The Popular Vote Fallacy

Some observers call it shaky support. Stumper calls it shaky reasoning.

Yesterday, a pair of elected officials (read: superdelegates) committed to Hillary Clinton seemed to waver in their backing of the New York senator. First up was New Jersey Gov. Jon Corzine, who in a CNBC interview reserved the right to switch to Barack Obama if Clinton does not win the overall popular vote. "I'm a very aggressive supporter of Senator Clinton," he said, "but I think you need at least a popular vote." Then came Pennsylvania Rep. Jack Murtha, who told the Huffington Post's Sam Stein (a NEWSWEEK alum) that Clinton "has to be ahead in the popular vote to have any chance at all of getting this nomination." Notice a pattern? Bingo: the "popular vote," which the Clinton campaign has eagerly pushed as an alternative to delegates in measuring who's winning and who's losing the Democratic nominating contest. But while most media types treated Corzine and Murtha's mirror-image remarks as evidence of an exit strategy, they failed to acknowledge the fallacious assumption underlying the whole discussion: that there's even a popular vote for Clinton or Obama to win.

Sadly, there's not.

Here's the math. To date, forty of the 46 states or territories to hold primaries or caucuses have released precise, undisputed popular-vote totals. In this count, according to RealClear Politics, Obama leads Clinton by 717,086 votes (13,355,209 to 12,638,123). So far, so good. But what, you ask, about the remaining six states? That's where we get into trouble.

First, there's Florida. Despite warnings from the Democratic National Committee, the Sunshine State scheduled its primary before Feb. 5--and true to its word, the party stripped the state of its delegates. That said, we're not talking about delegates; we're talking about votes. In Florida, where both Obama and Clinton were on the ballot, Clinton won by 294,772 (870,986 to 576,214). It's an open question, of course, whether a primary in which both candidates refrained from campaigning should even count. But let's say, for the sake of argument, that it should--which reduces Obama's popular-vote advantage to 422,314. Unfortunately, this doesn't help us much.

Next up is Florida's fellow gun-jumper, Michigan, where Clinton racked up 328,309 votes. Obama's total? Zero. That's because his name wasn't even listed on the ballot. On Jan. 19, Michiganders had two choices: Clinton or "uncommitted." And while "uncommitted" earned about 45 percent of the vote, it's impossible to determine what portion of that bloc backed Obama and what portion backed John Edwards, whose name was also absent. Ignoring the fact that Clinton herself said Michigan wouldn't "count for anything," this murkiness alone makes an overall popular-vote tally impractical: either you award all of the "uncommitted" votes to Obama, which would be grossly inaccurate; count Clinton's votes and leave Obama at zero, which would undoubtedly disenfranchise hundreds of thousands of Obama supporters; or don't include Michigan at all, which would disenfranchise even more, both pro-Clinton and pro-Obama.

That said, the worst is yet to come. The final four states--Iowa, Nevada, Maine and Washington--all held caucuses. But unlike Florida and Michigan, none of them even kept track of how many people voted for each candidate. (This is standard operating procedure in some caucuses, where delegates are awarded proportionally in thousands of precincts.) Wonks can devise equations to estimate the popular vote all they want, but mixing precise vote totals from other states with caucus approximations--which are, by definition, inaccurate--is mixing apples and oranges. Besides, thousands of voters in Iowa entered the caucuses intending to support Bill Richardson, Joe Biden, Chris Dodd and Dennis Kucinich, but were forced to jump to Obama, Edwards or Clinton once their preferred candidate didn't reach the 15-percent viability threshold; in Nevada, the same thing happened to Edwards supporters. How can you possibly pretend to count people required to resort to their second choices?

The fact is, the Democratic Party has only one mechanism in place for deciding the nomination: delegates. The system is simply not equipped to produce an accurate tally of popular votes. Clinton knows that no matter what happens in the final primaries, she won't catch Obama in the pledged-delegate race. But she also recognizes that Obama won't reach the magic 2,025 delegate majority needed to clinch the nomination by the end of regulation, and that only superdelegates (who are free to decide however they want) can put him or her over the top. So it's them she's appealing to. By becoming "the first candidate in the history of the modern (post-1968) nominating process to suggest that the popular vote total should be considered as important as the number of pledged delegates won (indeed, to suggest that the popular vote total has any relevance at all in the nominating process)," Clinton is hoping that some strange, hybrid form of overall ballot approximation will delegitimize Obama's inevitable delegate victory and spur the supers to declare her the "people's choice"-- however incomplete, imprecise or selective that approximation (necessarily) is. As Michael Barone of U.S. News notes, Clinton could conceivably "win" such a mythical popular vote. Now she just has to convince Murtha and Corzine's fellow superdelegates that it exists.

UPDATE: There's an even deeper inconsistency here: combining vote totals from primary states and caucus states (at least the ones that keep track) effectively underrepresents those caucus states simply because they chose to hold caucuses. I'll let reader "obholmen" make the point for me:

Allow me to use my own state of MN as an example. Minnesota is a caucus state with results of 66-32% in favor of Obama which yielded an excess of caucus voters of 73,115 for Obama.  This figure is included in the media created margin of approximately 700,000 for Obama.  But, because turnout in Caucus states is considerably lower than in primary states (Clinton is right about that), the margin of victory is also held down. Neighboring Wisconsin has a similar sized electorate. If the same number of voters turned out for a Mn primary as in Wisconsin and if Obama had the same 66-32% majority, his voter margin would have been over 340,00 rather than merely 73,115.  Or, if Obama carried a hypothetical Mn primary by the same pct as neighboring Wisconsin (58-41%), then Obama's popular vote margin would have been 190,000 instead of merely 73,115.  Even if Obama's margin of victory would only have been 10%, his popular vote margin would have been 110,000 --- still considerably more than he is credited with when only including caucus margin of victory.

Treating the vote totals in caucus states in the same category as primary vote totals significantly disadvantages the caucus states.  Even though Mn's electorate size is comparable to Wisconsin, there were 5 times as many voters in the Wis primary as in the Mn caucus.  This makes the Wisconsin primary 5 times more important than Mn even though the electorates are essentially equal. Another reason why it is a myth to suggest there is such a thing as a popular vote total.  If the rules are unfair, change them but do so before the primary season not after.  For this season, stick with the rules.

I tend to agree. Whether you think that caucuses are fair or not--and there are cases to be made on both sides--more than a dozen states or territories held them this year with the assumption that their results would be expressed in delegates, not votes. The Democratic Party will probably have to reform its nominating process at some point. But changing the rules in the middle of the game strikes me as unworkable. Then again, as I wrote earlier, the rules also say that the superdelegates are free to decide however they want--including on the basis of a chimerical popular vote.