Has anyone bothered to ask, How truly accurate are the polls?  If it is true that Obama's people have registered about 12,000,000 new voters in the last year or so, how -- if at all -- does this affect the validity of any national or state-wide polling results? Presumably, that number is statistically significant.  When setting sample sizes and generating lists of voters to poll, don't pollsters make certain evidence-based assumptions (premised on historical indicators, census data, etc.) regarding the number of people and types of people (e.g. likely voter, registered voter, etc.) in the state or national electorate? If they do not accurately perform that function -- by failing to take into correct account the numbers and locations of new voters -- how does this affect their poll results/conclusions (particularly if you assume 12,000,000 newly registered voters nationally who have no voting history/record and are of various ages, not just 18 year olds)?