POLITICS MAY 29, 2008
There's a strain of logic in recent presidential campaign
discourse that goes something like this: Though Barack Obama sports a modest
lead over John McCain in national polling, his apparent weakness in key
swing states like Pennsylvania, Ohio, and Florida could lead
to a loss in the electoral college even if he wins the national popular
vote by a wide margin. But his salvation could
lie in picking someone from one of those states, like Pennsylvania governor
Ed Rendell or Ohio governor Ted Strickland, as his running mate.
At first glance, it seems like a compelling argument. On
closer inspection, though, it's fundamentally misguided--on both counts. It's
highly unlikely Obama will win the popular vote while losing the electoral
college--in fact, it's all but impossible unless the popular vote is
exceptionally close, as it was in 2000. But, on the off-chance Obama's trouble
in those states does end up looming large, history gives little reason to
believe that putting Rendell or Strickland on the ticket would do much to help.
At the moment, Electoral College obsession
is once again overtaking the punditocracy, so please forgive me if I'm pointing
out the obvious: The Electoral College very rarely matters, and our current fixation
on it is mostly a product of memories from the Bush–Gore race. Before that year,
only once in American history--1888--had a candidate won a popular-vote
plurality while legitimately losing the presidency in the Electoral College.
(The election of 1876 doesn't
count, and in 1824 the vote went to
the House of Representatives.) In both 1888 and 2000, moreover, the national
popular vote was extremely close--a margin of 0.8 percent and 0.5
Once the national popular-vote margin gets much greater than
that, it quickly becomes prohibitively difficult for a losing candidate to
prevail in the Electoral College. Take, for example, the oft-heard refrain that
a swing of 60,000 votes in Ohio
would have handed John Kerry the election, even though Kerry lost by 2.4
percent of the vote nationally. This is true in a literal sense, but
meaningless from a practical standpoint. For one thing, you can just as easily
play the reverse game: A swing of 6,000 votes in Wisconsin
or 5,000 votes in New Hampshire would have made
George W. Bush the victor regardless of the outcome in Ohio.
More importantly, though, votes don't just spontaneously
shift in one key state. A major insight
from the 2004 campaign, on the part of strategists like Bush's Matthew Dowd, is
that votes are determined less by one's physical location than by factors like
demography and lifestyle choices: A Bush voter in Ohio
looks like a Bush voter in California.
As Bill Bishop argues in his recent book, The Big Sort , as Republicans and Democrats diverge from each other in their living
patterns, they increasingly resemble their partisan compatriots across state
As a result, any event or trend capable of producing a swing
of 60,000 votes in Ohio from Bush to Kerry
would almost surely have had some effect outside of Ohio. If the effect had been distributed proportionally
throughout the country, a swing of 60,000 votes in Ohio would correspond to a swing of around 1.5 million votes nationally--enough to erase Bush's 3-million-vote lead
in the popular vote. Or, in 2000, suppose Al Gore's margin of victory in the
national popular vote had been 1.5 percent, rather than 0.5. That amounts to a
net gain for Gore of more than 1 million votes, and about 60,000 in Florida, if distributed
equally throughout the country. Just a fraction of that figure would have given
him the presidency, recount or no recount.
For this reason, political scientists tend to discount the
likelihood of an Electoral College–popular vote split. "The consensus is
that there's a very narrow band where a split is really even possible--just a
one- or two-percent margin at most," says Daron Shaw, an Electoral College
expert at the University
Granted, at the margin, Obama currently looks stronger in
some swing states (Wisconsin, Colorado, Virginia) than
in others with more electoral votes (Ohio, Florida). But to lose
the Electoral College while winning the popular vote by any significant amount
would take a more contorted distribution of votes than that. "It would
require all of the battleground
states to be very disconnected from the national trends," Shaw says.
"It's just not realistic."
It's always tempting to believe this election will be
different--maybe Obama will prove uniquely able to win huge victories in blue
states and "waste" votes cutting into McCain's margin in red states,
while underperforming in battleground states. But this speculation has been
wrong before. In 2000, for instance, many observers thought
Bush would win the popular vote fairly easily by running up the score in the
South, but would lose the Electoral College.
Needless to say, it didn't turn out that way, and, despite
breathless predictions to the contrary ("The
coming Electoral College crisis"), it probably won't this time. As in
2000, the doomsayers rely on overblown regional stereotypes and underestimate
the degree of nationalization in the electorate. If massive support from
college-educated voters and record black turnout drive up Obama's totals in
non-battlegrounds like Connecticut and Mississippi, there are enough of those voters in Michigan and Florida
to put Obama over the top. Likewise, if Obama continues to struggle among
working-class whites, that will cut into his popular-vote margin in blue
states. Pundits may think Massachusetts is all
Cambridge and Ohio
is all Youngstown,
but it's just not so.
Of course, it's
certainly possible that the election will be so close as to make a split
conceivable. So Obama had better pick a Rendell or a Strickland as V.P., right?
Well, no. Summing up the work of the political scientists who have explored the
question, David W. Romero of the University
of Texas at San Antonio concluded in a 2001 paper that
"vice presidential candidates have
no influence on the voters' choice for president." (Bolding mine.) Events
like the announcement of a running mate or the vice presidential debates can
produce a momentary blip in polls--like the ten-point bounce Kerry got when he
selected John Edwards--but it soon disappears.
There's little evidence vice presidential candidates make a
difference even in their home states. A 1989 analysis by Robert Dudley and
Ronald Rapoport in the American Journal
of Political Science found that, on average, a vice-presidential candidate
improves his ticket's performance in his home state by only a statistically
insignificant 0.3 percent. (Presidential candidates, by contrast, get a sizable
four-percent boost in their home states.) Their result has been borne out in
the years since--remember when Edwards was supposed to put North Carolina in play for Kerry? They lost
by 12 points. What's more, the effect tends to be strongest in small
states--Edmund Muskie pretty clearly put the Democratic ticket over the top in Maine in 1968--but
nonexistent in large states where retail politics count for less. Even the edge
LBJ supposedly gave JFK in Texas
is debatable: It was a heavily Democratic state, and Kennedy fared well in other
Southern states with similar demographics.
These questions are part of a larger debate in political
science: Can the outcomes of presidential campaigns shift significantly as a
result of campaign quirks, or are they determined largely by underlying
economic and political fundamentals? For the most part, the latter view has won
out--and it suggests
that the Democratic nominee is headed for a relatively comfortable win. Of
course, the candidacy of Barack Obama (or Hillary Clinton, for that matter)
makes 2008 the first election that won't have two white male candidates, and therefore something of a historical anomaly. The race could end up being
a 2000-style nail-biter--and, in that case, there's a small possibility
that electoral math and running mates will make a difference. But if things do play
out as they have for decades, a lot of hyperventilating pundits will have egg on their faces.
is a reporter–researcher at The New Republic.