Public Policy Polling for Daily Kos (10/9-10, likely voters, no trendlines):
Scott McAdams (D): 26
Joe Miller (R): 35
Someone else: 35
Undecided: 5
(MoE: ±2.4%)
In looking at PPP’s poll released yesterday (on behalf of our Orange Overlords), let’s walk step-by-step through what they’re doing in this basically impossible-to-poll race. Their approach is fairly similar to what Ivan Moore did a few weeks ago: first run a heat without Lisa Murkowski. (The difference here is that PPP specifically ask “someone else,” while Moore only seemed to let people volunteer that on the first round.) With this approach, it’s a tie between Miller and “someone else.”
PPP then asked the “someone else” voters who they were planning to vote for: indie Tim Carter, indie Ted Gianoutsos, Libertarian Frederick Haase, write in Lisa Murkowski, or write in someone else? Murkowski won that round with 95%, with 2 going to Haase, 1 to Gianoutsos, 0 to Carter, and 2 to write in someone else. That was then imputed to find these results:
Scott McAdams (D): 26
Joe Miller (R): 35
Lisa Murkowski (WI): 33
Undecided: 4
(MoE: ±2.4%)
Miller leads narrowly, despite his atrocious 35/58 favorables among the general electorate – the worst PPP has found for a Republican senate candidate this year, including Christine O’Donnell. (Contrast that with 48/46 for Murkowski and 44/26 for McAdams, probably thanks to his cheerful ads.)
This contrasts a bit with Moore’s approach, which was to remind voters of Murkowski’s presence, ask all voters “stay the same” or “write in Murkowski,” and impute based on that, which seemed to yield a more pro-Murkowski result (44-36 in her favor). I don’t know which approach is the most optimal (maybe PPP’s, since it doesn’t prompt everyone that Murkowski’s in the race, only nudges them in that direction with “someone else” and the follow up question, which may more accurately reflect voting-booth dynamics where there’s no reminder about Murkowski), but at any rate, they’re both better than other polls which just list Murkowski with everyone else.
Ethan Berkowitz (D): 42
Sean Parnell (R-inc): 51
Undecided: 7Harry Crawford (D): 36
Don Young (R-inc): 58
Undecided: 6
(MoE: ±2.4%)
PPP also looks at the much-less-complicated Governor and House races, and, as one would expect, finds the incumbent Republicans in command here, although Ethan Berkowitz still looks surprisingly lively in the gubernatorial race.
I know which approach is more accurate. Will Murkowski’s name be on the ballot? No. Which approach more accurately represents the experience the voter will have at the polls. PPP’s.
If there is one thing I learned from college psychology classes, it is that recalling something is a much more difficult type of memory to form than is recognizing something. If anything, even PPP’s approach inflates Murkowski’s numbers because they give the voter Murkowski’s name from a list in the end.
My ideal poll is: “Who do you plan to vote for, Joe Miller, Scott McAdams or someone else? If someone else then who? ” Do not let the voter answer before both questions are asked. That approach most accurately reflects the experience in the ballot box, and explains better than spelling errors or general voter idiocy the reason why write-ins always underperform the polls.
Worse favorables than O’Donnel? That’s pretty bad.