• The Atlantic’s Contrarian Problem

    I’ve subscribed to The Atlantic for over 15 years. Some of the best things I’ve read have appeared in that periodical. I’ve learned a lot from it, or so I thought. But the recent article by Megan McArdle on the relationship between insurance and mortality begs the question whether I’ve actually learned anything (related posts).

    Because it’s in an area related to my own work, it is clear to me and should be to any health economist or health care researcher that McArdle’s scholarship was not up to snuff. The narrow slice of the literature she cited supported her conclusion but was not representative of the whole, which paints a completely different picture. And from that misleading and narrow slice her absurd conclusion followed–that we should have known more about the insurance-mortality relationship before we considered health reform.

    [Read about McArdle’s update of her thinking on this here.]

    One need only take an expansive look at the literature to reach the established conclusion that insurance improves health and reduces mortality. Michael McWilliams has done that thorough examination, on this blog, in his original published and peer reviewed work, and in a comprehensive literature survey. Stan Dorn has provided another blog-length literature review. McArdle’s falls fall short by comparison.

    McArdle’s piece is typical of a subset of articles in The Atlantic, the ones in which a journalist takes a look at a few papers, perhaps talks to a small number of researchers (or maybe not) and then spins a contrarian story. Such pieces have the veneer of legitimacy and seem to have the hallmarks of good scholarship. But in reality they posses neither.

    The next time I read such a piece in The Atlantic, can I believe it? If I didn’t know better I’d have believed McArdle’s. Perhaps many readers do or will. That may be a victory for McArdle and The Atlantic, but it is a defeat for truth and honesty.

    What’s the point of articles that cherry-pick research to support contrarian conclusions? They may be provocative, but a misleadingly narrow review of a body of research often can be. Being provocative is not necessarily bad, but doing so in a manner that misrepresents a body of work is an insult to science and does readers a disservice. In what way is that useful, unless your goal is to be misinformed and misled (as a reader) or to support otherwise unjustified policy conclusions (as an author)?

    I may have been an Atlantic subscriber for a long time, but if they print more pieces like McArdle’s I won’t be one for much longer.

    Share
    Comments closed
     
    • As one of the researchers who has followed this ONLY on your blog, and you and Michael have been eminently sensible while preaching to the choir, I find it difficult to believe the blogosphere has found this controversial. I am dimly aware of Megan McArdle’s general career writing for the Economist and the Atlantic, with the usual background of English/MBA, but not a deeper understanding of research methodology. In fact, of course, most researchers don’t have a good grasp of general issues of research methodology. This is hard, and thus it is not surprising that “Burgess’s Law” (newly coined here) that the more complex the issue (complex in the Dave Snowden sense such that it is impervious to simplification by modeling) the less the evidence based practice comes into play. Since in general, more health problems are complex than in most other fields, this is why evidence based practice in health is more rare, but easiest in considering broken arms than in mental health. What is the effect on mortality of care for depression? How much data does one need to address the question? What severity of depression are we talking about? One can barely even conceive how one would actually approach knowing anything on this.

      So, I guess I’m wondering two things, for those of us who love your blogs but don’t have time to follow everyone else’s — you get regular 6AM reads from me. One, is the Atlantic really standing by this story (I don’t think honestly that they intentionally fall in this trap, but they believe fairylike in the ability to model problems down to basic communicable points — sometimes they can and articles are good, and sometimes they can’t)? I did a little quick searching and couldn’t tell. They should be willing to re-address or back off on this point. Two, you allude to the idea that the blogosphere is mixed on this, without diving into it (again, a quick search agreed that there seems to be dizzying amount of politically motivated commentary), is there anything interesting in that debate or do people just line up on the political grounds of “I believe what I think about whether health insurance should be expanded to everyone and then I choose to believe the evidence that backs my opinion”. To me, the expansion of insurance issue is mostly a question of simple social justice and depends on one’s view of the health and welfare of one’s fellow citizens, on which we are mostly ambivalent. The rest of the arguments are thinly veiled arguments stemming from those underlying societal preferences.

      • @Jim Burgess – I haven’t seen anything interesting on the anti-reform side that both (a) makes sense, (b) is politically feasible, and (c) isn’t clearly a veiled attempt to maintain the status quo. Then again, I don’t read deeply into the anti-reform blogosphere. I do read Cowen, sometimes Hanson, and rarely McArdle. My usual sources are here: http://theincidentaleconomist.com/sources/

        Even though they are pro-reform, I do find that Drum, Cohn, and Klein are largely honest about the issues and the limitations of the legislation. Nobody who is intellectually honest will say it is perfect or is what they’d implement if they were king, and those three don’t say those things. Mostly, they and I feel it is far better than the status quo and would take some positive steps. But it isn’t the end of the necessary reforms. It is just the beginning.

        I’m afraid I can’t produce what you want here. If somebody can point to a more thorough synthesis of the pro-con on this, please do!

    • I hate to be a contrarian – but here is the abstract for the peer reviewed article you linked to..

      “Uninsured adults have less access to recommended care, receive poorer quality of care, and experience worse health outcomes than insured adults do. The potential health benefits of expanding insurance coverage for these adults may provide a strong rationale for reform. However, evidence of the adverse health effects of uninsurance has been largely based on observational studies with designs that do not support causal conclusions. Although recent research using more rigorous methods may offer a better understanding of this important subject, it has not been comprehensively reviewed.”

      Does the author really believe that ALL uninsured adults do not have access to “recommended care” [whatever the heck that is] – surely some percentage of those without insurance earn enough to pay for any care they require.

      And does the author really believe that just having health insurance leads to health – the recent death of John Murtha would seem to provide anecdotal evidence that having insurance can lead to premature death due to medical error [see the recent announced investigation into the causes of Congressman Murtha’s death]

      I think Megan McCardle and Tyler Cowen have done a great service by asking some very tough questions – which you and others answer obliquely at best – Dr. McWilliams makes a great case for providing care for chronic conditions like hypertension and diabetes – which would cost much less than the proposed health care reforms – but I did not find his article convincing on the general idea that insurance makes us healthier overall.

      The Atlantic is not a peer reviewed journal – and has no pretensions of becoming one. Ms McCardle did a great job in the context of asking for more data before taking such a huge step in transforming our economy and government.

      • @Lonely Libertarian – My advice to you, Megan, everyone, is to (a) read the literature, all of it, or (b) read a good summary of it (I’ve cited and posted some). If you can make a good case based on science for your position, good for you. Write it up, post it, publish it, do as you like. If not, well, I’m not going to repeat myself. I respectfully ask that you do not as well.

        (If you really need things repeated, see http://mdcarroll.com/2010/02/18/im-with-austin/ )

    • Thanks for advocating strongly for integrity, Austin.

    • I do really want to know what the effect is and it’s size – here is an article that did something like what I suggested earlier – looked at comparative populations of Christian Scientists and non – it does show a statistically significant difference in death rates between the two groups – much better proof of your case IMHO…

      Comparative longevity in a college cohort of Christian Scientists.
      Simpson WF.
      JAMA. 1989 Sep 22-29;262(12):1657-8. Erratum in: JAMA 1989 Dec 1;262(21):3000.
      PMID: 2769921 [PubMed – indexed for MEDLINE]

      That being said, I suspect some would be critical of this study as being flawed – it is not about insurance – but types of health care – CS versus “normal” and it is a bit dated.

      • @Lonely Libertarian – Now you’ve impressed me. I still do not believe this should be viewed as definitive. Nor should any study. I’ll see if I can get hold of this one and take a look though. Thanks.

    • The Ticker at The Atlantic says “Austin Frakt thinks The Atlantic has a contrarian problem, though in reality it’s just the opposite” and links to this post.

      I don’t get it. What’s the opposite? The contrarian has an Atlantic problem? The Atlantic has a contrarian solution? The Atlantic thinks I have a contrarian problem? I guess that makes sense.

      Still, now I think The Atlantic has a clarity problem.