Is Science Predictive?



Credit: Sam Wong, Princeton Election Consortium

First, a word of warning to the folks at home. This is going to be explicitly political. It is also going to be respectful. If you want to send or post some sort of response, I ask that you do the same.

As I write this, I’m hunkered down in Philadelphia, waiting for the full onslaught of Hurricane Sandy. It may be devastating, but who knows? I can’t help but think about how often I simply take other people’s scientific expertise for granted. I mean, it’s just plain prudence to accept that meteorologists will have a pretty good idea of how weather patterns work. If all of the warnings turn out to be a bit overblown (so to speak) this time around, where’s the harm? Better safe than sorry.

And prediction is like that. The New York Times columnist, Nate Silver, has a new book out describing how difficult it can be to make predictions in various areas. While it’s a bit self-aggrandizing, on the whole, it’s a pretty good read.

Science — and prediction in general — is tougher than it might appear, even to a very intelligent, fairly educated layperson. And yet, when we read about predictive behavior (in polling, evolution, climate, or even into which scientific theories are likely to be proven correct), there’s a kneejerk reaction that somehow these “experts” must be overlooking something obvious. I’ve seen it myself when writing about Dark Matter and Dark Energy, for instance, topics on which I’m an expert but most of my readers are not (check the comments sections of the linked articles if you don’t believe me).

Skepticism is a good thing in general, but there’s a point at which you’re simply being contrarian, and perhaps you should consider the possibility that you are arguing for what you’d like to be true, rather than what is supported by the evidence.

Take, for instance, the current election. I haven’t made any bones about the fact that I’m a lifelong liberal, and a strong supporter (on enough issues) of President Obama. I, like many on the left, have some very serious concerns about the pushback against science from the right. It isn’t simply a Congress that has demonstrated a willingness to slash funding for basic science or a presidential candidate who doesn’t understand the role of basic science in the academy. Rather (and I fully recognize that this isn’t a unique or new observation) the modern Republican party seems to cater to the idea that scientists who’ve spent years studying a particular problem have no freaking idea what they are talking about — or worse, that scientists have some sort of ideological agenda in presenting their results.

As I said, I’m a liberal, and it would come as little surprise that the majority of prominent science academics are (if not liberal) at least supportive of the president. You may have seen the letter in which 78 former Nobel Laureates endorsed Obama for reelection.

I’m not going to go into yet another tired tirade about how important science education is. Clearly I feel that it is; and both my choice of career and my writing on the side should demonstrate that I take science education education fairly seriously. If you’re reading this, odds are you feel the same way.

And while there are some outstanding scientists with conservative leanings, the fact remains that the Republican party routinely campaigns and legislates with profound skepticism toward science. I could point to individual examples of course. Todd Akin and his ludicrous understanding of reproduction (and the ensuing repercussions for policy) are among the best example from this election cycle. Sarah Palin was perhaps the best example from the last.

But it’s worse than that. The very narrative of the election cycle has been dominated by skepticism that people don’t know how to do their jobs. The funniest and most egregious has been the work of “Unskewed Polls” which suggested that pollsters couldn’t be trusted to even try to produce accurate predictions of the election. But why would anyone think that? Never mind that there are just as many pollsters (if not more) with conservative leanings than liberal ones; the real issue is that these guys want to be right. It’s how they earn their living.

The current consensus of state polls suggests that while the election will be close in terms of total votes, unless something significant happens in the next week or so, it is very likely that Obama will be reelected. Nate Silver has written extensively about the predictive power of the average of state polls, and my old pal (and, if must be said, Red-Stater) Wes Colley has done median statistics analysis of state polls. Any reasonably unbiased view of the polls suggests that there simply isn’t a very great likelihood of Romney winning enough states to put him over the top. There are lots of others as well, and all pretty much come to the same conclusion. Obama is very likely to win Ohio, Nevada, Wisconsin, New Hampshire and Iowa — and reasonably likely to pull out Colorado and Virginia as well. This puts him comfortably above the 270 electoral votes required for reelection. To conclude otherwise you either need to be willfully partisan, or simply believe that pollsters don’t know how to do their job. At all.

One reason I wanted to write this post now and not wait until after the election is that I am putting at least a little credibility on the line by making a prediction rather than a postdiction. The other reason is to point out that this skepticism from the right — dismissal of the data to get the answer that they’d like — is profoundly anti-scientific and is of the same sort that causes them to ignore all sorts of data.

A few weeks ago, for instance, there was an announcement from the Bureau of Labor Statistics that unemployment had dropped to 7.8%. As the BLS will freely admit, this number (much like polling numbers themselves) this is based on polls of companies and other measures that have uncertainty. Yes, the number contains noise. But it always has. That didn’t stop those on the right from insisting that the numbers had been cooked. This skepticism showed up on Fox, CNN, and other media outlets, despite there being no evidence that the BLS had changed their procedure or hadn’t done their job. The fact that they didn’t “feel right” shouldn’t negate the fact that somebody (in this case the BLS) has, you know, actually done the calcualtion.

While these electoral predictions (or the unemployment postdictions) have a fair amount of uncertainty, I hope they’re right for another reason beyond the reelection of Obama or lower unemployment being good news. I want further confirmation (in the form of the election) as a demonstration that experts actually know what they’re talking about.

I started by talking about the weather. While one cloudy day doesn’t make a Winter, there is some suggestion that these sorts of weather patterns are going to get more and more pronounced in the future. I’ve written about the difficulties in laymen assessing Global Warming before.

My point is that results of this election may (or may not) keep a party in power more likely to do something about Global Warming (or at least take heed to the evidence), but at very least I hope it serves as a reminder that people who devote their lives to becoming experts in a quantitative discipline might actually know what they’re talking about.

-Dave

This entry was posted in Uncategorized and tagged , . Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *