The Nader counterfactual

A pre-election Gallup survey asked Nader supporters what they would do without him. In the national sample of 193 supporters, 43 percent selected Gore, 21 percent Bush, 21 percent would not vote, and the rest selected another option or gave no answer. That supports a potential net benefit to Gore, but not a transfer of every Nader vote. [1]

The survey was small, national and hypothetical. It cannot supply an exact Florida vote total in an election that never happened. Still, the difference between a large third-party vote and a tiny major-party margin makes it reasonable to investigate a spoiler effect. Florida recorded 97,488 Nader votes, compared with the 537-vote certified Bush lead. [2]

The word “stole” obscures this distinction. A lawful vote for a third-party candidate was not a ballot secretly removed from another candidate. Nor did either major party own those votes. The historical question is how the presence of an additional candidate affected the outcome under plurality voting, not whether those voters committed an irregularity by expressing their preference.

Michael Herron and Jeffrey Lewis studied patterns across roughly three million Florida ballots, including choices farther down the ballot. Their model inferred that at least 40 percent of Nader voters would have chosen Bush rather than Gore if they participated in a two-candidate election. The authors nevertheless concluded that Nader’s presence was consequential because Florida was so close. [1]

That is a statistical counterfactual, not a set of ballots in which voters actually supplied a second presidential choice. It differs from a pre-election poll because it uses other recorded choices and modeling assumptions to estimate preferences. It must also distinguish a conditional preference among people who vote from the separate possibility that some would stay home. [1]

Many different changes can be sufficient to reverse a 537-vote margin. That does not make each one the exclusive cause of the outcome. A plausible Nader effect neither erases the ballot-design and voter-exclusion problems nor proves that a particular recount would have produced Gore. Those are different counterfactuals requiring different evidence.