When I turn my computer on, the first web page that comes up is the MSN homepage.  I don’t know that I actually ever chose this site, it’s just what was there.  But it’s got a good mix of news, entertainment, and random stories.  I check it several times a day because it changes and is updated and is a good way to check and see if anything crazy has happened since lunchtime. 

Yesterday, first thing in the morning I fired it up and found this story:

Obama Widens Lead.    http://firstread.msnbc.msn.com/archive/2008/10/22/1578991.aspx

It begins with this line.  “With 13 days to go, Obama has opened up his biggest lead over McCain in the NBC/WSJ poll, 52%-42%.”

Wow!  That is huge.  A 10 point lead is a huge number with only a week and a half to go in this marathon.  It’s hard to imagine how John McCain could possibly close that gap in such a short period of time.

Or is it?

At 3:00, on the same day, on the same website, I checked the updated headlines and saw this:

Candidates Running Nearly Even
http://www.msnbc.msn.com/id/27324419

It found that Obama was at 44% and McCain was at 43% with a margin of error of 3.5% which means Obama could be ahead by 8 or McCain could be ahead by 6.

So, what have we learned here?  I’m going to have to go with “not a hell of a lot.”

Either the electorate is wildly neurotic, with an inability to make up their mind for more than a few seconds at a time or the pollsters are incompetent fools who’s college GPA’s weren’t good enough to get real math jobs.

Honestly, I’d believe either.

But the more likely scenario is that at least some of the polling is wildly inaccurate.  Otherwise we would have to believe that 8% of the population went from supporting Obama to undecided in the matter of, at most, a few days, and, at worst, a few hours.

It all comes down to the sample of people they call.  These things only work, if the “representative sample of people” that they call is really a sample that accurately represents the people.  Each of these polls had a sample size of about 1,000 people.  That may seem like a lot, but it’s really only .0003% of our population.  I don’t know how good your math is (mine sucks) but that’s a really, really tiny number. 

It would be like going into a town of 10,000 people and asking 3 of them who they were going to vote for for Mayor and assuming that their opinions accurately reflect that of the whole community.

Sure, pollsters make some effort to make sure their sample is random and representative and blah blah blah, but come on, clearly someone is doing a lousy job.

It’s like their polling sample was all 3 year olds.  I know with my three year old, you can start off Monday morning with him dancing around the kitchen, excited over the fact that he’s going to have cereal.  And on Tuesday, when you put the same bowl of cereal on the table in front of him, he’s crying and saying “I hate cereal.”

If there were two separate polls of toddlers asking about presidents and cereal preferences, I would find these results completely plausible. 

But as far as polling adults goes, the only poll I would believe nowadays is one that says “84% of Americans put way too much emphasis on polls.” 

Well, at least for today.