Tuesday, May 26, 2009

More On Virginia Polls

Last week we criticized the Washington Post for not reporting on polls of the Virginia Democratic gubernatorial primary.

The Post has since endorsed Creigh Deeds in the race, but has yet to report on a number of recent polls taken in the race.

Mark Blumenthal, the force behind the popular Pollster.com website, comments this week in the National Journal on the Post's decision to exclude the polls.

Like the Curmudgeon, Blumenthal favors reporting on the polls, with appropriate caveats to educate readers.

Blumenthal also points out--as did the Post--that polling in a primary where turnout is quite unpredictable, but likely small, is perilous under the best of circumstances. We agree, but that's where the power of multiple polls comes into play.

If in the Virginia race we had one poll saying Moran was ahead, one with Deeds ahead and one with McAuliffe ahead, all with somewhat different methodologies, it would be pretty easy to say the polls don't tell us anything.

Instead, we now have six straight polls--since mid-April--by four different pollsters with different methodologies, all showing McAuliffe ahead. See the Pollster.com chart for details. That's pretty consistent. We also see consistency of trends in successive polls taken by the same pollsters--if we saw inconsistency there, we'd have to conclude that the samples they were taking were unstable.

In any event, the Post may have valid reasons for not reporting on each individual poll, but at some point the political analysts at the Post owe their readers a more in-depth review of the polling results.

McAuliffe may not win the election. That wouldn't necessarily mean the polls are incorrect--they all show large numbers of still undecided voters, and turnout will still be key, so things could change easily. Our bet, however, is that right now McAuliffe is truly leading, with just a couple weeks to go.

No comments: