Yesterday I reported that internal polls of different candidates in the US Senate race seemed to be saying very different things. Today I can update that, but the discrepancy still remains.
Tom Kiley, pollster for Michael Capuano, and Celinda Lake, pollster for Martha Coakley, both confirmed for me today what I reported yesterday. Kiley says his poll, taken Sunday and Monday of this week, has Coakley around 35%, and Capuano 7 points behind, in the high 20s. Lake says her poll, taken Sunday through Tuesday, has Coakley at 41%, and Capuano at 20%. (Both have Pagliuca roughly around 10%-12% and falling slightly; Khazei around 7-10% and rising slightly; and some 15%-20% undecided.)
So one poll has a 7 point lead, and the other has a 21 point lead. That's an enormous discrepancy, and I haven't been able to determine any differences in their assumptions or methods that might account for it.
Lake also insists that she is seeing absolutely no trending decline in Coakley's support, whereas Kiley says Coakley's support has dropped slowly but steadily. Again, I don't know what might account for the difference.
As for the third candidate's internal polling that I mentioned yesterday, I have not gotten any further confirmation yet. A source with that other campaign had told me that their polling -- like Capuano's -- also shows Coakley's support dropping, and her lead over Capuano down to around 7 points.
To me, the Coakley trend is the biggest mystery. Capuano may or may not be consolidating support and rising; but if Coakley is holding steady at or above 40%, it's hard to see her losing, regardless of how the undecideds break (or what the turnout numbers are, or whether Capuano's GOTV is better). However, if her support is steadily dropping, then we don't know how much further it might drop, and that means anything can happen.