Splitsville

If you want a good example of massaging data to fit your argument, take a look at in today's Washington Post. In an op-ed, pollster and political consultant Mark Penn writes:

According to the American National Election Studies at the University of Michigan, the number of split-ticket voters in the electorate -- meaning people who vote for a Democrat for president and a Republican for Congress, or vice versa -- has gone up 42 percent since 1952. That shows a radical new willingness on the part of Americans to look at individual candidates, not party slates. It is a sign of a thinking electorate, not a partisan one.

I checked the NES. In 1952, 12 percent of voters split their ticket. In 2004, the number went up to about 17 percent--a 42 percent increase as Penn points out. But this obscures the fact that the vast majority of voters, 83 percent in the last election, still vote the party line. More importantly, this chart shows that the percentage of split ticket voters has been declining pretty steadily over the last 30 years. Penn could have just as easily and more accurately written that the percentage of split ticket voters has declined by 42 percent since 1972.



Penn was a consultant to Bill Clinton and has a reputation for trying to push the Democrats to the middle, hence his desire to show that split-ticket voters are an important constituency. That may or may not be good politics, but Penn doesn't help his argument by playing fast and loose with the numbers.

Update: Over at The Plank Michael Crowley points out that Penn is Hillary Clinton's pollster. Is his op-ed a way of under-cutting the argument that Hillary is becoming too centrist?