Wednesday, April 25, 2012

Poll Watch Smarts


Stuart Rothenberg makes a good point in a recent post that the results from a single poll don’t tell us very much about a presidential race—or any topic for that matter.  So, don’t’ get too excited if the media announces some dramatic new finding in a poll that Obama is suddenly up dramatically and Romney is down or Romney is up and Obama is down.  Keep in mind two things about polls.  (1) Pay attention to trends over several weeks, not the results of a single poll, and (2) Be suspicious of any poll with results that differ more than 6 or 7 points from the results in other polls.  Chances are there is something about that poll—the way the questions were asked, the order of questions, the way the sample was put together, the sample size, etc.—that makes that poll unreliable. 

Additionally, remember that U.S. presidents are NOT elected by popular vote but rather by votes in the Electoral College and the popular vote is usually much closer than the electoral vote.  For example, in 2008 Obama won 68% of the electoral votes but only 53% of the popular vote.  In 1996, Clinton won 70% of the electoral votes but only 49% of the popular vote.  Interestingly, G.W. Bush won the 2000 election with a bare 50% of the electoral vote but actually lost to Gore in the popular vote by about 500,000 votes.  The point is that head-to-head national polls comparing the two top contenders don’t tell you much about the margin of victory of the winning candidate—and sometimes even whether the candidate who is ahead in the national polls will actually win. 

No comments: