A selection from today’s Morning Jolt . . .

Monday Survey USA released a poll of Kentucky voters showing Democrat Alison Lundergan Grimes up by 2 in the Senate race. Not only is that the first poll to show Grimes ahead since June, every other poll since July had Mitch McConnell ahead by at least 4. The previous most recent poll, by the CBS News/New York Times, had McConnell up by 6.
Wednesday SurveyUSA had Charlie Crist up by 6; two weeks earlier, they had Rick Scott ahead by 1, and two weeks before that they had Scott ahead by 5. Other pollsters have not seen the same dramatic shift in Crist’s favor.
To their credit, the Survey USA release mentions a possible reason for volatility:
The volatility and abnormal pattern of who is home and who is not home will continue next week, when normal SurveyUSA polling operations are confounded by Yom Kippur, the Jewish Day of Atonement. Some schools will close, and some may use the long weekend to travel, and be away from their home telephones, altering for yet another week the pattern of who is and who is not reachable by pollsters.
Their September 26 poll of Oregon showed incumbent Democrat John Kitzhaber up by 12 in the governor’s race and incumbent Democrat Jeff Merkley up 20 in the Senate race, around what the other pollsters have found.
Wednesday, Sept 26, SurveyUSA released a poll that had Republican David Purdue up 1 in Georgia’s Senate race; that’s the smallest lead any pollster has found in the past six polls.
They found Jason Carter, the Democrat, up 1 in the governor’s race. That was the only poll in the last five to have Carter ahead.
Again, this doesn’t mean that they’re “skewed” or “biased” or trying to depress Republican turnout or trying to lull Democrats into a false sense of security.
At first glance, I would wonder if they were projecting or expecting a higher turnout — with an electorate closer to the makeup of presidential years — than other pollsters.

How do we know if a poll is an “outlier”? Wait for more polls, see what they show . . .