This is why we don't, and shouldn't, use polls for decisionmaking OVER election results.
California voters would change their February primary vote for Hillary Clinton to a vote for Barack Obama if the vote were held again, according to an exclusive poll commissioned by CBS 5.While voters in the California Democratic Presidential Primary backed Clinton by a 10-point margin, a new SurveyUSA poll shows that if given the chance to vote again, Californians would choose Barack Obama by a 6-point margin, 49%-43%.
The poll was conducted on May 7 and 8 and has a margin of error of 4%.
One of the arguments that I find rather unconvincing is that superdelegates should somehow weigh an unofficial poll as highly as an official election/caucus. What the hell is that all about? Do we throw out the votes for any election because of a contrasting poll? Regardless of how supposedly scientific a poll may be, why is it even considered to be on a par with an official election?
And yet, the argument about electability is just that. Polls saying that one candidate does stronger than another candidate with certain segments of the population. Polls saying that one candidate does better than another against McCain.
Polls are unofficial. Elections and caucuses are official. They do not carry anything close to the same weight.