The Law Matters

This is a guest post by Michael Bailey and Forrest Maltzman, authors of The Constrained Court.  In November 2011, they predicted that the Court would uphold the Affordable Care Act, based on modeling of the justices’ ideological preferences and attitudes toward precedent.


After months of excruciating waiting, guessing, and waiting some more we now have heard from the Supreme Court on possibly the biggest case of a generation. We are definitely happy to claim that its decision is consistent with our prediction and our book.

There are many fascinating aspects of the case. First, it is simply not true that the Court is simply and exclusively a policy-oriented institution.  The Court is often political.  Some justices can be very political.  (Indeed, it is hard to read Scalia’s scathing attack on President Obama’s immigration policies as anything other than political.)  But the Supreme Court is not completely political.  Other factors – a complex combination of legal considerations and external constraints – can push the Court to do things differently than nine legislators would do.  Policy preferences alone do not dictate Court outcomes.

Second, this case reinforces our book’s argument about the idiosyncrasy of the Court.  We find again and again that justices differ in the degree to which any given legal or political factor matters.  We can partially predict this variation—justices with previous bench experience are more likely to respect precedent; justices appointed around the New Deal era were more likely to defer to Congress—but the variation is vast.  At some level this is consistent with the attitudinal model’s claim of broad discretion for justices, but while the attitudinal model claims justices use that discretion to pursue policy goals, we believe justices use their discretion to pursue a broad range of goals that includes their values about law and views about the appropriate role of the Supreme Court.

The most idiosyncratic choice was Kennedy’s.  Few thought the mandate would survive without his vote.  Our prediction (which was based on some judgment calls about the case) was that he would vote to strike with about 50 percent probability.  Coming off a week in which he voted against mandatory life sentences for juvenile offenders it cannot be said that he is reliably liberal or conservative.  He’s just hard to predict.

As for predictions in general, I’m glad that ours worked out, but I wouldn’t want to claim that this was the goal of what we were doing.  The goal of our models was to test theories about Supreme Court decision-making.  The prediction was a good exercise and, as it happened, adding legal elements in this case made us more confident about the Court upholding the law than we would have been from a purely policy perspective.  There is, however, always noise and I definitely wouldn’t want to make a living off of betting on Kennedy’s choices.

The political constraints might be where much of the action was.  As we predicted, Roberts wrote an opinion that “cabins Commerce Clause more than it is now.”  This is a big deal.  If Roberts limited the Commerce Clause as he did and struck the law, the Court’s legitimacy would come under a fierce assault.  The right comparison may not be Dickerson (where the Court upheld precedent that the justices likely did not agree with on policy grounds), but rather Marbury v. Madison, where the Court conceded on a particular policy in the interest of establishing a more general constitutional principle.

Roberts began his opinion by asserting, “We do not consider whether the Act embodies sound policies. That judgment is entrusted to the Nation’s elected leaders. We ask only whether Congress has the power under the Constitution to enact the challenged provisions.”  We concur.  Justices are constrained.

10 Responses to The Law Matters

  1. Political "Scientist" June 28, 2012 at 3:44 pm #

    These chimps sure are getting pretty good at throwing darts!

  2. Dan Nexon June 29, 2012 at 9:22 am #

    Mike et al: would you explain why this is a vindication? You predicted 7-2 or 6-3 to uphold. The justice you had as most likely to uphold was in the 5-4 minority. The majority (4 dissents + Roberts) broke from settled commerce clause jurisprudence. Etc.

  3. idiot June 29, 2012 at 9:59 am #

    Your original “ideological preferences” model was more accurate than your ‘precedent’ model suggests, 5-4 reversing PPACA. That model merely failed to predict one justice vote (Roberts), as opposed to your ‘precedent’ model, which failed to predict two justices’ votes (assumed Kennedy was leaning for PPACA, and that Roberts was leaning against).

    In fact, there is some speculation (mostly caused by the dissent referring to Ginsberg’s argument as a “dissent” and that Ginsberg’s concurring opinion attacked the Chief Justice’s arguments rather than the actual dissent), that the conservative ‘dissent’ was originally a majority opinion, and if that was the case, that meant that John Roberts had originally sided with the conservative opinion and voted to overturn, and for whatever reason, decided to uphold the law on a different technicality. That scenario however does point to the idiosyncratic nature of the court (and its “political” nature) as well. Though most of the time, you can predict what happens, there will be a few exceptions.

    But your own predictions was itself flawed, and I would have hoped at least an attempt at being humble.

  4. LAC SCHOLAR June 29, 2012 at 10:16 am #

    Came here to echo the thoughts of Dan and idiot.

  5. Michael Bailey June 29, 2012 at 11:17 am #

    With the proviso that “who predicted what in a blog post” is less consequential than the big questions about role of policy vs. law vs. institutions, here goes my bragging rights response. I think we did alright – we predicted Roberts would not vote with conservatives b/c, as per our book argument, he isn’t pure policy maximizer. Check. No, he did not go directly with the precedent argument, but I didn’t see many predictions of him going the tax route. Our blog post may have led people on in this respect, but I think in our book we’re very clear that we don’t think precedent (or deference or 1st amendment) are the only non-policy values that matter. These were the ones we could measure and therefore test. Lots of factors, like institutional legitimacy may well matter too, but we couldn’t measure them to our satisfaction.

    For Kennedy our textual prediction was to uphold. Obviously wrong. The model’s prediction was a 54% probability of upholding. If I were really nice to myself I could call that a half-check (or 0.46 of a check).

    The textual Alito prediction was an add-on against the model prediction which had Alito at 80% to overturn (what were we thinking??). That would have been 0.8 of a check, but half off for the textual add-on. So in total 1.86 checks. Not so bad, I think.

    We didn’t predict exactly what happened, but I do think our theory pointed us toward a much different and more accurate view of the process than the straightforward policy-preferences only view.

    Bottom line: I think the Constrained Court approach to the Court did better than the attitudinal model. As for predictions of a single case – we did fine, better than most. Not exactly right by any stretch, but you can see I’m a generous grader and I’m happy to compare our predictions to any you made before the decision was made!

  6. Michael Bailey June 29, 2012 at 11:26 am #

    Plus, show us some love for saying of Roberts that he would “possibly writ(e) an opinion that cabins the Commerce Clause more than it is now” while upholding.

    I agree that it was a close run thing with the dissent appearing to have been the initial majority. I do believe the political science approach is able to put perspective on all cases, not just a single case.

  7. Dan Nexon June 29, 2012 at 5:07 pm #

    Thanks for the clarification.

  8. Lorenzo from Oz June 30, 2012 at 5:46 pm #

    I would note that the 7-2 decision to protect the States against punitive defunding of existing Medicaid funding possibly muddied the prediction somewhat.

  9. Forrest July 1, 2012 at 10:07 am #

    I should also note, that while we did not go out on a limb and post prior to oral arguments on a blog a predicted case outcome for any other case, it is very clear that the last week of the term was not a good week for the attitudinal model. In Arizona, Stolen Valor, and ACA, it appears that the majority of the bench came out and supported a position that runs contrary to what I think most people would have predicted based upon policy preferences alone. What Mike and I argue in our book the CONSTRAINED COURT (buy it and send our kids to college), is that justices do indeed allow legally based decision rules to guide the decisions they make. Although three cases in a row clearly could be an outlier, in our book we present systematic evidence that justices are constrained.

  10. LAC SCHOLAR July 8, 2012 at 10:27 am #

    I still find your claim of victory a little troublesome. The reason you correctly predicted the outcome is because you forecast Kennedy voting to uphold. Based on that, you (and not your model) predicted that Roberts would go along with him in order to keep the opinion. If your model would have predicted Kennedy’ vote accurately, you never would have predicted that Roberts would uphold and thus your forecast would have been incorrect. So, it seems that there was more luck involved than a correct/accurate forecast.