By Selwyn Duke

As some of you know, RealClearPolitics.com provides a daily-updated average of a great number of presidential-election polls.  And, at the moment, their methodology has Barack Obama up by 6.9 points.  Now, although what I studied in "Probability and Statistics" class a couple of decades ago is a little fuzzy, it occurs to me that it’s not logical to include polls at extreme ends of the spectrum (the CBS News/NY Times poll has Obama up by 14), as it’s possible that many of them either use non-representative samples of the population or are anomalies.  So I have calculated the average taking this factor into consideration.

I omitted the two highest and lowest polls from Real Clear Politics’ list of 13; consequently, 2 polls showing the gap at 3 points, 1 showing it at 9 and 1 showing it at 14 were taken out of the equation (there are 2 polls with a 9-point spread, so 1 remains).  Here is the site’s list of polls with the excluded ones in red:

Using this methodology, which I think is superior, Obama’s lead shrinks a bit to 5.8 points.  It’s still not within any reasonable margin of error, but I think it just might be a more accurate characterization of the state of the presidential race.