April 09, 2007
Cannon and Tanner Part 2
Michael Cannon has a defense of his op-ed up over at Cato's blog. He focuses mainly on the argument that health insurance isn't the most cost effective way to improve health, which is certainly a supportable point. But that wasn't the portion of his op-ed I primarily objected to. What angered me was the sentence that read, ""[y]ou may think it is self-evident that the uninsured may forgo preventive care or receive a lower quality of care. And yet, in reviewing all the academic literature on the subject, Helen Levy of the University of Michigan's Economic Research Initiative on the Uninsured, and David Meltzer of the University of Chicago, were unable to establish a "causal relationship" between health insurance and better health."
This is why I called the op-ed misleading, rather than a deliberate lie. What Cannon (and Tanner) are doing is exploiting the public's unfamiliarity with the concept of "causality." I'd bet a maxed-out HSA that most readers understood that passage to say that the researchers couldn't establish a relationship between health insurance and health. As I showed in my rebuttal, the study argued quite the opposite. The problem is, health insurance doesn't lend itself to randomized experiments that can demonstrate causality, and so there's no real way to test the assumption. The closest we've come, the RAND experiment, "provide[d] consistent evidence that health insurance improves health." Cannon and Tanner make it sound like the evidence suggests health insurance doesn't improve health. The study says the evidence suggests the opposite, but causality is difficult to discern because no one wants to deprive individuals of health insurance for the sake of experimentation. So their representation of the conclusion wasn't a lie, per se, it was just very, very misleading, and it was misleading in the direction of their argument.
April 9, 2007 | Permalink
question: why is his/their misdirection not a lie?
here are some of the dictionary definition of the word:
1. a false statement made with deliberate intent to deceive; an intentional untruth; a falsehood.
2. something intended or serving to convey a false impression; imposture: His flashy car was a lie that deceived no one.
3. an inaccurate or false statement.
4. the charge or accusation of lying: He flung the lie back at his accusers.
Are you saying they aren't being deliberate in their attempt to use misdirection to deceive through confusion?
This interest me because I see this as the primary issue with the left- why say misdirection when everyone else in the day to day world would call this lie?
Posted by: akaison | Apr 9, 2007 3:52:26 PM
As a philosopher, I'm becoming curious about what 'causality' amounts to in social science contexts like this one.
You shouldn't have to actively set up experiments to prove causation. We know that lunar eclipses are caused by the earth blocking sunlight and casting a shadow on the moon, but obviously we can't do experiments where we move astronomical bodies around. Observation that doesn't involve actively setting up an experiment should suffice.
Posted by: Neil the Ethical Werewolf | Apr 9, 2007 3:59:28 PM
To mislead people by any means is the same as lying. These bastards need to be called on their crap.
Posted by: MarvyT | Apr 9, 2007 4:12:53 PM
the causality in hard sciences are different from the social sciences. you can't have true causality in the social sciences because you can't observe society like you can the eclipsing of the moon. part of it is my view that in the social science there can be no true objectivity. its like that principle in physics where you cant both observe an event, and know where it is at the same time (or something like that). society is like that but more complex. we as participants can't fully know objectively whats happening in society. what we do doesn't effect the moon like policies that are constantly being changed and reshaped fundamentally so that what you started with has morphed into something else. the laws of nature of more simple in thier approach although hard for us to fully understand whereas social forces are not.
Posted by: akaison | Apr 9, 2007 4:26:48 PM
Social science is forever on the lookout for hidden "third variables" that might actually be causing the effect you're seeing. You think it's what you're trying to measure (whether having health insurance improves health outcomes), but it could be related to something else (how much money you have).
If someone were to look at data and say "see, people with health insurance have better health outcomes across the board" the obvious rebuttal would be "but people who have health insurance are also typically employed, earn at least a middle class salary, and do not live in highly stressful urban areas (i.e. bad neighborhoods). Any one of these variables could be explaining the variation, rather than the presence of health insurance." And they're right. Unless your study controls for those factors, you really don't know why people with health insurance would do better than those without. Is it simply because they're wealthier, and the wealthy do better than the poor in ALL aspects?
It's surprising to me that you'd have such trouble coming up with good data, though. All you'd have to do is get a sample of Medicaid-eligible people, and other people who are near-poor, but not poor enough to qualify for Medicaid. The poorer people with Medicaid should have better health outcomes than the slightly less poor people who do not have Medicaid. If that were the case, then you could prove conclusively that health coverage is beneficial.
If the Medicaid population comes out less healthy, you have tons of effects that could make that so, like the sicker people are the ones who go through the trouble of getting Medicaid, or the reason lots of people get that poor is because they have health issues, etc. But if it turns out the opposite way, that Medicaid-eligibles have better health outcomes than near-poor non-eligibles, then you have pretty conclusive evidence that health coverage improves health outcomes.
Posted by: spike | Apr 9, 2007 4:56:48 PM
And Ezra, you're right about how misleading that statement is. "You might think the uninsured forgo primary care but there is no causal link that's proven that the health outcomes are different"? In addition to the causation/correlation dichotomy, his argument is really implying that there's no causal link between primary and preventive care and health outcomes, which is even more misleading.
Posted by: spike | Apr 9, 2007 5:09:08 PM
Dear Neil the Philosophical Werewolf:
It is not possible to observe causality, experimentally or otherwise. Cause and effect are the bastard children of a marriage between analysis and narrative. Various analyses of what light is and how it works, and of the relation of cosmic bodies, lies behind the story of the earth passing between moon and sun.
Analysis, by itself, is always a priori, and a lot of work went into convincing people, through careful observation that, for example, the earth revolved around the sun, and the moon around the earth -- that is, that the analytical model proposed by Copernicus and revised by Galileo and Kepler, was, in fact, in some important and practical sense, the correct one, and the one, which we should use to interpret what we observe.
Social science is engaged in a similar enterprise. Various analytical models (or sometimes, just notions) may be proposed, which would establish a relationship between, say, having health insurance and the state of one's health, and then people try to make systematic observations, concerning whether any of those analyses are better than any other.
In the first instance, are people with health insurance measurably healthier on average (or in terms of some other statistical measure of general tendency) than people without? If so, is that because some people lose their health insurance when they get sick? Is that because people, who don't take care of themselves competently get sick more, and, not incidentally, don't buy health insurance? And, on and on.
In science, people usually have a surplus of speculative analyses and hypotheses. In social science, where human behavior is generally overdetermined, there is almost always not enough observable data.
Posted by: Bruce Wilder | Apr 9, 2007 5:18:32 PM
I disagree with Bruce- causality can be proven in the sciences- but not in the social sciences. Its the hallmark of science that any law must be based on proven repeatable events. Whereas in our imagination it maybe possible that X was not cause by Y, that's what we have math to get around. I suppose the point is that math isn't actuality, but that seems like an extreme requirement to prove something.
Posted by: akaison | Apr 9, 2007 5:23:39 PM
I just love it when well fed fat cat republicans who never had to worry about money a day in their lives dare to make statements about those of us who have to struggle with a budget and worry over the portions not covered by insurance.
Posted by: vwcat | Apr 9, 2007 7:52:04 PM
Please will you all repeat after me-- it's not the health of the uninsured, it's the wealth of the uninsured. The Cato always deliberately ignore the impact of the cost of care on those who don't have insurance-- it is of course the biggest problem and the cause of many bankruptcies, blah blah blah.
but Ezra there is no excuse for you ignoring the issue!
Posted by: Matthew Holt | Apr 9, 2007 9:33:33 PM
Posted by: judy | Sep 28, 2007 5:22:07 AM
The comments to this entry are closed.