The problem with taking polls at their face value

[BzaDem]

Anyone looking at the latest USA Today/Gallup poll (a 3-day poll, Feb 22-24) would think "Wow, Obama has a huge national lead over Clinton." That's because the result is 51 Obama, 39 Clinton.

But if you dig a little deeper, you will find that the USA Today/Gallup poll doesn't screen for likely voters. In fact, it doesn't even screen for registered voters. It doesn't even ask the people it calls whether they have ever voted before!

This definately matters. To see that, look at the Gallup Daily tracking poll (also a 3-day poll, same Feb 22-24). Note that both polls are associated with Gallup. The latest Gallup tracking poll has Obama at 47, Clinton at 45. Within the margin of error, as opposed to a blowout with the other Gallup poll.

Furthermore, hypothetical GE matchups in February consistently favor the unknown candidate when there are any negatives for the known candidate. This is because voters know very little about unknown candidates' negatives. Kerry was ahead of Bush in Feb by 7 points, and we know what happened. Even as late as May 1988, Dukakis was ahead of Bush by 10 points, and we know what happened. This effect might be especially present in current hypothetical matchups between Obama and McCain, since the media has not covered heavily some of Obama's negatives that swiftboaters will bring up in the fall.


My point certainly isn't that Hillary will win the primary, or McCain will win the general. It is that you must look beyond the surface of polls to have any chance at getting accurate information. Anyone who thinks Obama has this election in the bag vs. McCain based on polls, as another DU poster suggested, needs to take a vacation. That kind of attitude gets us complacent as a party, and contributes to us losing elections. Obama may still very well win the GE, but polls at this point are NOT evidence of that. Hypothetical GE matchus in February are simply not accurate, and even primary polls neet to be closely looked at.


Links:

Kerry ahead of Bush in Feb:
http://www.cnn.com/2004/ALLPOLITICS/02/02/elec04.poll.prez/index.html

Dukakis ahead of Bush in May:
http://query.nytimes.com/gst/fullpage.html?res=940DEEDD1F3EF934A25756C0A96E948260

Analysis of the two recent Gallup polls:
http://www.pollster.com/blogs/re_dueling_gallups.php