As for the use of criminal statutes: We did these experiments with students in their first week of law school and with students at the end of their first year. The latter group might be aware of the rule of lenity that Professor Eskridge mentions, but the first group would not be. But there was no significant difference between the responses from those two groups. This also suggests that law school doesn't do much to erase the cognitive tendencies we think best explain the results. We also used a couple of questions based on civil rather than criminal statutes, with the same results shown here.
Now maybe more years of living and practicing law do have that effect, as Professor Eskridge suggests. Or maybe they don't. We can't prove anything about judicial behavior by studying what non-judges do; and we don't claim to have done so. But we think our results are suggestive, and might help explain a discrepancy familiar to all students of judicial decisionmaking. On the one hand, there have been studies of judicial behavior from a distance that suggest judges probably give effect to their policy preferences when they interpret ambiguous statutes; and Professor Eskridge thinks this, too. Yet when you talk to judges you often find that they don't experience themselves as doing any such thing, and some of them get mad when an academic suggests otherwise.
Our experiments suggest -- in a stylized but plausible way, with a non-judicial population -- some mechanisms by which this discrepancy can be produced. Respondents in our new study were told not to let their policy preferences infect their judgments about meaning. It happened anyway. The subjective experience of separating one's judgment about meaning from one's preferences is not a reliable indicator that the separation is actually happening. We doubt (and the empirical studies of judicial behavior do not suggest) that this changes all that much when a lawyer gets older and becomes a judge. It would be like imagining that judges are better than the rest of us at avoiding the endowment effect, or self-serving bias, or other stubborn sources of cognitive trouble. Maybe they are, but we doubt it.
Professor Eskridge questions whether our “external” question would be as helpful to judges as it is to our respondents. But he describes the external question as just a way of wording the inquiry, which makes it natural to suggest (as he does) that judges are too “savvy” to be affected by this. But it isn’t just a matter of wording; the external question really is a different inquiry than simply asking people how clear the statute seems to them. Sometimes judicial arguments about ambiguity are arguments partly about whether the external question (as we style it) is the right question ¬ as in the Smith case Anup mentioned (the one where the defendant was charged with “using” a gun because he tried to trade the gun for a bag of cocaine).
Another thing to remember is that judges aren’t the only ones who interpret statutes.
Professor Eskridge cites a fascinating article by Brudney & Ditslear (89 Judicature 220 (2005-2006)) that examines the relationship between use of legislative history and the ideological tilt of opinions in the Supreme Court. That article does not, however, demonstrate that use of legislative history mitigates the effect of policy preferences on judgments. What it reports is that legislative history is used at roughly the same rate in opinions with liberal outcomes and opinions with conservative outcomes. That result is also consistent with the prediction that liberals can use legislative history to reach liberal outcomes and conservatives can use it to reach conservative outcomes. Fortunately the data set that Brudney and Ditslear have gathered may be able to test Professor Eskridge's claim. We recommend a regression of whether an opinion is liberal on the interaction between use of legislative history and whether the author is a liberal. If the interaction has the opposite as the main effect on the author being liberal, it means it mitigates the effect of policy preference on judgments.
Finally, in comparing different methodologies, both Professors Eskridge and Elhauge note that our findings suggest that textualism appears not to mitigate the effect of policy preferences. This is correct. But to provide a complete picture, it should be admitted that our data do not suggest that textualism is any worse than purposivisim or intentionalism, which are captured by the our drafters' intent question in the second set of surveys. Both the textual cue and the drafters' intent cue allow policy preferences to affect judgments at similar rates.