Tuesday, February 16, 2010

Truth With An Error Range.

A recent article by Megan McArdle cause a mini storm in the health blogging circles. Moved by the assertion made by Ezra Klein back in the fall that Joe Lieberman was willing to cause the deaths of hundred of thousands of people by threatening to block reform, McArdles goes on to examine the link between mortality and health insurance and after finding various reasons to ignore or disqualify various studies on the issue, she concludes that while there might be some effect, it's hard to tell how big it is and because it's hard to measure exactly how big it is, it's probably too small to matter.

Needless to say, McArdle has been attacked from all sides (actually, mostly from the left) for this article and it's easy to see why. While her article doesn't insist that health insurance is irrelevant and she does say "even if we did agree that insurance rarely confers significant health benefits, that would not necessarily undermine the case for a national health-care program" by launching a pedantic attack on a statistic, she helped to undermine what many would consider a "logical" case for expanding health coverage to everyone.

There are two reasons why I find this whole debate incredibly annoying and offensive.

First of all, by making a reasonable case that the effect of health insurance on mortality are hard to determine precisely she then goes on to argue that we can't say how large it is. If one study says 50, another study says 200, it could probably as well be 0, seems to be the argument. The truly annoying thing about this argument is that from everything McArdle says, she doesn't seem to actually believe that there is no relationship, but she is suggesting that it might be small and either way no one can claim with any certainty what exactly it is.

I guess McArdle deserves credit for correctly identifying a problem with the studies of this sort - they aren't perfect and they aren't precise. However, does it mean they are thus entirely irrelevant? That's what she seems to be suggesting in her concluding paragraph: "we should have had a better handle on the case for expanded coverage—and, more important, the evidence behind it—before we embarked on a year-long debate that divided our house against itself." Translation: why didn't you pro-reform morons get your facts straight before wasting everyone's time with this non sense for a whole year?

Therein lies the insane contradiction of what McArdle is saying: on the one hand, she discredits the assertion that expanding health insurance saves lives by showing how evidence is hard to attain; on the other hand, she says we need better evidence to show exactly how many lives were saved by health insurance. Unless we have an exact measurement, it's useless; if it pretends to be exact and precise it's probably wrong.

McArdle's sophistry is equivalent to asking a man on a boat in the middle of an ocean how deep the water underneath is. The man clearly sees that the water is very deep but he has no good way to measure it until he realizes that he can tie a heavy rock to one end of a thread, drop it in the ocean and measure the length of the thread until the rock hits the bottom. Of course the bottom of the ocean is uneven, so multiple measurements will result in different results. Faced with this imperfect technique and variation in results, McArdle would probably say that we can't conclude that the ocean is very deep. Even though it is intuitive and obvious to all the rest of us that the ocean is very deep, because one can't measure exactly how deep, she would suggest there's some likelihood it's shallow, and maybe we're all just imagining the ocean in the first place. In the case of health care, McArdle's logic would suggest that because we can't exactly measure the link between insurance and health outcomes, we shouldn't be so obsessed with providing insurance to everyone. In the case of our man on a boat, McArdle would probably send the man without an oxygen tank to find something on the bottom of the ocean because he can't prove exactly how deep it is. Despite the preponderance of logic, common sense, intuition and countless anecdotes of people who lose their lives because they can't get the care they need, McArdle is totally unmoved and declares the deep ocean of the problems that uninsured have a mere mirage.

The other reason why this debate is incredibly annoying is because it's entirely besides the point. Yes, it would be great if we could say with total certainty how many people die each year because of lack of insurance and it would be awesome if we could therefore accuse someone blocking reform of killing hundreds of thousands of people. But even in the absence of that, the case for reform is no weaker. In fact, disproving (or undermining) any one "estimate" does not change the fact that the system is incredibly broken and unfair. And while it might be fun to play around with statistics to make a case one way or the other, the truth has an error range and no estimate or projection can be relied upon with absolute certainty. Ultimately, this debate is about values and trade-offs and I wish those hiding behind pseudo-statistical analysis would debate it as such.

No comments:

Post a Comment