Monday, May 25, 2009

Predicting Appellate Results from Quantity (Not Quality) of Questions

The New York times today here has a very interesting article (here) addressing the thesis that: “The bottom line, as simple as it sounds, * * * is that the party that gets the most questions is likely to lose.” It is a short article and I recommend the reader link to it and read it in its entirety.

I decided to make a few brief comments based on my appellate experience while with the DOJ Tax Division Appellate Section in the early 1970s. My experience was before 3 judge panels that, in oral argument, functioned similarly to the way the Supreme Court works in oral argument. Hence, I observed the dynamics of oral argument and, even in some cases, understood them and applied them to my client's perceived advantage. My brief comments are:

1. To state the obvious, the quantity of questioning is not a perfect predictor (hence "likely" in an important qualifier). But quantity is more than just slightly statiscally significant (86% is quite good). Most of us would like odds that good in placing any type of bet (or just living life). See Leonard Mlodinow, The Drunkard's Walk: How Randomness Rules Our Lives (2008) and Mlodinow's recent comment on the New Times Happy Days Blog.

2. The article suggests that justices often ask questions for which they already know the answer and are just using the question and answer to convince one or more other justices. Often, it does not matter whether the responsive answer is consistent with the questioning justice's perception of the right result. Even wrong answers can be an important teaching tool, as many of us learned as we participated in the Socratic method of teaching in law school.

3. While serving as an appellate attorney, I was aware of the dynamic in a general sense. By way of background, judges who read the briefs or have been briefed by their clerks, in most cases, have an idea of what they will hold before oral argument. Their minds can be changed (and I experienced one known such significant instance), but usually they go out as they came in. This comment is directed to the odds of changing a panel's mind, and is a different, but related, issue to predictability from the quantity of questions asked.

4. My personal experience is that, across the board, counsel for the party appealing (appellant) usually get more questions than counsel for the party not appealing (appellee). Keep in mind that the party appealing has to change the status quo and given the dynamics of systemic deference to the result below (whether to the jury or to the judge) (perhaps, heaven forbid, a form of inertia), I think that the party appealing is statistically more likely to lose, wholly independent of how many questions the panel asks. (This observation is consistent with the following observation in the article that, "If the two sides receive the same number of questions, the likelihood of reversal is 64 percent, which is in line with the usual probabilities; the court reverses more often than it affirms.") But the judge's perception of need to test whether that party should really lose, perhaps drive the dynamic of a larger quantity of questions.

5. Further dynamics are at issue in tax cases. The Government wins more far more times in taxpayer appeals in tax cases than it loses. In tax cases, taxpayers do not generally exercise the same restraint that the Government does in taking appeals which must be authorized by the Solicitor General; hence, the universe of taxpayer appeals present statistically more ill-advised -- unpersuasive -- appeals than the universe of Government appeals. Focusing on the opposite circumstance -- a Government appeal -- Government appeals are generally more meritorious because, in part, they are preceded by fairly rigorous review before the Solicitor General is persuaded to even authorize the appeal. I don't know the statistic, but I would be stunned if, statistically, the Government did not win significantly more of its appeals than the taxpayers win of their appeals. But, the Government on its appeal faces the same dynamic of trying to change the status quo, and thus the Government is bound to lose a number of its appeals, regardless of the rigorousness of the internal DOJ review process. The dynamic noted previously of deference will inevitably give some tilt to Court of Appeals' results in favor of affirmance rather than reversal.

6. I now provide some anecdotal experience based on the some of the preceding observations. On some taxpayer appeals where I perceived from the dynamics of the opening appellant argument the panel understood the issues and the Government would win, my appellee argument was simply to invite questions if the panel had any but otherwise to present no argument at all when the panel had no questions. (I usually flattered or pandered the panel by saying that I perceived from the panel's questions that they understood the issues in the case (just a euphemistic way of saying I thought the Government would win).) Almost invariably, the panel asked no questions and the Government won. Sometimes I varied this approach if the dynamics of the opening argument were generally favorable but I felt that only one or two points needed clarification. I would make those one or two points (once less than 30 seconds with no panel questions after a 25 minute appellant opening argument) and sit down. I developed the strategy to do this type of limited appellee oral argument after my first case with the Government -- a taxpayer appeal to the Second Circuit. After the taxpayer's counsel used all his time, I got up with my canned argument (that I had rehearsed with a panel from the Appellate Section and went, as canned, for about 15-18 minutes), but realized within three minutes that each member of the panel was (i) not paying a lot of attention to what I said and (ii) indeed was reading something that I quickly surmised or guessed or speculated was related to the next argument. I already knew from the court's questions on the opening argument that the Government would win, so I quickly (but not abruptly) closed my argument out, covering only the major outline. Perhaps 5 minutes out of a total permitted of 30 minutes. I speculate that the panel really appreciated that I did not further occupy their and our time (but I could be wrong if the panel had anticipated having that time to further prepare for the next argument). I don't think this approach would work in the Supreme Court, although I could imagine it working if the opening argument isolated some single limited issue that was concerning the justices and then only addressing that issue (provided of court that the issue for which the Supreme Court accepted certiorari was clearly understood).

7. On Government appeals, I approached it differently. I did not perceive a bias in the quantity of questions asked. Since the Government had the burden to persuade the court to change the status quo, I felt that I wanted questions from the court. I felt that, as attorney for the Government appellant, fewer questions meant loss. Of course, the Government lost some, perhaps most of the Government appeals I argued (I really did not keep count), and in those cases, overall, I am certain that I got more questions than counsel for the appellee. And, as I now recall it, in those successful Government appeals, the taxpayer's counsel got more questions from the panel than I did. So this observation is consistent with the statistic.

8. I offer one more observation in trying to use such statistics at the Supreme Court level to extrapolate conclusions about the dynamics of Courts of Appeals. Appeals to the Courts of Appeals are of right. The Courts of Appeals must hear and decide them. Appeals to the Supreme Court through the certiorari process are not appeals of right. The Supreme Court must affirmatively decide to hear the case and determine which issues it will hear. I perceive that the Courts of Appeals now are much more stringent in determining which cases justify oral argument than they were in the days I was doing appellate work on a daily basis. This can be a process with a dynamic not dissimilar to the Supreme Court's determining which cases it will hear. Presumably, Courts of Appeals are much more likely to hear oral argument in cases which, in the review process for determining which cases will be scheduled for oral argument, are identified as involving important issues (important in their own right or involving some potential conflict with other circuits or even the own circuit) for which oral argument can be useful in understanding the bases and ramifications of whatever action the Court of Appeals ultimately takes. In this sense, the designation for oral argument in the Court of Appeals suggests some statistical correlation with the likelihood for some form of reversal or some type of revision to the result or reasoning below (i.e., if the lower court action were clearly correct, oral argument is not needed), just as the Supreme Court's acceptance of certiorari suggests a more than random statistical possibility of reversal or revision of the result or reasoning below (as noted in the article, "the [Supreme] [C]ourt reverses more often than it affirms."). The Supreme Court's rigorous selection process makes it much more likely that the Court perceive an importance in correcting errors in result or reasoning below than it just pronouncing that the lower courts got it right in result and reasoning. That dynamic in the Courts of Appeals in assigning cases for oral argument is, I suspect, much less pronounced, simply because I don't think that the selection process is nearly so rigorous. But, having said that, there might be some positive correlation that would be shown with empirical analysis.