Much is made of the number of undecideds this time around and what I am asking is, for historic comparisons to past election, which sets of polls should be looked at?
Current 2-ways are averaging 11% not selecting one of the two with the recent ABC/WaPo being 6%.
Current 4-ways are averaging 18% not selecting one of the two, and 6% not selected any of the four, with the recent ABC/WaPo being 13 and 2%.
Early September 2012 had 5 to 7% not selecting one of the two. Very similar in 2008. ABC/WaPo in 2008 same time had 7%. (And only 3% in 2012.)
In any case there are more undecideds in the aggregates now than in either of those two cycles but the choice of poll sets informs how much more.
Apples to apples it seems like it’s only a few percent more to me.