Scientists and Politicians

(This column was first published in the January 20, 2003 issue of The Buffalo News.)

Are we experiencing global warming? Is air pollution an increasing health threat? Are our forests damaged by acid rain? Is the quality of our oceans declining? Does increasing human population pose serious problems? Or more generally, is our environment in trouble?

Those are important questions even in these times of global terrorism and the possibility of wars. And unhappily, the answers of scientific specialists comes down heavily on the response "yes" to each of them.

For example, two major studies reported in the January 2 issue of the British journal Nature strongly support global warming predictions. The data included "earlier frog breeding, bird nesting, first flowering, tree budburst, and arrival of migrant birds and butterflies," as well as new species colonizing "previously 'cool' regions" and the range of Arctic species contracting.

Despite this kind of evidence, such important environmental questions remain politically controversial. Citizens are perplexed by this situation and have every right to ask: Why is this?

I suggest that there are many reasons, but two are of special importance.

First, the answers to these large questions are not "settled" by individual experiments, the kind of experiments that are carried out in physics lab. Rather, they rely on correlational evidence. Consider those data of the Nature articles: the earlier arrival of bird migrants is happening while temperatures are increasing, but that doesn't establish that the heat is causing this change. Perhaps there are other causes: it might be, for instance, that there are simply more birders watching for early migrants.

My favorite example of how correlation can mislead is a "proof" that the best basketball players are those who play dirty. Take any basketball team records and list individual players' scores and fouls. You will find that the players who foul most often generally also score more points. But does that correlation prove that contention? No. The correlation derives from another cause. Better players play longer and thus also have more time to commit fouls.

Despite such examples (much beloved of tobacco companies) correlation remains a useful statistical tool and, especially when the evidence comes from a wide range of sources, it can become compelling. That is the case of the Nature articles that derived from hundreds of independent tests.

The second problem arises from the differing traditions of science and law. Scientists are trained to examine "X" and "not X", treating both equally in order to determine (whenever possible by experimentation) which is correct. On the other hand, most politicians are lawyers. As such they are trained to defend "X" or "not X"; that is, to marshal evidence for their preferred choice, disregarding questions of which is right.

And so we have conservatives starting from "no" answers to all of those initial questions and gathering arguments in support of those views. The best recent example of this is a 1998 book by Danish political scientist Bjorn Lomborg, The Skeptical Environmentalist. That title makes clear his posture.

But Lomborg has been in the news this month as well. A Danish Research Agency committee, whose members included many of that country's senior scientists, after spending a year studying his book has condemned it as "scientifically dishonest" and "clearly contrary to the standards of good scientific practice," their decision based largely on his selectivity of sources.

This action follows an earlier series of rebuttals in the January 2002 issue of Scientific American under the heading, "Misleading Math about the Earth."

No matter how strongly the evidence supports one position, however, usually today no action is taken. Instead, a committee is appointed to study the matter further.-- Gerry Rising