Contributors
John C. Eastman- Contributor

Dr. Eastman is a Professor of Law at Chapman University School of Law and Director of the Claremont Institute Center for Constitutional Jurisprudence. The views expressed here are his own.

Judicial Mischief
An election almost halted because of 3 judges' statistical ignorance
[John C. Eastman] 9/30/03

Mark Twain noted in his autobiography that there are three kinds of lies: lies, damned lies and statistics. The three-judge panel of the 9th Circuit Court of Appeals, which on Sept.15 put the gubernatorial recall election on ice (temporarily), scored a trifecta on Twain's "lie" scale. Parts of its opinion were merely false, parts were patently false and parts were based on statistical manipulations.

The panel decision had a shelf life of a mere eight days, of course, vacated by a unanimous ruling from an 11-judge en banc panel of the 9th Circuit that included seven judges appointed by President Clinton. One might be tempted, therefore, to forget the mischievous opinion, but that temptation should be resisted, for several reasons.

First, the opponents of the recall have claimed they would litigate the recall election before, during and even after the election. With a dozen lawsuits filed up and down the state within days of the recall being certified ("before") and extraordinary appeals being taken even while absentee voting was underway ("during"), there is every reason to believe they will continue the litigation even "after" the election (assuming, of course, that the results are not to their liking).

Second, the en banc panel did not actually reject the three-judge panel's equal protection holding. Instead, it specifically acknowledged that the equal protection argument accepted by the three-judge panel "is one over which reasonable jurists may differ." It simply held that, at the preliminary injunction stage, and in light of the Supreme Court's admonition that Bush vs. Gore did not necessarily bar local entities from using different voting systems, the ACLU plaintiffs had not "established a clear probability of success on the merits of their equal protection claim."

Moreover, the en banc panel described the ACLU's claims under the Voting Rights Act as even "stronger," explicitly holding that the "plaintiffs have shown a possibility of success on the merits" of those claims, and apparently inviting post-election litigation by stating that the ACLU's allegation of vote denial amounts "at this time" to "merely a speculative possibility – that any such denial will influence the result of the election."

For plaintiffs bent on either stopping, postponing, or invalidating the recall election, there is more than enough room in the en banc opinion to continue the litigation after the election.

Finally, although the three-judge panel decision was vacated by the en banc court, there is no question that it will continue to have sway over any post-election phase of this litigation. After all, if the complaint is amended after the election to allege that the "speculative possibility" of alleged residual vote differentials has become fact, any decision by the district court rejecting the equal protection and/or Voting Rights Act claims will undoubtedly be appealed to the 9th Circuit.

There, per the court's rules, the appeal will most likely be heard by the same three-judge panel that rendered the initial decision. The errors, outright falsehoods and statistical manipulations in that decision therefore remain an important part of this case.

Some of the errors appear to be based on nothing more than ignorance or simple misunderstanding. The three-judge panel noted, for example, that "the sheer number of gubernatorial candidates will make the antiquated voting system far more difficult to use," but election officials have expressed just the opposite concern. The number of candidates exceeds the capacity of the optical scan and other equipment being phased in for the March 2004 election, not the capacity of the "well-tested and reliable" punch-card system.

And the court's claim that "the punch-card voter does not have an opportunity to inspect the ballot for vote accuracy" is likewise erroneous. Any voter can conduct, and is encouraged to conduct, a visual inspection of his punch-card to ensure that chads are completely removed. He is also free to check the card to make sure that the number of the removed chad corresponds to the candidate or position for whom he intended to vote. With only four questions on the special election ballot, this task is much simpler than in the normal election.

More troubling, though, are the errors that are egregious enough to be presumed deliberate. The panel found "almost dispositive" the fact that the former California secretary of state withdrew certification of punch-card voting machines effective March 2004, falsely claiming that "the secretary was required by state law to find that a voting system was 'defective' and 'unacceptable' before issuing an official proclamation of decertification."

What the decertification statute actually requires, in language that the panel reprinted elsewhere in its opinion, is that voting systems should be decertified "if they are defective, obsolete, or otherwise unacceptable." As the panel, again, well knew, the decertification at issue was based on grounds of obsolescence, not of defect.

In a portion of the decertification order also reprinted elsewhere in the panel opinion, then-Secretary of State Bill Jones compared the punch-card systems to typewriters – "they worked well for many years but are now obsolete in the world of the personal computer" – and he expressly warned that a "poorly planned rush to implement a new voting technology ... could easily result in great harm to the most fundamental right of the people, the right to vote."

To find such a certification "almost dispositive" on the question of whether the punch-card systems are so error prone as to amount to an Equal Protection violation is intellectual dishonesty of the highest order.

Which leaves us with the statistics. The ACLU's allegations that the punch-card systems are roughly twice as prone to error is based on a comparative study of residual vote rates in areas using different voting systems. There are several flaws in the allegations and the study on which they rely, all of which undermine the ACLU's claims. First, and most fundamentally, the allegation of "error" is simply unproven, for it is based on studies that simply did not distinguish between intended undervotes, on the one hand, and accidental undervotes (whether due to voter error or machine error), on the other. But in a world such as ours, where some voters simply choose not to cast a vote in races where they do not like any of the candidates and where this "undervote" might well be larger in minority areas when no candidate for the particular office is a member of the voter's racial or ethnic group, the implications of this flaw on Equal Protection claims in the existing studies is huge.

Second, the "twice as prone to error" claim is highly misleading, suggesting, through use of a favorite statistical trick, an order of magnitude in the different error rates that is simply not true. The exact same claim could be made whether punch card machines had a 50 percent error rate (compared to a 25 percent error rate for other systems), or a one-tenth of one percent error rate (compared to a one-twentieth of one percent error rate). When the relevant question is the impact that the differential error rates might have on the outcome of the election, the relevant statistic is the percentage point difference in accuracy rates.

Here, even assuming the ACLU's allegations are true, the difference is between a 98 percent and 99 percent accuracy rate, a difference of one percentage point, and even much of that difference would be eliminated by any recount in close elections.

Moreover, even the one percentage point differential overstates the effect on the election outcome. The panel decision, following the ACLU's argument, repeatedly stated that more than 40,000 votes might be lost. But the differential, assuming the ACLU's own claims, would be 20,000 votes once the allegedly lower error rate elsewhere in the state was taken into account. And the outcome-affecting differential is even lower whatever the margin of victory is in those areas using the punch-card system.

Suppose statewide voters vote "Yes" on the recall, but that in punch-card counties the vote is 60-40 against. The differential error rate would amount to a swing of only 4,000 votes, or one-half of one-tenth of one percent of the roughly eight million votes expected to be cast. Such near perfection in any election system is about all that can be expected.

Which brings me to a final point. What is really being targeted here, in my view, is not any marginal erroneous undercount, but the deliberate undercount.

Democrats have long been able to rely on overwhelming support, often exceeding 90 percent, from African-American voters. In recent years, turnout has started to decline in elections in which an African-American is not on the ballot, so a good portion of the undercount in any given race without an African-American candidate can likely be attributed to voter intent. The new voting machines have software designed to encourage voters to cast a vote even in races for which they did not intend to vote.

If the software succeeds, the alleged "error" rate between the new machine and the punch-card ballots would appear to be even greater, when in fact the new machines are less accurate at recording the actual intent of the voter.

What matters, in such a world, is not voter intent, but what the ACLU and courts like the three-judge panel think the voter intent ought to be. If we ever reach that point, we should simply bypass the voting booth altogether, for the litigation in the courts that will result will be much more of a coronation than a vindication of the rule of law.

This article first appeared in the San Diego Union Tribune on 9/28/03

copyright 2003 by the San Diego Union-Tribune, posted with permission

§