I'm still dumbstruck that we aren't having a bigger discussion in the media about the accuracy, or the lack thereof, by some major polling companies on the Democratic race for President in 2008.

The latest miscue was in Wisconsin this week, where many polls gave Barack Obama only a small lead of around 4-5 points, almost within the margin of error.

OOPS!  Obama wins by 17 points.

That's not small change.  Most of the Wisconsin polls were off by more than 10 points!

We have seen it swing both ways in this year's race.  Before Super Tuesday, the Reuters/CSPAN/Zogby poll had Obama winning California by double digits.  Instead, Hillary Clinton won the Golden State by 8.  It was an over 20-point mistake.

We certainly remember New Hampshire, where the only issue on Primary Day was "how much" Obama would win by.  He lost.  Not only did he lose, but the exit polls had shown that he was going to win, raising the question, are the voters lying to pollsters both on the phone and in person?

South Carolina showed an Obama lead, but nothing like the drubbing of Clinton that ensued on that last Saturday in January.

So it leads to me ask, why should I be at all confident about the nationwide summaries that I see floating around every other day.  Gallup yesterday had a national poll that was a dead heat with Obama and Clinton, and then hours later he is winning a state by 17.

One of the March 4 states gives us a good example, as pollster Scott Rasmussen has a poll that shows Clinton up by 17 points in Texas on Barack Obama.  I will also note that Rasmussen had Obama winning in Wisconsin by only four points.  So what do you believe?