What's the goal of writing this sort of book? The main audience has got to be young adults (and their parents) who are already pretty well fixed, financially. Students at Yale, for example. And the book must be intended for people who are already beyond the standard recommendations of personal-investment books (pay off your credit card debt, don't waste so much money on restaurant meals and fancy clothes, buy 1 used car instead of 2 new ones, etc). Basically it sounds like they're talking to people who have a lot of money but want to make sure that they retire rich rather than merely middle-class.

I can't say that I'm morally opposed to helping the rich get richer. After all, I'm not out there lobbying to cut my own salary, and I teach at Columbia, where I have no problem encouraging students to go into well-paying jobs in applied statistics.

But I can't really see what would motivate somebody to write a whole book on helping rich kids prepare for retirement. I can see how people might want to buy such a book, and how it might make economic sense to publish it, and how it could get reviewed in the New York Times, but who'd care enough about the topic to write the book in the first place? There must be something I'm missing here. (The book might sell well but I can't imagine it will make the authors rich, so I doubt it's simply a financial motivation. It's also hard for me to believe they're planning to use this book as a foundation for a lucrative consulting business, although maybe that's what they're thinking.)

P.S. I'm surprised that Cowen didn't remark on this aspect of the Ayres and Nalebuff book--although some of his commenters did. Perhaps as an economist he is thinking of this as a purely technical problem without questioning the larger goals.

P.P.S. Just to be clear on this: I'm not saying that it's morally wrong to give financial advice to rich people (or even to "rich kids," which somehow sounds even worse). The question is why would two intellectually able economists, who can study anything they want, bother to write a book on the topic? (And, yes, you could ask why I am bothering to write a blog entry on the question. But I have a good answer to that, which is that it was bugging me, and a blog entry is the perfect vehicle for writing about something that bugs you.)

Michael Spagat notifies me that his article criticizing the 2006 study of
Burnham, Lafta, Doocy and Roberts has just been published. The Burnham et al. paper (also called, to my irritation (see the last item here), "the Lancet survey") used a cluster sample to estimate the number of deaths in Iraq in the three years following the 2003 invasion. In his newly-published paper, Spagat writes:

[The Spagat article] presents some evidence suggesting ethical violations to the survey's respondents including endangerment, privacy breaches and violations in obtaining informed consent. Breaches of minimal disclosure standards examined include non-disclosure of the survey's questionnaire, data-entry form, data matching anonymised interviewer identifications with households and sample design. The paper also presents some evidence relating to data fabrication and falsification, which falls into nine broad categories. This evidence suggests that this survey cannot be considered a reliable or valid contribution towards knowledge about the extent of mortality in Iraq since 2003.

There's also this killer "editor's note":

The authors of the Lancet II Study were given the opportunity to reply to this article. No reply has been forthcoming.

Ouch.

Now on to the background:

More than six-and-a-half years have elapsed since the US-led invasion of Iraq in late March 2003. The human losses suffered by the Iraqi people during this period have been staggering. It is clear that there have been many tens of thousands of violent deaths in Iraq since the invasion. . . . The Iraq Family Health Survey Study Group (2008a), a recent survey published in the New England Journal of Medicine, estimated 151,000 violent deaths of Iraqi civilians and combatants from the beginning of the invasion until the middle of 2006. There have also been large numbers of serious injuries, kidnappings, displacements and other affronts to human security.

Burnham et al. (2006a), a widely cited household cluster survey, estimated that Iraq had suffered approximately 601,000 violent deaths, namely four times as many as the IFHS estimate, during almost precisely the same period as covered by the IFHS study. The L2 data are also discrepant from data provided by a range of other reliable sources, most of which are broadly consistent with one another. Nonetheless, there remains a widespread belief in some public and professional circles that the L2 estimate may be closer to reality than the IFHS estimate.

But Spagat says no; he suggests "the possibility of data fabrication and falsification." Also some contradictory descriptions of sampling methods, which are interesting enough that I will copy them here (it's from pages 11-12 of Spagat's article):

Lets say you are repeatedly going to recieve unselected sets of well done RCTs on various say medical treatments.

One reasonable assumption with all of these treatments is that they are monotonic - either helpful or harmful for all. The treatment effect will (as always) vary for subgroups in the population - these will not be explicitly identified in the studies - but each study very likely will enroll different percentages of the variuos patient subgroups. Being all randomized studies these subgroups will be balanced in the treatment versus control arms - but each study will (as always) be estimating a different - but exchangeable - treatment effect (Exhangeable due to the ignorance about the subgroup memberships of the enrolled patients.)

That reasonable assumption - monotonicity - will be to some extent (as always) wrong, but given that it is a risk believed well worth taking - if the average effect in any population is positive (versus negative) the average effect in any other population will be positive (versus negative).

If we define a counter-factual population based on a mixture of the study's unknown mixtures of subgroups - by inverse variance weighting of the study's effect estimates by their standard errors - we would get an estimate of the average effect for that counter-factual population that is minimum variance (and the assumptions rule out much - if any bias in this).

Should we encourage (or discourage) such Mr P based estimates - just because they are for counter-factual rather than real populations.

K?

I'm not sure how the New York Times defines a blog versus an article, so perhaps this post should be called "Bayes in the blogs." Whatever. A recent NY Times article/blog post discusses a classic Bayes' Theorem application -- probability that the patient has cancer, given a "positive" mammogram -- and purports to give a solution that is easy for students to understand because it doesn't require Bayes' Theorem, which is of course complicated and confusing. You can see my comment (#17) here.

Prolefeed

From Anthony Burgess's review of "The Batsford Companion to Popular Literature," by Victor Neuberg:

Arthur J. Burks (1898-1974) was no gentleman. During the 1930s, when he would sometimes have nearly two million words in current publication, he aimed at producing 18,000 words a day.
Editors would call me up and ask me to do a novelette by the next afternoon, and I would, but it nearly killed me. . . . I once appeared on the covers of eleven magazines the same month, and then almost killed myself for years trying to make it twelve. I never did.

[Masanao: I think you know where I'm heading with that story.]

Ursula Bloom, born 1985 and still with us [this was written sometime between 1978 and 1985], is clearly no lady. Writing also under the pseudonyms of Lozania Prole (there's an honest name for you), Sheila Burnes and Mary Essex, she has produced 486 boooks, beginning with Tiger at the age of seven. . . .

Was Richard Horatio Edgar Wallace (1875-1932) a gentleman? . . . . In the 1920s and 1930s, Mr Neuburg tells us, one in four of all books was the work of Wallace. [How did Neuburg estimate this? I guess I'll have to track down his book and find out.] Everybody, especially the now unreadable Sir Hugh Walpole, looked down on this perpetually dressing-gowned king of the churners, who gave the public what it wanted.

Burgess continues:

What the public wanted, and still wants, is an unflowery style woven out of cliches, convincing dialogue, loads of action. Is the public wrong.

The added emphasis is my own.

I'll have more to say at some point about the popular literature of the past, but for now let me just note the commonplace that once-bestselling melodramas often seem unreadable to present-day audiences. I'm guessing that it has something to do with the cliches not working any more and the dialogue no longer being convincing. I'm sure there will even be a day when Eddie Coyle's words no longer sound natural. (Not that that book was ever extremely easy to read, nor was it a major bestseller.)

Don Green and Holger Kern write on one of my favorite topics, treatment interactions (see also here):

We [Green and Kern] present a methodology that largely automates the search for systematic treatment effect heterogeneity in large-scale experiments. We introduce a nonparametric estimator developed in statistical learning, Bayesian Additive Regression Trees (BART), to model treatment effects that vary as a function of covariates. BART has several advantages over commonly employed parametric modeling strategies, in particular its ability to automatically detect and model relevant treatment-covariate interactions in a flexible manner.

To increase the reliability and credibility of the resulting conditional treatment effect estimates, we suggest the use of a split sample design. The data are randomly divided into two equally-sized parts, with the first part used to explore treatment effect heterogeneity and the second part used to confirm the results. This approach permits a relatively unstructured data-driven exploration of treatment effect heterogeneity while avoiding charges of data dredging and mitigating multiple comparison problems. We illustrate the value of our approach by offering two empirical examples, a survey experiment on Americans support for social welfare spending and a voter mobilization field experiment. In both applications, BART provides robust insights into the nature of systematic treatment effect heterogeneity.

I don't have the time to give comments right now, but it looks both important and useful. And it's great to see quantitatively-minded political scientists thinking seriously about statistical inference.

Pretty pictures, too (except for ugly Table 1, but, hey, nobody's perfect).

Delia Baldassarri and Amir Goldberg write:

Americans' political beliefs present a long observed paradox. Whereas the mainstream political discourse is structured on a clearly defined polarity between conservative and liberal views, in practice, most people exhibit ideologically incoherent belief patterns. This paper challenges the notion that political beliefs are necessarily defined by a singular ideological continuum. It applies a new, network-based method for detecting heterogeneity in collective patterns of opinion, relational class analysis (RCA), to Americans' political attitudes as captured by the American National Election Studies. By refraining from making a-priori assumptions about how beliefs are interconnected, RCA looks for opinion structures, belief networks, that are not necessarily congruent with received wisdom. It finds that in the twenty years between 1984 and 2004 Americans' political attitudes were consistently structured by two alternative belief systems: one that is strongly aligned with the liberal-conservative divide, another that rejects the association between moral (e.g. abortion) and economic (e.g. redistribution) attitudes. An interaction between class and religious participation explains this variance. High earners with weak religious commitments, and vice versa, tend to exhibit the latter approach; those with seemingly conflicting moral and economic worldviews are also more likely to self-identify as Republicans. These findings are particularly relevant for recent debates on the roles of moral and economic beliefs in shaping political behavior, suggesting that different sociodemographic groups understand the political debate differently, and consequently emphasize different issues in deciding their party identification.

Here's what they do:

We [Baldassarri and Goldberg] identify three groups of respondents: Ideologues, who organize their political attitudes according to the prevalent liberal-conservative polarity; Alternatives, who reject the traditional prescriptive association between moral and economic attitudes, and are instead morally conservative and economically liberal, or vice versa (e.g. they tend to be pro-abortion but against economic redistribution); and Agnostics, who exhibit weak associations between political beliefs.

And here's (some of) what they find:

delia2.png

delia3.png

This is fascinating work, and I'm wondering if it can explain some of the different patterns of voting, income, and religiosity in different states (see especially Figure 6.12 of Red State, Blue State). I'll have to think more about this.

Erik Voeten describes a new plan in which rejected submissions to the American Economic Review are automatically submitted to lower-status journals, and he asks whether political science could do the same thing. This would be fine, I think, but in any case I suspect it's less of a big deal in political science. My impression is that the goal of "publishing in the top journal" is more of a big deal in economics than in political science. It's just not such a big deal to publish in the #1 journal. It's certainly not like in medicine, where if you publish in the New England Journal, you might hit page 1 of the newspaper. Or like physics or biology, where it's a big deal to publish in PRL or Cell or whatever.

Actually, economics seems to me more competitive than political science in general. I remember seeing, several years ago, a recommendation letter for an economist applying for a postdoctoral position. One of the letters of recommendation described the candidate as not being good enough for a faculty position at one of the top 8 programs but good enough for anything lower. I mean, c'mon, what kind of silly precision is this?

P.S. Statistics is even less hierarchical than political science. There is no agreed-upon top statistics journal. It depends if you're doing probability theory, theoretical statistics, or applied statistics, and even then, within each of these subfields there are multiple journals that could be considered as the best. In statistics, we also have the opportunity to publish in subject-matter journals (like the APSR!) or in computer science, engineering, and so forth. So much less pressure.

P.P.S. I'm not saying that statisticians and political scientists are better than political scientists, or that we're nicer people, just that the fields have different cultures. I'm actually surprised that the academic field of economics is so competitive since I'd assume that, as with statistics, many of the people competing for these jobs could get well-paid non-academic positions easily enough.

Anthony Goldbloom writes:

I'm writing because:

a) you may have some interest in our new project, Kaggle, a platform for data prediction competitions; and

b) to get your input.

First, the summary: Kaggle allows organizations to post their data and have it scrutinized by the world's best statisticians. It will offer a robust rating system, so it will be easy to identify those with a proven track record. Organizations can choose either to follow the experts, or to follow the consensus of the crowd -- which, (at least according to James Surowiecki) is likely to be more accurate than the vast majority of individual predictions. (It'll be interesting to see who triumphs - the crowd or the forecasters with a track record.) The power of a pool of predictions was demonstrated by the Netflix Prize, a $1m data-prediction competition, which was won by a team of teams that combined 700 models.

Now, for my questions:
1. Can you think of any interesting problems that would be ripe for a data-prediction competition?
2. Can you blog about Kaggle? We've had some interest from the general tech community and the data mining community, but it'd be great to get statisticians involved.

For interest, we're currently running a competition to forecast the voting in May's Eurovision Song Contest. As you may know, Eurovision pits performers from all over Europe against each other, producing voting outcomes which are widely believed to be influenced by politics and alliances rather than performance quality. Contestants in Kaggle's 'Forecast Eurovision Voting' competition will attempt to exploit these regularities to predict the voting matrix (who votes for who) for the 2010 Eurovision Song Contest. The winner of the Kaggle contest will collect a US$1,000 cash prize, which we hope to recoup by laying a bet based on the competition's consensus forecast.

My reply:

This looks like fun. From a statistical perspective, one thing that interests me is the extent to which different methods would be useful for different problems.

I remember several years ago talking with a professor of computer science who'd told me about some machine learning methods he was using, and I told him about the hierarchical interaction models that I was playing with (and which I'm still struggling to fit, so many years later . . . but that's another story, about the slowness of research in statistics compared to the rapid progress in computer science). Anyway, he told me about a problem he was working on--something to do with classification of proteins based on their molecular structure--and I told him about my problem--modeling voters based on geographic and demographic predictors. It turned out that his methods were useless for my problems and my methods were useless for his. Or, to be precise, his problem was so complex that I couldn't easily figure out how to apply my ideas, and he felt that my problem was so small and noisy that his methods wouldn't work for me.

At a deep level, I can't believe this could really be so. I'd think that a fully fleshed-out machine learning method would work on my little survey analysis problems, and that a fully-operational hierarchical modeling approach would work on his huge-data problems. Or, to put it another way, there should be some larger structure that includes these different approaches as special cases. But, in the meantime, it would be interesting to see the extent to which different methods work better on different sorts of examples.

P.S. Goldbloom adds that they're now offering a "spotting fee" for interesting competition ideas.

An eruption of data

Frank DiTraglia writes:

This might interest you and readers of the blog. I [Frank] can think of so many interesting ideas for this dataset; too bad they're all a distraction from my PhD!

As with those who manipulate symbols without reflective thought, that Andrew raised, I was recently thinking abouts students who avoid any distraction that might arise by their thinking about what the lecturer is talking about - so that they are sure to get the notes just right.

When I was a student I would sometimes make a deal where someone else would take the notes and I would just listen - then I would correct the notes they took for misconceptions later - there were almost always quite a few.

But not to be disparaging of students - they learned this somewhere/how and there must be advantages.

In fact - in someways math is a discouragement of thinking - replacing thinking with symbol manipulation thats assured to avoid wrong answers ... to the now zombified assumptions.

K?

Inappropriate parallelism

I've been teaching at elite colleges for over twenty years, and one thing that persistently frustrates me is the students' generally fluent ability to manipulate symbols without seeming to engage with the underlying context. Colorless green ideas sleep furiously, and all that. My theory is that in high school these students were trained to be able to write a five-paragraph essay on anything at all.

I was reminded of this when reading an article on the recent airline disruptions in Europe, where Washington Post columnist Anne Applebaum writes:

A friend with no previous interest in airline mechanics explained over the phone how two planes had already been affected. Another proffered a detailed description of the scientific process by which the ash enters the engine, melts from the heat, and turns back into stone--not what one wants inside one's airplane engines, really.

Others have become mystics. A British friend sees this as "judgment for the bad things we have done to the Earth." . . .

So far, so good. But then:

Though it is uncanny, I [Applebaum] do understand why some want science to explain this odd event and why others see the revenge of the volcano gods.

Huh? It seems a bit . . . anthropological of her to put science and "the volcano gods" in this sort of parallelism. It's no big deal, really, it just reminds me of a remark I once read that newspapers were better in the old days: Back when "journalists" were "reporters" and didn't have college educations, they just reported the facts and had neither the obligation to understand the world nor the inclination to smooth out reality to fit the contours of their well-rounded sentences. As a college teacher, I'm the last person to endorse such an idea, but it does have its appeal.

P.S. No, I doubt that Applebaum herself thinks of scientific and volcano-god explanations as equivalent. But that's my point: she wrote something that she (probably) doesn't really believe and, I suspect, she didn't really think clearly about, just because it fit the flow of her article. It was symbol-manipulation without full context.

?

Can somebody tell me why, when they're talking about those fancy financial products, people use the word "tranche" rather than "slice"? No big deal, I'm just curious.

BDA online lectures?

Eric Aspengren writes:

I've been attempting to teach myself statistics and I've recently purchased your book Bayesian Data Analysis. I have a small problem, unfortunately. I tend to need to have things explained to me by an actual, physical person (or a video). I've been able to use MIT's online video courses to help with my learning and was wondering if Columbia may have videos of your lectures available. If not, maybe there is a professor who teaches using your text that might have videos of their lectures available somewhere. I tend to be quite thick-headed when learning math.

I'm utterly fascinated by Bayesian methods. I work in politics and I feel there tends to be a lack of quantitative bases for decision making in this field. Conventional wisdom holds sway in many circles and I'm attempting to change that in my area.

My quick reply: I'd recommend starting with my book with Jennifer before moving on to Bayesian Data Analysis. Beyond this, no, I don't have any lectures online (except these). Actually, some of the online material on Bayesian statistics doesn't make me so happy (recall our blog discussion on the relevant Wikipedia articles). So I think you have to be careful what you listen to. Or, to put it another way, there's probably a lot of good stuff out there, but be careful not to take something seriously, just because someone says it in an authoritative manner.

P.S. Also, I write good books and give good one-hour lectures, but I'm not always so great over a one-semester course. I think you're better off taking a course out of my book but from a different lecturer who can present it from his or her own perspective.

Mark Girolami sends along this article, "Riemann Manifold Langevin and Hamiltonian Monte Carlo," by Ben Calderhead, Siu Chin, and himself:

This paper proposes Metropolis adjusted Langevin and Hamiltonian Monte Carlo sampling methods defined on the Riemann manifold to resolve the shortcomings of existing Monte Carlo algorithms when sampling from target densities that may be high dimensional and exhibit strong correlations. The methods provide fully automated adaptation mechanisms that circumvent the costly pilot runs required to tune proposal densities for Metropolis-Hastings or indeed Hamiltonian Monte Carlo and Metropolis Adjusted Langevin Algorithms. This allows for highly efficient sampling even in very high dimensions where different scalings may be required for the transient and stationary phases of the Markov chain. The proposed methodology exploits the Riemannian geometry of the parameter space of statistical models and thus automatically adapts to the local structure when simulating paths across this manifold providing highly efficient convergence and exploration of the target density. The performance of these Riemannian Manifold Monte Carlo methods is rigorously assessed by performing inference on logistic regression models, log-Gaussian Cox point processes, stochastic volatility models, and Bayesian estimation of dynamical systems described by nonlinear differential equations. Substantial improvements in the time normalised Effective Sample Size are reported when compared to alternative sampling approaches.

Cool! And they have Matlab code so you can go try it out yourself. If anybody out there knows more about this (I'm looking at you, Radford and Christian), please let us know. I care a lot about this right now because we're starting a big project on Bayesian computation for hierarchical regression models with deep interactions.

P.S. The tables are ugly but I forgive the authors because their graphs are so pretty; for example:

rmhmc.png

Some new articles

A few papers of mine have been recently accepted for publication. I plan to blog each of these individually but in the meantime here are some links:

Review of The Search for Certainty, by Krzysztof Burdzy. Bayesian Analysis. (Andrew Gelman)

Inference from simulations and monitoring convergence. In Handbook of Markov Chain Monte Carlo, ed. S. Brooks, A. Gelman, G. Jones, and X. L. Meng. CRC Press. (Andrew Gelman and Kenneth Shirley)

Public opinion on health care reform. The Forum. (Andrew Gelman, Daniel Lee, and Yair Ghitza)

A snapshot of the 2008 election. Statistics, Politics and Policy. (Andrew Gelman, Daniel Lee, and Yair Ghitza)

Bayesian combination of state polls and election forecasts. Political Analysis. (Kari Lock and Andrew Gelman)

Causality and statistical learning. American Journal of Sociology. (Andrew Gelman)

Can fractals be used to predict human history? Review of Bursts, by Albert-Laszlo Barabasi. Physics Today. (Andrew Gelman)

Segregation in social networks based on acquaintanceship and trust. American Journal of Sociology. (Thomas A. DiPrete, Andrew Gelman, Tyler McCormick, Julien Teitler, and Tian Zheng)

Here's the full list.

Tyler Cowen links to a news article about a biology professor who got removed from a class after failing most of her students on an exam.

The article quotes lots of people, but what I wonder is: what's on the exam? It would seem to be difficult to make a judgment here without actually seeing the exam in question.

The Two Cultures: still around

David Blackbourn writes in the London Review of Books about the German writer Hans Magnus Eisenberger:

But there are several preoccupations to which Enzensberger has returned. One is science and technology. Like left-wing intellectuals of an earlier period, but unlike most contemporary intellectuals of any political stamp, he follows scientific thinking and puts it to use in his work. There are already references to systems theory in his media writings of the 1960s, while essays from the 1980s onwards bear the traces of his reading in chaos theory.

For some inexplicable reason, catastrophe theory has been left off the list. Blackburn continues:

One of these takes the topological figure of the 'baker's transformation' (a one-to-one mapping of a square onto itself) discussed by mathematicians such as Stephen Smale and applies the model to historical time as the starting point for a series of reflections on the idea of progress, the invention of tradition and the importance of anachronism.

Pseuds corner indeed. I can hardly blame a European intellectual who was born in 1929 for getting excited about systems theory, chaos theory, and the rest. Poets and novelists of all sorts have taken inspiration by scientific theories, and the point is not whether John Updike truly understood modern computer science or whether Philip K. Dick had any idea what the minimax strategy was really about--these were just ideas, hooks for them to hang their stories. All is fodder for the creative imagination.

I have less tolerance, however, for someone to write in the London Review of Books to describe this sort of thing as an indication that Enzensberger "follows scientific thinking and puts it to use in his work." Perhaps "riffs on scientific ideas" would be a better way of putting it.

P.S. See Sebastian's comment below. Maybe I was being too quick to judge.

Saw a video link talk at a local hospital based research institute last Friday

Usual stuff about a randomized trail not being properly designed nor analyzed - as if we have not heard about that before

But this time is was tens of millions of dollars and a health concern that likely directly affects over 10% of the readers of this blog - the males over 40 or 50 and those that might care about them

Its was a very large PSA screening study and and

the design and analysis apparently failed to consider the _usual_ and expected lag in a screening effect here (perhaps worth counting the number of statisticians in the supplementary material given)

for an concrete example from colon cancer see here

And apparently a proper reanalysis was initially hampered by the well known - "we would like to give you the data but you know" .... but eventually a reanalysis was able to recover enough of the data from the from published documents

but even with the proper analysis - the public health issue - does PSA screening do more good than harm ( half of US currently males get PSA screening at some time? ) will likely remain largely uncertain or at least more uncertain than it needed to be

and it will happen again and again (seriously wasteful and harmful design and analysis)

and there will be a lot more needless deaths from either "screening being adopted" if it truly shouldn't have been or "screening was not more fully adopted, earlier" when it truly should have been (there can be very nasty downsides from ineffective screening programs, including increased mortality)



Nate asks, "Why is writing a 100,000-word book about 10 times harder than writing 100 1,000-word blog posts?" I don't know if this is true at all. Writing the book might be less fun than writing the blog posts, but is it really harder? If you really want to write the book, maybe the trick is to make blogging feel like work and book-writing feel like a diversion.

1, 2, 3, Many Tea Parties?

Thomas Ferguson and Jie Chen wrote an article tying the recent Massachusetts Senate election to national political trends:

Passage of the health care reform bill has convinced some analysts that the Massachusetts Senate election might be a fluke. In fact, polls taken after the legislation passed show Republicans widening their lead in fall congressional races. This paper takes a closer look at the Massachusetts earthquake. It reviews popular interpretations of the election, especially those highlighting the influence of the "Tea Party" movement, and examines the role political money played in the outcome. Its main contribution, though, is an analysis of voting patterns by towns. Using spatial regression techniques, it shows that unemployment and housing price declines contributed to the Republican swing, along with a proportionately heavier drop in voting turnout in poorer towns that usually provide many votes to Democratic candidates. All these factors are likely to remain important in the November congressional elections.

Lucia de Berk found not guilty

Maarten Buis writes:

A while ago you blogged on the case of Lucia de Berk:

Maybe you would like to know that her conviction has been overturned today.

Cheap talk and nukes

Sandeep Baliga presents an argument that is interesting but, I think, flawed. Baliga writes:

Obama's Nuclear Posture Review has been revealed. The main changes:
(1) We promise not to use nuclear weapons on nations that are in conflict with the U.S. even if they use biological and chemical weapons against us;

(2) Nuclear response is on the table against countries that are nuclear, in violation of the N.P.T., or are trying to acquire nuclear weapons.

This is an attempt to use a carrot and stick strategy to incentivize countries not to pursue nuclear weapons. But is it any different from the old strategy of "ambiguity" where all options are left on the table and nothing is clarified? Elementary game theory suggests the answer is "No".

Why, according to Baliga, does a nuclear policy have no effect? In short, because the government is not in fact constrained to follow any specified policy in the future:

Contracts we write for private exchange are enforced by the legal system. . . . But there is no world government to enforce the Nuclear Posture Review so it is Cheap Talk. . . .

What if our targets do not know our preferences? Do they learn anything about our preferences by the posture we have adopted? Perhaps they learn we are "nice guys"? But even bad guys have an incentive to pretend they are nice guys before they get you. Hitler hid his ambitions behind the facade of friendliness while he advanced his agenda.

I commented:

Is this really true? My impression was that Hitler was pretty aggressive, in his rhetoric as well as his actions.

One could argue that Hitler was so evidently a bad guy, even at the time, that he had nothing to gain by talking nice. But, the point remains, that if "cheap talk" is really so valueless as all that, there'd be no reason for anyone to do it at all. I think it might be more useful to model actors (such as "the U.S." or "Germany") as complex entities, and to consider that stated policies and goals might, at the very least, affect he balance of power in future intra-organizational struggles.

To which Baliga replied:

My point is not that cheap talk is useless in ALL games, only in some games. . . . But there is one key case where cheap talk is useless even in games of incomplete information: when a player i's preferences over player j's actions do not depend on player i's preferences. In the nuclear story, this arises if the player i prefers that player j not acquire nuclear weapons, whether player i is itself rapacious, conciliatory or something in between. Then, player i will always send the message that minimizes the probability that player j arms and cheap talk cannot be informative.

This clarification helps, but I still think the analysis breaks down in this example.

One way to see this is to ask, if this "cheap talk" is useless, why it's done at all! Or, conversely, why it wasn't done earlier. Baliga's analysis seems to me to rely on there being some "suckers" somewhere who don't realize what's going on.

Perhaps, for example, the leaders of Iran, Russia, etc., aren't fooled by the cheap talk.--after all, they run countries and have incentives to understand the relevant signaling--but maybe it could sway American voters, who don't have the time and inclination to gain a deep understanding of power politics. But . . . if it could fool the voters, it could change U.S. policy, and in that sense the stated policy does mean something. Beyond this, there are default effects and status quo effects and costs to violating or altering a stated policy. So I don't think such public statements are necessarily meaningless, especially considering the many players involved in policymaking in any country.

On the other hand, I know next to nothing about international relations, so I could well be missing something important. I don't see Baliga's conclusions as following from basic game theory but maybe there's something about this particular setting that changes things.

Forecasting market future

McKinsey had a great chart today:

mckinsey2.png

To explain the chart - the green are the predictions for the next year at various times in the past. The blue is the truth that eventually comes along. So, the predictions do tend towards the truth as time goes on.

They're consistently optimistic - but does the data justify a more sophisticated model? Or is the design of markets inherently biased towards rewarding optimists more than realists?

Random matrices in the news

Mark Buchanan wrote a cover article for the New Scientist on random matrices, a heretofore obscure area of probability theory that his headline writer characterizes as "the deep law that shapes our reality."

It's interesting stuff, and he gets into some statistical applications at the end, so I'll give you my take on it.

But first, some background.

About two hundred years ago, the mathematician/physicist Laplace discovered what is now called the central limit theorem, which is that, under certain conditions, the average of a large number of small random variables has an approximate normal (bell-shaped) distribution. A bit over 100 years ago, social scientists such as Galton applied this theorem to all sorts of biological and social phenomena. The central limit theorem, in its generality, is also important in the information that it indirectly conveys when it fails.

For example, the distribution of the heights of adult men or women is nicely bell-shaped, but the distribution of the heights of all adults has a different, more spread-out distribution. This is because your height is the sum of many small factors and one large factor--your sex. The conditions of the theorem are that no single factor (or small number of factors) should be important on its own. For another example, it has long been observed that incomes do not follow a bell-shaped curve, even on the logarithmic scale. Nor do sizes of cities and many other social phenomena. These "power-law curves," which don't fit the central limit theorem, have motivated social scientists such as Herbert Simon to come up with processes more complicated than simple averaging (for example, models in which the rich get richer).

The central limit theorem is an example of an attractor--a mathematical model that appears as a limit as sample size gets large. The key feature of an attractor is that it destroys information. Think of it as being like a funnel: all sorts of things can come in, but a single thing--the bell-shaped curve--comes out. (Or, for other models, such as that used to describe the distribution of incomes, the attractor might be a power-law distribution.) The beauty of an attractor is that, if you believe the model, it can be used to explain an observed pattern without needing to know the details of its components. Thus, for example, we can see that the heights of men or of women have bell-shaped distributions, without knowing the details of the many small genetic and environmental influences on height.

Now to random matrices.

After pointing out that getting a true picture of how log prior and log likelihood add to get the log posterior - was equivalent to getting a fail safe diagnostic for MCMC convergence

I started to think that was bit hard - to just get a display to show stats was easy ...

But then why not just subtract?

This blog has previously included discussions of prediction markets.

A new article in Slate claims that Intrade and perhaps some other markets have an "anti-Obama" bias. According to the article, a wide swath of people (including Intrade participants, economists, Fox and CNBC financial pundits, and the Wall Street Journal) systematically underpredict the Administration's likelihood of advancing its agenda, and are systematically gloomy about the outcome of the Administration's successes. The article points out that in the year since a Stanford economist wrote an op-ed entitled "Obama's Radicalism is Killing the Dow," the Dow is up 72%.

A question about scaling

Brian Kundsen writes:

I'm no expert on British politics, so maybe I'm actually the confused one. Anyway . . .

John Lanchester writes in the London Review of Books:

Labour have an enormous statistical advantage going into the election. The simple way of putting this is to say that votes in the country are worth less than votes in the city. That's because the Boundary Commission has struggled to keep up with the historic drift of Britons out of cities into the country . . . Country constituencies are bigger, in population as well as geographical terms, than urban ones . . . Because Labour's support skews urban and the Conservatives' skews rural, this translates into a big advantage for Labour. How big? Well, this non-partisan article from the House of Commons magazine, dating from 2006 when the election was a long way off, reckoned that the Tories needed to win the election by a margin of 10 per cent in order to have any majority at all.

If you follow the link, though, it appears that (a) the boundaries were redrawn in 2005 or 2006, so the boundaries are only 4 or 5 years out of date, and (b) the "10 percent majority" thing is not coming from any imbalances in district sizes:

The Conservatives will still need a swing of about 10 per cent to win power outright . . . A swing of just over one per cent will now cost Labour its overall majority, compared to 1.8 per cent with boundaries unchanged. The changes reduce the swing needed by David Cameron to secure an overall majority from 11 per cent under the old boundaries to nine or 10 per cent with the constituencies which will be used in the next general election . . .

Got that? The effect of redrawing the district lines is estimated to be something like 0.7% of the vote, not 10%. (I'm getting 0.7% by subtracting "just over one per cent" from "1.8 per cent.") If the boundary change in 2005 or 2006 comes to the equivalent of 0.7% of the vote, then I'd expect any population shifts since then to account for less than 0.7% in partisan advantage.

Less than 0.7%. Not 10%.

The 10% is coming from the multiparty system, which has at times benefited the Labour party and at times benefited the Conservatives and doesn't seem to have benefited the Liberal Democrats at all (yet).

Bill James has a dictum that the alternative to doing good statistics is not no statistics but bad statistics. People who choose not to take numbers seriously--that is, as numbers--do not simply ignore numbers; rather, they treat them as words. In baseball, that meant, for example, going gaga over free-swingers who hit .300 in a hitter's park and underrating low-average power hitters who draw walks. (Or, at least, that's how things used to be before James and others popularized on-base percentage, slugging average, and all the rest.)

In the political context, Lanchester is pulling out numbers without regard to their magnitude--1%, 10%, what's the difference? He wants to understand the political system--I respect that--but he doesn't know where to start, so he picks out numbers where he can. I don't think he's trying to mislead, I just think this is what happens when you take numbers as words.

P.S. I happened to notice this one because I followed the link from Helen DeWitt's blog. More generally, though, I notice a fair amount of innumeracy in the London Review of Books and (back when I used to subscribe to it) the New York Review of Books. For example, here, here, here, and here (from Gary Wills, Frederick Crews, David Runciman, and Samantha Power).

These people are all busy writers, and I can hardly expect them to find the time to think quantitatively, so I think a statistical copy editor would be a good idea. I think they could hire Ubs to do this for a reasonable fee, for example.

P.P.S. Yes, I know, hitting .300 in the majors is no mean feat. There's no way I could hit .003 in any park. You could put me at bat every day for the rest of my life against anybody--major league, minor league, whatever--and I'd never even come close to getting a hit. The point above, though, is about comparisons of world-class athletes, which we can attempt even if we are leagues and leagues below them in ability.

Interesting discussion (by Dave Armstrong, Will Wilkinson, and somebody called Sebastian) here.

OK, OK . . .

Donovan Crew writes:

I was wondering if you saw this, and what your take would be on it.

My reply: Yes, everybody's been emailing me this! I have a post prepared on it but am waiting until current backlog dissipates.

Elizabella

Sethi on Hirschman

Exit, Voice, and Loyalty, by Albert Hirschman, has a title that's too grabby for its own good.

You hear about the book and think: Yeah, that makes sense. In a difficult situation I can get out ("exit"), speak up ("voice"), or try to strengthen the organization ("loyalty"). We experience this choice in so many different areas of life, from kindergarten playgrounds to marriages to business dealings.

But Hirschman's book does much more than lay out these choices. Read Rajiv Sethi's fascinating appreciation of Exit, Voice, and Loyalty to see what's there.

P.S. I'm sensitive to this issue of too-grabby-a-title partly because this happened, at a much lower level, with our book Red State, Blue State, Rich State, Poor State. Our findings on individual income, state income, and voting were already well known, and a lot of people just assumed that this was all that was in the book, not noticing all the material on religion, economic and social issues, and political polarization. I think we would've been better off with a vaguer title such as "Political Polarization: How Americans are Divided and How They're Not."

I increased the range of the plot from Statistics is easy! part 2 and added the 2.5% and 97.5% percentiles from a WinBugs run on the same problem ... using bugs() of course

And then started to worry about that nasty bump on the right of the 97.5% percentiles

plot3.png

Controversy over social contagion

Dan Engber points me to an excellent pair of articles by Dave Johns, reporting on the research that's appeared in the last few years from Nicholas Christakis and James Fowler on social contagion--the finding that being fat is contagions, and so forth.

More precisely, Christakis and Fowler reanalyzed data from the Framingham heart study--a large longitudinal study that included medical records on thousands of people and, crucially, some information on friendships among the participants--and found that, when a person gained weight, his or her friends were likely to gain weight also. Apparently they have found similar patterns for sleep problems, drug use, depression, and divorce. And others have used the same sort of analysis to find contagion in acne, headaches, and height. Huh? No, I'm not kidding, but these last three were used in an attempt to debunk the Christakis and Fowler findings: if their method finds contagion in height, then maybe this isn't contagion at all, but just some sort of correlation. Maybe fat people just happen to know other fat people. Christakis and Fowler did address this objection in their research articles, but the current controversy is over whether their statistical adjustment did everything they said it did.

So this moves from a gee-whiz science-is-cool study to a more interesting is-it-right-or-is-it-b.s. debate.

Matthew Yglesias writes that, setting aside the merits of the issues, Republicans should not worry about potential electoral losses arising from a move to the right:

It's not at all clear to [Yglesias] that the heart of [David Frum's] criticism--that Republicans need to moderate in order to become electorally viable--is really true. The empirical evidence to me [Yglesias] suggests that our default view about the relationship between ideology and electability ought to be one of nihilism--any challenger can win provided the economy is doing poorly, and any incumbent can get re-elected provided things are going allright.

Yes, but not completely.

There is definitely some evidence that moderate candidates do better.

A centipede many times over

I recently came across this presentation by Charlie Geyer on diagnosing problems with Markov chain simulations. Charlie gives some good advice, and some advice that's not so relevant to my own experience, and misses some other ideas that I and others have come across.

To a repeat a frequent theme of mine (see, for example, meta-principle #3 in this discussion), I suspect that many of the differences between Charlie's approaches and those more familiar with me stem from the different sorts of problems we work on. Charlie works on big, tangled, discrete models in genetics--the sorts of problems with multiple modes and where, not only can computations be slow to converge, then can easily get stuck in completely unreasonable places. To solve these problems, you have to write complicated computer programs and solve problems specific to your model and your application. In contrast, I work on smaller, continuous problems in social science where, if you find yourself really far from where you want to be in parameter space, you'll probably realize it. I develop relatively simple but general algorithms that are intended to work for a wide range of examples and applications.

Neither my approach nor Charlie's is "better"; we just do different things.

Here are three examples.

First, on page 3 of his slides, Charlie writes:

Sex ratios in the GSS

Jim Manzi writes:

I just read your post on the recent paper on the causal effect of daughters and sons on political orientation. It struck me as really weird that the sex ratio among respondents to both questions was almost exactly 1.1. Isn't it very stable around 1.05 for the US population? If so, couldn't there be a very biased sample of people who are willing to respond to these questions that could account for the difference? (Though it does seem weird - though not inconceivable - that such a bias would be almost identical for both questions.)

My reply: I dunno, perhaps some differences in the rate at which people give up kids for adoption? Or maybe it's some sort of nonresponse issue. It would make sense to see if something similar is happening in other surveys that ask about the sex of respondents' children, and it would also make sense for the researchers mentioned in the cited post to do some sensitivity analysis regarding these selection issues.

Ted Dunning points me to this article by J. Andres Christen and Colin Fox:

We [Christen and Fox] develop a new general purpose MCMC sampler for arbitrary continuous distributions that requires no tuning. . . .The t-walk maintains two independent points in the sample space, and all moves are based on proposals that are then accepted with a standard Metropolis-Hastings acceptance probability on the product space. . . . In a series of test problems across dimensions we find that the t-walk is only a small factor less efficient than optimally tuned algorithms, but significantly outperforms general random-walk M-H samplers that are not tuned for specific problems. . . . Several examples are presented showing good mixing and convergence characteristics, varying in dimensions from 1 to 200 and with radically different scale and correlation structure, using exactly the same sampler. The t-walk is available for R, Python, MatLab and C++ at http://www.cimat.mx/~jac/twalk/.

This looks pretty cool to me! I asked Christian Robert what he thought, and he referred me to this blog entry where he presents a bit of a skeptical, wait-and-see attitude:

The proposal of the authors is certainly interesting and widely applicable but to cover arbitrary distributions in arbitrary dimensions with no tuning and great performances sounds too much like marketing . . . there is no particular result in the paper showing an improvement in convergence time over more traditional samplers. . . . Since the authors developed a complete set of computer packages, including one in R, I figure people will start to test the method to check for possible improvement over the existing solutions. If the t-walk is indeed superior sui generis, we should hear more about it in the near future...

I have a lot of big models to fit, so I expect we'll be trying out many of these different methods during the next year or so.

P.S. Update here.

Political leanings of sports fans

A few months ago, Yu-Sung and I summarized some survey results from the 1993-1996 General Social Survey. 56% of respondents said they attended an amateur or professional sports event" during the past twelve months, and it turned out that they were quite a bit more Republican than other Americans but not much different in their liberal-conservative ideology:

sport.png

Then, the other day, someone pointed me to this analysis by Reid Wilson of a survey of TV sports watchers. (Click the image below to see it in full size.)

Sports-Stats_900-thumb-300x210.gif

The graph is very well done. In particular, the red and blue coloring (indicating points to the left or right of the zero line) and the light/dark (indicating points above or below the center line on the vertical axis) are good ideas, I think, despite that they convey no additional information, in that they draw attention to key aspects of the data.

An embarrassing question

Someone asks:

Do you have, and would you be willing to share, the code used to generate figure 1 of your paper "Lets practice what we preach: Turning Tables Into Graphs"?

My response: I'd be happy to share the code, if I had it! The program was written by my collaborator, and I doubt he's saved it. We do have coefplot() in the arm package, though. Beyond this, I've heard people say good things about ggplot2.

In any case, I think the hardest part is figuring out what graph you want to make, and the second hardest part is making all the little choices to make the display work well. Once you have the basic idea, you can sketch it out and then just draw in all the lines and points one at a time pretty easily.

Aleks (I think) sent me this link to a paper by Jianiang Shi, Wotao Yin, Stanley Osher, and Paul Sajda. They report impressive computational improvements based on "a novel hybrid algorithm based on combining two types of optimization iterations: one being very fast and memory friendly while the other being slower but more accurate." This looks awesome. And they even test their method on the UCI database, which is what Aleks used in the cross-validation studies for bayesglm.

Here are my questions:

1. Could their algorithm work with other regularizations? L1 corresponds to a folded-exponential (Laplace) family of prior distributions? I prefer the t model. We implemented the t in bayesglm using an adaptation of the standard weighted least squares algorithm. That's fine for our little examples, but for big problems, where speed is a concern, maybe this new method is much better I imagine it could be adapted to our family of prior distributions, but I don't know enough about the method to be sure.

2. What about hierarchical models? That's the next thing to do. lmer/glmer are running pretty slow for us.

P.S. The funny thing is that Shi and Sajda are at Columbia (at the biomedical engineering department). But, to the best of my knowledge, we've never met. I don't even know where their office is located. According to their addresses in the article, they have a 10025 zipcode. But the engineering departments I've been to are all in 10027.

Tyler Cowen reports the following claim from sociologists Dalton Conley and Emily Rauscher:

Using nationally-representative data from the [1994] General Social Survey, we [Conley and Rauscher] find that female offspring induce more conservative political identification. We hypothesize that this results from the change in reproductive fitness strategy that daughters may evince.

This surprised me, because less than a year ago, we reported here on a study by economists Andrew Oswald and Nattavudh Powdthavee with the exact opposite finding:

We [Oswald and Powdthavee] document evidence that having daughters leads people to be more sympathetic to left-wing parties. Giving birth to sons, by contrast, seems to make people more likely to vote for a right-wing party. Our data, which are primarily from Great Britain, are longitudinal. We also report corroborative results for a German panel.

Understanding the results (possibly) in terms of "family values"

This is a fun problem: we have two different studies, both by reputable researchers, with opposite results! I took a look at both papers and can't immediately see a resolution, but I will offer some speculations, followed by some scattered comments.

We already know that the Republicans are the party of families much more than Democrats are. Just for example, John McCain did 20 percentage points better among married than unmarried voters.

The way I'd interpret Conley and Rauscher's finding is that having a child is likely to make you much more focused on family issues, and that having a daughter (as compared to a son) might make this effect even larger. To the extent that the Republicans were seen (as of 1994) as the family-values party, this could swing some votes (or, at least, some party identification). Commenter DN below also suggests a connection to views about crime. It should be possible to look into this by studying GSS responses to some different survey questions.

At least in the U.S. context, it makes much more sense to to me to see this sons/daughters thing as being a spinoff from the huge differences between married and unmarried voters than in terms of evolutionary strategies (more on this below).

From Anthony Burgess's review of "Best Sellers: Popular Fiction of the 1970s," by John Sutherland, I learn that The Godfather sold 300,000 hardcover and 13 million paperbacks during that decade, and Fear of Flying, the book at the bottom of the New York Times's bestselling list (#10? #15?), sold 100,000 hardcover and nearly 6 million in softback. Comparing to the list from 1965, we see that absolute sales increased rapidly. And the numbers have shot up still more if it's true what they're saying about James Patterson.

What I want to know is, what happened to all those copies of "Alive!" Back in the 1970s, you used to see copies of that book everywhere. Did everyone who owned that book throw it out? Or was it printed on some sort of special disintegrating paper guaranteed to fall apart after twenty years? I guess I should also ask what happened to those two hundred million Perry Mason books. Are they all in Grandma's attic somewhere?

What should they teach in school?

Bill Mill links to this blog by Peter Gray arguing that we should stop teaching arithmetic in elementary schools. He cites a research study from 75 years ago!

L. P. Benezet (1935/1936). The teaching of Arithmetic: The Story of an Experiment. Originally published in Journal of the National Education Association in three parts. Vol. 24, #8, pp 241-244; Vol. 24, #9, p 301-303; & Vol. 25, #1, pp 7-8.

I imagine there's been some research done on this since then?? Not my area of expertise, but I'm curious.

P.S. You gotta read this anonymous comment that appeared on Gray's blog. I have no idea if the story is true, but based my recollection of teachers from elementary school, I can definitely believe it!

I've made this point several times recently but just want to get it down one more time.

Kaiser writes that "the gulf between infographics and statistical graphics remains wide," and he provides some examples of eye-catching charts that don't do a very good job at all of presenting information.

I think the following two ideas are helpful:

1. Tufte had it write when he wrote of the visual display of quantitative information. Not "data" but, more generally, "quantitative information." Which can include all sorts of derived quantities, ranging from the Consumer Price Index to logistic regression coefficients (generally best to divide these by 4, of course). Much confusion (not from Kaiser, but from others) has arisen because people think of statistical visualization to be about showing the raw data. No, it's more general than that.

2. Newspaper and magazine articles are often illustrated by photographs and cartoons which are pretty or shocking or attention-grabbing or interesting and in some way complement the substance of the article. We don't generally criticize such illustrations as being noninformative; they're not really expected to convey information in the first place.

I think of infographics (for example this sort of thing featured by Nathan Yau) as a cross between 1 and 2 above. Another example is Harper's Index (which I always thought was a cool idea, and it even featured my own research once!). It conveys a bit of information but not systematically--it's more a way to stimulate some thought than to draw any conclusions. It plays the role of a good photographic illustration.

P.S But some displays are cool and informative! I am sad that a cool-but-ultimately-limiting gimmick such as Wordle has become so popular, but I'm happy that there's also room in the world for something like the Baby Name Wizard, which is cool and also informative and leads to new and interesting ways of thinking about the world.

From Anthony Burgess's review of John Updike's The Coup:

This emboldens me [Burgess] to set some of Mr Updike's prose as verse, making such small emendations as are necessary to regularize the rhythm:
The piste diminished to a winding track,

Treacherously pitted, strewn with flinty scrabble
That challenged well the mettle of our Michelin
Steel-belted radials. Distances grew bluish;
As we rode higher, clots of vegetation,
Thorny and leafless, troubled with grasping roots
The rocks. In the declivities that broke
Our grinding, twisting ascent, there were signs
Of pasturage: clay trampled to a hardened
Slurry by hooves, and also excrement
Distinguishable still from mineral matter,
Some toppled skeletons of beehive huts,
Consumed, their thatches, as a desperate fodder.
Aristada, which thrives on overgrazed
Lands, tinged with green this edge of desolation.

I see Burgess's point. More than this, I'm reminded of the very low standards of contemporary poetry. If you want to write a novel or even a short story, it better be interesting in some way. For a poem, though, it just has to be . . . not too embarrassingly bad. The above passage, as poetry, would fit just fine in the New Yorker or elsewhere--it follows all the rules (imaginary gardens with real toads and all the other life-affirming b.s.). And, in fact, if I saw it there and troubled to read it in that format, I'd probably think of it as pretty good, in the sense that I could actually figure out what it is talking about. Personally, I trace this back to what we were taught by our high school English teachers about poetry being intense, poetry being a puzzle, and or course good poetry being something you're supposed to like. Now, I fully admit that T. S. Eliot has been admired by many people whose literary skills, tastes, and judgments I respect much more than my own--still, I think of him not so much as a great poet but as somebody who was well connected and got off a few good lines. I'm down on the whole poetry-as-puzzle thing. It worked OK for Michael Stipe, but he had that music thing going on.

OK, now on to Topic #1

The above is all set-up to my main point. My real goal here is not to bash poetry but to reflect upon a related item from Burgess's review, where he writes:

[Updike] is committed also to a kind of poetic unit, a verse paragraph that, in certain contexts of action or even speech, seems excessively long. And there is a basic melody which seems to require otiose adjectives:
. . . a poster of Elvis Presley in full sequinned regalia, Marilyn Monroe from a bed of polar bear skins making upwards at the lens the crimson O of a kiss whose mock emotion led her to close her greasy eyelids, and a page torn from that magazine whose hearty name of Life did not save it from dying. . . .

That hearty [continues Burgess] is surely unecessary . . . yet without the adjective the prosody falters.

I have a few comments:

1. I think Burgess is right. "Hearty" does not really make sense there, but it wouldn't work to simply remove it. For one thing, "whose name of Life" would sound too much like "the game of life." The word "hearty" nudges the reader and keeps the sense of the sentence moving along.

2. Updike's rhythm really works, allowing the reader (that is, me) to follow a complex sentence straight through the first time. And I know, from my own experience, that it's not easy to get that sort of flow.

Just for example, why did Updike put "upwards at the lens" before "the crimson O"? In spoken English it would be natural to set out the object of the verb right away (that is, to say, "Making the crimson O of a kiss" and go from there). But, no, that wouldn't work: if you put the "upwards at the lens" phrase after the kiss, it won't be clear that it should be modifying Marilyn, not her kiss.

I'm aware of this particular rearrangement trick because I do it all the time in my own writing. My point is that it takes a lot of work to get the sentences to just flow on the printed page, where you don't have timing, intonation, facial expressions, and gestures to help clarify your intentions.

3. There are other ways to keep the logical flow, for example the sort of frank discursiveness most impressively (to me) done by Nicholson Baker in The Mezzanine.

4. My main reaction, though, is the thought that Burgess's argument applies to me as well! At a much lower level than Updike, sure, but still. I put so much effort into the flow of my sentences (even, I'm embarrassed to admit, when blogging), and in writing about technical matters (as I usually do), I have the further constraint of not wanting to get anything wrong. In Bayesian Data Analysis in particular, I went carefully over everything to make sure I was not saying anything sloppy. I'd noticed that a lot of the statistics literature had such sloppiness, and I wanted to be careful to label rigorous results, conjectures, and opinions as such. Along with all of this, I try to avoid cliches, especially those sloppy expressions such as "the lion's share" (one of my pet peeves, for some reason--and, no, I don't really think of "pet peeve" as a cliche even though, yes, I know that it is) which are vaguely--but only vaguely--quantitative.

Anyway, in making sure my sentences are readable, I sometimes lapse into too rhythmic a style. I hadn't really thought of this except in special cases, but reading this Burgess review of Updike, it all suddenly makes sense. There can be a tradeoff between rhythm and meaning. And not just in a simple way that you can choose between words that are true and words that are beautiful. Rather, the same rhythm that I rely on to make my words clear on the page (or the screen) has the effect of requiring me to add unnecessary words. And there's no simple solution, because if I just remove those extra words, all of a sudden it can be hard for people to follow what I'm saying.

As they say in the stagecoach business, remove the padding from the seats and you get a bumpy ride.

P.S. This all reminds me . . . I'm a big Updike fan (despite being unimpressed by his book titles); I love Rabbit, Run and like its sequels, many of the short stories are just amazing . . . a few years ago I picked up a copy of Roger's Version, which I recalled has having received good reviews. I read a few pages but just couldn't continue--nothing about the writing seemed remotely plausible. It didn't seem anything like how a real person would talk (nor was it interesting enough to realize for other reasons). Rabbit, Run, though, that was great. I like how it was written and also its ideas. I completely disagree with the notion that Updike was merely a painter of pretty word-pictures with nothing to say.

P.P.S. Speaking of Updike, here's my favorite poetry-related item from recent years (from the New Yorker in 1994):

baker.png

Statistics is, I hope, on balance a force for good, helping us understand the world better. But I think a lot of damage has been created by statistical models and methods that inappropriately combine data that come from different places.

As a colleague of mine at Berkeley commented to me--I paraphrase, as this conversation occurred many years ago--"This meta-analysis stuff is fine, in theory, but really people are just doing it because they want to run the Gibbs sampler." Another one of my Berkeley colleagues wisely pointed out that the fifty states are not exchangeable, nor are they a sample from a larger population. So it is clearly inappropriate to apply a probability model for parameters to vary by state. From a statistical point of view, the only valid approaches are either to estimate a single model for all the states together or to estimate parameters for the states using unbiased inference (or some approximation that is asymptotically unbiased and efficient).

Unfortunately, recently I've been seeing more and more of this random effects modeling, or shrinkage, or whatever you want to call it, and as a statistician, I think it's time to lay down the law. I'm especially annoyed to see this sort of thing in political science and public opinion research. Pollsters work hard to get probability samples that can be summarized using rigorous unbiased inference, and it does nobody any good to pollute this with speculative, model-based inference. I'll leave the model building to the probabilists; when it comes to statistics, I prefer a bit more rigor. The true job of a statistician is not to say what might be true or what he wishes were true, but rather to express what the data have to say.

Here's a recent example of a hierarchical model that got some press. (No, it didn't make it through the rigor of a peer-reviewed journal; instead it made its way to the New York Times by way of a website that's run by a baseball statistician. A sad case of the decline of intellectual standards in America, but that's another story.) I'll repeat the graph because it's so seductive yet so misleading:

A couple weeks ago I blogged on John Gottman, a psychologist whose headline-grabbing research on marriages (he got himself featured in Blink with a claim that he could predict with 83 percent accuracy whether a couple would be divorced--after meeting with them for 15 minutes!) was recently debunked in a book by Laurie Abraham. Discussion on the blog revealed that Laurie Abraham had tried to contact Gottman but he had not replied to the request for an interview.

After this, Seth wrote to me:

Well-connected journalists

I learned recently that the wife of the foreign minister of Poland is a columnist for the Washington Post. Or, should I say, the husband of a Washington Post columnist is the foreign minister of Poland. (I also learned that said columnist is not a fan of Hilary Clinton--perhaps not the most diplomatic thing to say right now, foreign policy-wise?)

What I'm wondering, though, is how common this sort of thing is, for a major journalist to have open political connections. Here I'm not thinking of retired politicians who write for the press (an op-ed by Newt Gingrich, a column by Eliot Spitzer) or pundits and politicos such as William F. Buckley and Pat Buchanan who appear on TV, but someone with this kind of political role who has what seems to be more of a straight journalist job.

I'm certainly not trying to imply that there's anything wrong here--I'd be the last to claim that relation by marriage to the foreign minister of a midsize country is any disqualification from commenting about politics--I'm just wondering how often this occurs, and whether it's more common now than in earlier decades. I recall seeing occasional conflict-of-interest items of this sort before, but always on a case-by-case basis.

P.S. When I posted this at the sister blog, several commenters named the husband-wife couple of NBC's Andrea Mitchell and Federal Reserve chair Alan Greenspan. There must be a lot more though, no? What I was imagining was not a list of individually prominent cases, but some kind of count, even if incomplete.

Babies as vote-getters??

From Freakonomics blog:

In the U.K., however, Conservative leader David Cameron -- the likely winner, per the prediction markets, in the yet-to-be called election -- has just unleashed a doozy: his wife Samantha is expecting the couple's fourth child. Their first-born, Ivan, recently died at age 6 from a rare neurological condition. It is hard to imagine that a pregnant wife won't help Cameron a bit more.

"It's hard to imagine"? Not that hard. I can well imagine that voters in Britain (as in the U.S.) will choose their candidates based on their records, their party positions, and so forth. Also I've heard that the economy can be a factor. I'm not saying Freakonomics is wrong, I've just never heard of the pregnant-wife-wins-votes theory before.

The hidden side of everything, indeed.

That said, I'm not current on the political science literature, so I posted this on the sister blog to see if anyone knew of any research on the topic. No bites yet, but I'm still willing and ready to be educated on the topic.

Well can we at least make it look easy?

For the model as given here, there are two parameters Pc and Pt - but the focus of interest will be on some parameter representing a treatment effect
- Andrew chose Pt - Pc.

But sticking for a while with Pt and Pc - the prior is a surface over Pt and Pc as is the data model (likelihood)

In particular, the prior is a flat surface (independent uniforms)
and the likelihood is Pt^1 (1 - Pt)^29 * Pc^3 (1 - Pc)^7 (the * is from independence)

(If I reversed the treatment and control groups - I should be blinded to that anyways)

What made the clockmaker tick?

In a comment on a note on Anthony Burgess, Steve writes:

Burgess famously was told once by his doctor that he had a year to live. So, to provide for his family after his death, he wrote five novels in one year, all published. He turned out to be perfectly healthy and went on to publish countless words.

I just read a biography--The Real Life of Anthony Burgess, by Andrew Biswell--which suggests that it didn't quite happen like that.

The mystery of mysteries

Commenter Maxine on Jenny's blog writes:

I find that the endings are the worst things about crime fiction. Harlan Coben for example writes great posers, but then the end.....hmmm. Sophie Hannah seems to be doing something similar (just read her most recent, A Room Swept White). . . .

One quite frustrating thing about reviewing crime fiction is that one cannot criticise these silly endings as you would then give away the whole point.

This all sounds reasonable, but . . . why is it that revealing the ending of a crime novel would "give away the whole point" more than revealing the end of a non-crime novel?

Is there something inherent in crime that makes it more suspenseful than other endeavors, or just some sort of literary convention? I don't see it. There can be just as much suspense in a story of love, or illness, or animals, or war, or all sorts of other human endeavors, no? So maybe it's just a tradition, that crime stories are expected to be wrapped around a "whodunit?" or "how will he get caught?" structure, whereas other sorts of stories are expected to work off of a known outcome.

I have this image in my mind of the central thread of a crime story tied down at one end (usually, although not necessarily, the chronological beginning), and the central thread of a traditional non-crime story being tied down at two ends, so you know where it will start and where it will end.

I was thinking about this general topic the other day while reading snippets of a sci-fi novel that I've been carrying around in my jacket pocket. The novel is from 1950--reprinted in the 70s, I think, but still old enough that it's satisfyingly short and easy to carry. It's fun to see all the conventions being followed, one of which is that it will have a happy ending, in the sense that the hero will be alive and victorious at the story's end. Well, maybe not--I still have another 10 or so pages to go, will need to wait for a few more long lines at the market--but I'm pretty sure it'll go that way. This novel is like a string tied at both ends. It has points of suspense--as in Dr. Jekyll and Mr. Hyde, I don't know how it's gonna get to where it's going--but the basic structure is clear.

One other thing--reading these oldstyle books often diminishes my appreciation of newer works in the same genre. The original can be so much fresher.

"...it has become standard in climate science that data in contradiction to alarmism is inevitably 'corrected' to bring it closer to alarming models. None of us would argue that this data is perfect, and the corrections are often plausible. What is implausible is that the 'corrections' should always bring the data closer to models." - Richard Lindzen, MIT Professor of Meteorology

Background:

Back in 2002, researchers at NASA published a paper entitled "Evidence for large decadal variability in the tropical mean radiative energy budget" (Wielicki et al., Science, 295:841-844, 2002). The paper reported data from a satellite that measures solar radiation headed towards earth, and reflected and radiated energy headed away from earth, and thereby measure the difference in incident and outgoing energy. The data reported in the paper showed that outgoing energy climbed measurably in the late 1990s, in contradiction to the assumptions of predictions from climate models that assume positive or near-zero "climate feedback."

It keeps on coming

I just received the following email:

Dear Prof. Gelman,

I am sending you the latest work by Prof. XXX (whose work was mentioned recently on the blog), this time on the intelligence of liberals and atheists. Perhaps this will be of interests to readers of your blog.

I do applied work in neuroimaging analysis.

Yours,
YYY
University of ZZZ

I replied with this link.

P.S. I was surprised at how far down the page I had to go to find that entry, which I felt I'd written only yesterday. I'm definitely blogging too much.

From the New York Times, 9 Sept 1981:

IF I COULD CHANGE PARK SLOPE

If I could change Park Slope I would turn it into a palace with queens and kings and princesses to dance the night away at the ball. The trees would look like garden stalks. The lights would look like silver pearls and the dresses would look like soft silver silk. You should see the ball. It looks so luxurious to me.

The Park Slope ball is great. Can you guess what street it's on? "Yes. My street. That's Carroll Street."

-- Jennifer Chatmon, second grade, P.S. 321

This was a few years before my sister told me that she felt safer having a crack house down the block because the cops were surveilling it all the time.

I love reading old book reviews

I encountered this book in the library, "Homage to Qwert Yuiop: Selected Journalism 1978-1985," by Anthony Burgess. It's just great. I think somebody should collect and print the rest of Burgess's journalism too. (There do seem to be one or two other collections out there but I doubt they have the density of this overflowing volume of 589 pages of small print. I have a feeling they're leaving out a lot of good stuff.) I'd also gladly buy an edition of the uncollected book reviews of Alfred Kazin or of Anthony Boucher. I'm sure a lot of editing would be required, though, but I think it would be worth it.

It would be fun to spend some months going through old book reviews from the 1950s-1980s and collecting the most interesting parts for a collection. But I suspect that almost nobody but me would be interested in such a thing.

P.S. The library had another book of Burgess's previously uncollected essays--a collection prepared posthumously. I was surprised to find that I did not like these essays at all! The problem, I think, is that the editors tried to collect what seemed to them to be the most important of his works, which led to the inclusion of a lot of pretentious crap. I think Burgess's workaday book reviews were much more interesting and judicious than these pieces where he had more freedom to say what he wanted.

To commenters

I often (although not always) respond, sometimes directly, other times with a longer response to several different comments in the thread. So please check back after a day or two to see if I answered your question. I'd hate to think that this effort was all wasted.

Note to blog: Add quizzes

After seeing the 200 or so responses to this entry by Tyler Cowen, I'm thinking that maybe we could get some more participation by adding reader-response activities.

For example, I could've posted this, but replacing the three blogs mentioned there by X, Y, and Z. I'm pretty sure no one would've guessed X or Y, let alone the completely obscure Z. An opportunity wasted, I suppose.

P.S. What Cowen wrote was:

I [Cowen] read many blogs but there is only one . . . which I find truly obnoxious. . . . I never read it regularly in the first place, but I don't think I can stand to read this blog any more ever again.

It's hard for me to relate to this. I've surely read many fewer blogs than Cowen has, but of the ones I've read, there are a lot that I find so obnoxious that I have decided to never read them again. Maybe Cowen's read enough blogs that mere everyday bloggy obnoxiousness doesn't bother him anymore.

A huge ad in the subway . . .

. . . for Paul Auster's new book.

Only in Paris, huh?

How hard is it to say what you mean?

Deep in a long discussion, Phil writes, in evident frustration:

I don't like argument by innuendo. Say what you mean; how hard is it, for cryin' out loud?

Actually, it is hard! I've spent years trying to write directly, and I've often noticed that others have difficulty doing so. I always tell students to simply write what they did, in simple declarative sentences. (It can be choppy, that's fine: "I downloaded the data. I cleaned the data. I ran a regression" etc.) But it's really hard to do. As George Orwell put it, good prose is like a windowpane, but sometimes it needs a bit of Windex and a clean rag to fully do its job.

P.S. I feel similarly about statistical graphics. (See also here.)

Some thoughts on ethics

Kaiser writes:

If everyone agrees that taking Avandia, the blockbuster G.S.K. diabetes drug, increases the risk of heart attacks, what's the problem? Why is Avandia still being prescribed? Read this New York Times article [by Gardner Harris] to find out.

It's pretty amazing:

Three years ago, Dr. Steven E. Nissen, a cardiologist at the Cleveland Clinic, conducted a landmark study that suggested that the best-selling diabetes drug Avandia raised the risk of heart attacks. . . . on May 10, 2007, 11 days before Dr. Nissen's study was published in The New England Journal of Medicine, he and four company executives met face to face in a private meeting whose details have not been disclosed until now.

Fearing he would face pressure and criticism from executives, Dr. Nissen secretly recorded the meeting -- which is legal in Ohio as long as one party to the conversation is aware of the taping. . . . GlaxoSmithKline had threatened scientists who tried to point out Avandia's risks . . . Dr. Ronald L. Krall, GlaxoSmithKline's chief medical officer, predicted almost exactly the results of another crucial study of Avandia that was two months from publication and whose results, according to scientific protocols and the company itself, should have been kept secret from the company. In an interview, Dr. Nissen said the recording showed that the executives hoped to persuade him not to publish his study by suggesting that they had contradictory information they would share with him in a joint study.

"In retrospect, it seems clear that neither statement was true," Dr. Nissen said. "They did not have contradictory data, and they never intended to cooperate in any analyses."

But I guess I shouldn't be surprised:

This article by Ryan Streeter summarizes some demographic trends highlighted by demographer Joel Kotkin which he describes as showing that "America is evolving in a conservative direction." I'm not quite convinced by that, but I do think these statistics give some insight into how the Republican Party maintains its strength among high-income Americans.

Colorless green ideas sleep furiously

Kaiser nails the silliness.

Our story begins with this article by Sanjay Kaul and George Diamond:

The randomized controlled clinical trial is the gold standard scientific method for the evaluation of diagnostic and treatment interventions. Such trials are cited frequently as the authoritative foundation for evidence-based management policies. Nevertheless, they have a number of limitations that challenge the interpretation of the results. The strength of evidence is often judged by conventional tests that rely heavily on statistical significance. Less attention has been paid to the clinical significance or the practical importance of the treatment effects. One should be cautious that extremely large studies might be more likely to find a formally statistically significant difference for a trivial effect that is not really meaningfully different from the null. Trials often employ composite end points that, although they enable assessment of nonfatal events and improve trial efficiency and statistical precision, entail a number of shortcomings that can potentially undermine the scientific validity of the conclusions drawn from these trials. Finally, clinical trials often employ extensive subgroup analysis. However, lack of attention to proper methods can lead to chance findings that might misinform research and result in suboptimal practice. Accordingly, this review highlights these limitations using numerous examples of published clinical trials and describes ways to overcome these limitations, thereby improving the interpretability of research findings.

This reasonable article reminds me of a number of things that come up repeatedly on this blog and in my work, including the distinction between statistical and practical significance, the importance of interactions, and how much I hate acronyms.

They also recommend composite end points (see page 418 of the above-linked article), which is a point that Jennifer and I emphasize in chapter 4 of our book and which comes up all the time, over and over in my applied research and consulting. If I had to come up with one statistical tip that would be most useful to you--that is, good advice that's easy to apply and which you might not already know--it would be to use transformations. Log, square-root, etc.--yes, all that, but more! I'm talking about transforming a continuous variable into several discrete variables (to model nonlinear patterns such as voting by age) and combining several discrete variables to make something continuous (those "total scores" that we all love). And not doing dumb transformations such as the use of a threshold to break up a perfectly useful continuous variable into something binary. I don't care if the threshold is "clinically relevant" or whatever--just don't do it. If you gotta discretize, for Christ's sake break the variable into 3 categories.

This all seems quite obvious but people don't know about it. What gives? I have a theory, which goes like this. People are trained to run regressions "out of the box," not touching their data at all. Why? For two reasons:
1. Touching your data before analysis seems like cheating. If you do your analysis blind (perhaps not even changing your variable names or converting them from ALL CAPS), then you can't cheat.
2. In classical (non-Bayesian) statistics, linear transformations on the predictors have no effect on inferences for linear regression or generalized linear models. When you're learning applied statistics from a classical perspective, transformations tend to get downplayed, and they are considered as little more than tricks to approximate a normal error term (and the error term, as we discuss in our book, is generally the least important part of a model).
Once you take a Bayesian approach, however, and think of your coefficients as not being mathematical abstractions but actually having some meaning, you move naturally into model building and transformations.

P.S. On page 426, Kaul and Diamond recommend that, in subgroup analysis, researchers "perform adjustments for multiple comparisons." I'm ok with that, as long as they include multilevel modeling as such an adjustment. (See here for our discussion of that point.)

P.P.S. Also don't forget economist James Heckman's argument, from a completely different direction, as to why randomized experiments should not be considered gold standard. I don't know if I agree with Heckman's sentiments (my full thoughts are here), but they're definitely worth thinking about.

When I saw this op-ed by Phil Kiesling the other day (recommending nonpartisan primary elections as a way of reducing polarization in Congress), I had several thoughts.

John Sides offers a more thorough, research-based discussion of the effects of open primaries, which I'll discuss below. But first my immediate reactions to Kiesling's op-ed:

  1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48  

Recent Comments

  • Anonymous: I live in Washington State where we currently use the read more
  • Sean Walker: @Ian I agree that changing the voting system is the read more
  • Bill Jefferys: "My personal preference would be to just dispose of the read more
  • Bill Jefferys: "There's a theorem..." Indeed, and Kenneth Arrow won a Nobel read more
  • Ted: My personal preference would be to just dispose of the read more
  • Ian Fellows: @Cody That just moves the spoiler effect problem to the read more
  • Ian Fellows: I agree with the above posters. Changing the primary system read more
  • Cody Custis: @Anne The editorial is advocating a 'non partisan blanket primary' read more
  • Janne: You're attacking the wrong problem. The issue is that your read more
  • Bob O'H: Looking from abroad, this is just bizarre. The purpose of read more
  • Anne: Point 4 is exactly why primaries exist, isn't it? So read more

Find recent content on the main index or look in the archives to find all content.