Posted tagged ‘scientific thinking’

How to Think: Public Policy edition.

May 27, 2008

One of the running strands of this blog is the notion that the public’s interest in science is at least as much in the ways scientists think as it is in the particular discoveries that emerge over time. Not that the latter are unimportant — far from it: they are rather the currency with which science buys and holds the attention of the culture that supports it.

But to take a recent example, the uncovering of a fossil animal intermediate between a fish and a land-living creature was the fact that got Neil Shubin’sYour Inner Fish off and rolling. But the real story Shubin told, excellently (despite this snark), was the process by which Shubin and others put themselves in a position to anticipate and appreciate the significance of that fossil.

It’s that old chestnut: “The King died; the Queen died…” is a list of facts. “The King died; the Queen died of grief…” is a story. The story to be told by science writing is one that allows its reader to enter into the means of discovery, ideally in ways that such a reader can use, even if she or he never confronts a limbed fish.

More specifically, I’ve emphasized a couple of attributes of science that need to get more play in our broader culture — empiricism (rigorous observation and experiment) and abstraction, by which I usually mean some kind of quantitative analysis.

Then along comes my younger brother, Leo. He’s a senior civil servant for a California county, running a huge budget. On the side he teaches a course in public policy at a local college.

This year, a student’s question prompted him to put what he hoped his class had learned into capsule form — and in a few words it captures what I’ve been trying to say in too many more, lo these many months. He wants his students to impose as much discipline as possible on what they think we know. It should not be limited to those few fortunate enough to learn from him. Here’s his valedictory to this year’s class:

1. When people are talking about public policy issues, always remember the question—what is the underlying problem or opportunity they are proposing to address? How are people modeling the problem in their minds (often unstated)– – as to what are the problem’s causes and what could various interventions hope to achieve? Once you state the underlying assumptions out loud, do they make sense; are they reasonable? So often we jump to a particular solution and advocate it without stepping back to think whether there might be other ways to address the same issue

2. Quantify wherever possible so that you think about how big the problem really is and how much difference you can expect from the different approaches people may take. If you have no way of measuring the problem, you are unlikely to be able to prove to people tackling it is worth the effort, and no way of judging the success of any pilot programs you might have a chance to implement.

3. Clarify the trade-offs between doing nothing and the various alternatives. Whether a public policy approach is worth doing can’t be answered without comparing it to something else. You need clear criteria and to compare approaches against each other as to how well they meet your criteria. Just advocating a course of action by itself is not convincing unless you can compare the outcomes to doing nothing or other alternatives.

Amen and Amen. Here endeth the lesson.

Image: Edward Hopper, “The El Station,” 1908. Source: Wikimedia Commons.

Why Can’t Republicans (and Harvard Economists) Count? Housing edition

March 8, 2008

I’ve focused a lot on the importance in thinking in numbers in a variety of blog posts.  (This one is my personal favorite). As I’ve done so, I’ve emphasized that this kind of thinking is one of two real pillars of scientific thinking. (The other one is empiricism — actually going out and in ways you can check getting information about the real world.)

The larger point I keep sniffing around is the notion that this is what a real definition of science literacy means: it’s not what facts you know (or think you know — see this post for a gory view of truthiness in science). Rather — its how you approach facts as you learn them, what sense you or I make of our experience that counts.

Counts — there’s the word again. Apparently uber-economist Martin Feldman, late of Ronald Reagan’s administration and now professing to unsuspecting Harvard undergraduates, doesn’t do that so good. He’s got a nifty proposal to address the mortgage crisis in America — a massively complex scheme of government intervention and subsidy (waittaminute — ain’t that for Atrios’s DFHs?) that will, in the end, in the real world, add up to…

Bupkis. Tanta over at Calculated Risk has run the numbers. Putting the absolute best possible framework around Feldman’s idea (he wants the feds offer a 15-year second mortgage loan at a highly subsidized rate, with a number of restrictions, to cover 20% of existing mortgages), Tanta works out what all the details actually mean.

You can mess about a bit with the assumptions in the examples worked out there, but the bottom line remains the same. The sucker don’t work. Plausibly, it will increase monthly payments for many borrowers (total interest will go down; but the real-world economic crisis derives from the fact that folks can’t pay what they owe now, not fifteen years down the road). One case study ends up with a home owner forced to buy 12 fewer lattes per year … which, as Tanta notes, hardly advances the cause of economic stimulus.

Not to spill two many bytes on this — after all, this is a proposal so dumb it has nowhere to go, despite the bar being set pretty low on stupid over the last several years — but why is this so hard to figure out?  Feldman can in fact do his sums — I’m sure.  Why not actually run a few tests against his hypothesis (subsidizing a fraction of mortgage interest costs will make a difference to the economy — yes or no?) and quietly trashcan the idea himself, without wasting time the rest of us could use …say … meeting the book deadline whose breath I feel hot against my neck.

Count, man! Count.  (You’ll still respect yourself in the morning.)

(h/t Atrios)

I don’t know nuthin’ ’bout economics, but…: NPR/Henri Poincaré/Mortgage follies edition

February 25, 2008

Innumeracy is a problem I have and will come back to a lot here. But as I listen to more and more popular presentations of technical subjects, I still get astonished by the intersection of two structural problems in the media.

That is: many reporters — not so high a proportion of self-described science writers, though still plenty there — have trouble with even the most elementary uses of quantitative approaches to their stories because they just don’t think in numbers at all. That’s the negative way of framing the problem; journalists have a lack that inhibits their capacity to do good work in an ever-more technically imbued world.

Then there’s the affirmative problem. Reporters establish stories by anecdote, by individual bits of data, single episodes. They’re called stories for a reason: the goal is to perform one of the most powerful acts of communication humans have figured out, to convey information that compels belief because its hearer can place themselves right into the narrative.

That’s why, to edge closer to the real subject of this post, so much of the reporting on the mortgage crisis (fiasco) centers on some family that’s about to lose a house, and spend little time, on the meaning of the big numbers, like the implications of a repricing of US housing on a large scale.  The point is that not only do many journalists not know a set of ideas that could help them figure out such things;  what they do know leads them away from the kind of approach to their work that more mathematical sophistication would provoke.

But there’s a wonderful passage that bears on this from the great French mathematician Henri Poincaré in a collection of essays that greatly influenced the young Albert Einstein:

We can not know all facts, and it is necessary to chose those which are worthy of being known.

Choose? Worthy? Surely Poincaré is not going prematurely po-mo on us here?

Not really. The notion embedded in his deliberately provocative turn of phrase is that facts need form, some apparatus that can incorporate a given datum into a richer story — one with a meaning larger than that of a single incident. That apparatus is quantitative.

(BTW — I use the word “quantitative” rather than mathematical, because for a great deal of human experience, the math needed to make sense of what’s going on is not that complicated.  It’s often a matter of counting, sorting, and extracting relationships within the formal limits of what you learn by the end of high school.  I have posted on a couple of such examples from great scientists — Freeman Dyson, for one, and J.B.S. Haldane for another.  There are lots more — perhaps readers could be persuaded to post examples of what they think are elegant, simple insights a bit of math can give us ?)

All of this  into mind while I listened to NPR this morning.

This is the story that got me going — a short (1 minute, 10 seconds) reporter-voiced account of what seemed to the Morning Edition team to be something strange: Even though the Fed is cutting interest rates, mortgage rates went up sharply last week. That ain’t how its supposed to be, according to the reporter, Adam Davidson, because when the Fed lowers its rates, other rates are supposed to drop.

The reason Davidson gave for what he saw as weird is not all wrong: he said that lenders are newly afraid of inflation, and hence want to charge a higher price for money that is going to be paid back over time.

But look at the unexamined assumption: that the Fed can control rates in general. That’s not true.

What’s missing here? An understanding of the real importance of time.

The Fed mostly exerts its influence on interest rates through the shortest of short-term instruments, the overnight federal funds rate — which is just the price banks pay for extremely brief loans required to keep their minimum reserves up to snuff.

But real people borrow money for houses on long time scales, most famously through 30 year mortgages. The enormous difference between the types and uncertainties of risk between those two scales of time serve at least partially to decouple the two rates — see the data to be retrieved here for a survey view of this.

So it is true that fear of inflation could keep push term rates up, whether or not the Fed was playing around with short term rates. But so could lots of other things.

Perhaps that the value of US real estate is unclear in a falling market, and thus lenders demand a risk premium before they lend against such difficult-to-value assets. Perhaps the overall credit worthiness rating of American real estate borrowers has dropped in the aggregate.  Perhaps lenders fear that the secondary market for mortgages is going to get a bit less liquid.  Lots of factors play into long term interest rates that have nothing to do with the reasons the Fed makes its interest rate decisions.

In other words: and the NPR story was either meaningless or misleading. And it failed because the reporter glossed over or did not fully understand what the mortgage rate summarizes as a single number — all the complex calculations of risk and profit that underpin the decision of whether or not to make a loan.

What I would have loved to hear instead of a “this fact is strange” report would be that story: how do interest rates express quantitatively our ideas about the future.  It’s still a good, fully human story:  Those numbers tells us a tale about what we think we know about what’s coming down the pike — and how much in dollars and cents we fear changes in our perception of what we don’t know.

Image: Rembrandt van Rijn, “The Money Changer,” 1627. Source: Wikimedia Commons.