One has to have sympathy for a pollster. It is hard enough to get a representative sample and accurate polling in a “normal” election season. But in an election season featuring the first African-America, possible strong youth and minority voting, and more mobile phone users than land-lines, the task of accurate polling becomes near impossible.

American pollsters were humiliated this past primary season by first predicting a strong Obama win in the New Hampshire primary (only to have Hillary win by about seven points) and then by consistently underestimating Obama victory margins in several states, predominantly Souther states.

Now some pollsters are reporting a 14-point Obama lead over McCain (NY Times/CBS News poll). But wait. That same day another poll reported a Obama lead of only 3-points, within the margin of error. Why are the polls so divergent from one another? There are several factors for this:

1) The Bradley Effect. Named after African-American California gubernatorial (which means Governor for non-Americans) candidate Bradley, this effect states that an African-America will always do better in polls then in actual votes due to the fact that many White voters will lie to a pollster about voting for an African-American so that they do not appear to be racist. In the end, though, the White voter will cast a vote for the White candidate. Thus pundits always consider the Bradley effect when looking at polls, if an African American is leading by, say, 4 points; when one factors in the Bradley effect the African American could be up by just 1 or even losing. This is what happened to Bradley in California, he was leading in all the polls only to lose on election day: hence the Bradley effect.

Obama could have the same problem: his numbers might be below what is being registered. If Obama’s polling puts in 4 points above McCain, he could be in trouble. But if Obama is 8 points, then the consensus is that he is fine because the Bradley effect is never that big so Obama will win just with a smaller percentage victory. In any case, the polling numbers not could be (and most likely are to a degree) falsely altered due to the Bradley effect.

2) Youth and Minority Voters: Pollsters often downplay the representation of young and minority voters in their polls because these demographics do not vote in significant numbers. A lot of pollsters are keeping this tradition by recording only a small amount of youth and minority voters in polls. But they could be underestimating them this election season. Obama has, of course, energized African Americans like no one before and it is almost a sure thing that they will turn out in record numbers to support him. During the Democratic primary, pollsters underestimated Obama’s victory points in Souther states (which have a large black population) by not properly factoring in African American voters. They could engaging in the same misleading tactics now as well. Second, all the young often do not vote, they do love Obama. And they might vote for him in record numbers as well. If the polls are underestimating the share of the young and black vote this election, Obama could be doing better than he thinks.

3) Mobile phones: An increasing amount of Americans are ditching land-line phones and only maintaining a mobile phone. Polling is done mostly over land-line. That means pollster are ignoring a large segment of the American public: one that is young, technologically ‘down with it’, and city people; in other words: almost pure Obama voters. Most polls do factor in some mobile-phones and these show a significant preference for Obama. So if there is underreporting of mobile phones, as there is in most polls, then once again Obama is mostly likely doing better with his numbers.

Even his the Bradley effect, Obama still has best 2-out-of-3. Come election day the only surprise might be that the polls underestimated Obama’s victory margin not overestimated it. This was often the case in the primaries and Obama must pray that it will be on November 4.