Those Pesky Snap Polls
Posted Oct 08, 2008 at 11:39 AM by Maurice Berger
Another reason to be wary of post-debate snap polls: they are limited to voters who actually watched the debate. And that's where the trouble begins. According to Nielsen, viewers of the first two debates did not exactly reflect the voting population at large: "Both debates drew audiences made up mostly of white viewers with higher levels of income ($100,000+) and education (4+ years of college). Older viewers (age 55+) made up the largest portion of the TV audiences for both debates (42% - 46%). However, the Biden-Palin V.P. debate (median age: 52) drew a slightly larger portion of younger viewers than the first Obama-McCain debate (median age: 54)." Thus, these flash results--already suspect, since voters are being asked to respond instantaneously to a complicated political event--represent a relatively unrepresentative sample. So even if an instant poll is correctly weighted vis-a-vis party affiliation, it may miss the complexities of race, age and class, enormous factors in the way voters are thinking about this election.
