Anchorage based pollster Marc Hellenthal released a poll of 403 like Alaska voters interviewed October 15- 21. The poll asked about the Senate race, the House race, and the governor’s race. It was paid for by lobbyist Sam Kito. The polls show Gov. Sean Parnell with a slim lead over unaffiliated and Alaska Democratic Party-endorsed candidate Bill Walker. Rep. Don Young has an 18 point lead over Democratic challenger Forrest Dunbar. What’s most surprising, however, is that it shows GOP Senate candidate Dan Sullivan down 10 points against Democrat Sen. Mark Begich. That number’s greatly at odds with other polls released. However, Begich has been fighting hard for the seat in the last weeks, particularly in rural Alaska. That said, the sample size is small and the time in the field is long. Here are screen-grabs of charts from Hellenthal’s poll. The “general percent,” category is Hellenthal’s prediction if the undecideds were divvied up between the candidates.
CORRECTION: I got all sorts of things wrong about Sam Kito. Initially, I said it was Kito III who paid for the poll. That’s not true. That’s his son. Also, I said that he was a former legislator, which also isn’t true. However, it’s true that his son was a former lobbyist and is now a legislator.
Apparently likely voters are determined by asking people if they’re going to vote. I didn’t note any questions about past voting history, particularly midterm election voting history. I’m interested to know how many people were reached to come up with the 403 likely voters. Mr. Hellenthal? Just how many people did you reach and question who answered that they weren’t going to be voting in the election? Or did every single person you interviewed answer that they were definitely, probably or somewhat likely to vote?
Polls showing such close margins may have some value to campaigns reflecting effectiveness of strategy but when published to the general public do they only have propaganda value? Is the idea to announce that since my candidate is “ahead” (regardless of the margin) undecided voters might want to be on the side of the winner? Why else would you publish a poll with a margin of less than the statistical error allowed by the pollster?
I wonder if bias from “polling fatigue” is destroying the validity of this polling endeavor by use of the telephone. I am impressed with how many times now have I heard people brag about being disturbed by these incessant phone calls then lying to the pollster just to skew the results.
The famous 1936 telephone polling error that predicted Ralph Langdon over Franklin Roosevelt was biased, not because of the responses, but because only the relatively affluent could afford phones and the affluent tended to support Langdon. Isn’t s similar bias to be found in the population that no longer has a traditional “land line” and only uses a Cell Phone. Perhaps a roulette wheel would be just as accurate……….
A poll measures an accumulative record of a person’s thought and emotions up to the point in time when they are polled. A poll is a historical summary, not a futuristic prediction. Sometimes the future is not like the past.
The principle goal of a political campaign is to alter reality. Your candidate may start out way behind. Your goal is to put your candidate way ahead.
Large survey samples have mathematically smaller margins of error, but not that much smaller. A one thousand sixty seven (1,067) sample has a margin of error or + 3.00%, a six hundred (600) sample + 4.00%, a three hundred eighty four (384) sample + 5.00%, and a two hundred sixty seven (267) + 6.00%. Everything else being equal, you gain a little more accuracy but at considerable monetary cost.
Often overlooked in evaluating the validity of a survey is “response rate,” which is the percentage of persons actually surveyed out of those that could have been surveyed. Right now, Alaskans with listed landline telephones have been inundated with surveys and “robo” calls. Survey firms with a large percent of landline completions have very low response rates. You must always assume that the answers from those who respond are just like those who did not respond. When the response rate is low, this assumption is highly suspect.
Surveys done quickly (under three days) sacrifice response rate for timeliness. On any given day, there is a significant percent of those you attempt to survey who are not home or not (for whatever reason) willing to do a survey. It is better, when you can, to stretch a survey over several days to have a higher response rate.
Fortunately, cellular phones have not been over surveyed or “robo” called to death. Seventy nine percent (79.0%) of our completed calls on this survey were on cell phones. We found that 39.0% of respondents could have only been surveyed on their cell phone. They did not have a landline. Also, when calling cell phones, we ask for the zip code where the respondent lives so we can correctly assign them to their proper area of Alaska.
Found a slew of poll numbers.
http://polltracker.talkingpointsmemo.com/contests/ak-senator-14
I would not bet on any of them.
We’ll see. I’ll bet Young is a whole lot closer to 70 than 60 and way over the 53 that Hellenthal is claiming.
@AH Ha: Sometimes he’s right. Sometimes he’s wrong. It is, however, a small sample size and it was in the field for a very long time, in polling terms. I updated the story to reflect that.
This seniliity report is brought to you by Marc Hellenthal and Sam Kito. Even Geritol wouldn’t help. Do either of them know that we had the millenium? The poll isn’t worth looking at. Sorry.
Amanda, Do you have any track record for this guy’s polls? has he been accurate in the past ? I agree with the prior comment… This is a fairly small sample size and it disagrees fairly significantly with ALL other polls I’ve seen recently….Even PPP has Sullivan in the lead.
The PPP poll from September didn’t even have Don anywhere close to 70%.
Against Dunbar who no one has heard of and has spent zero money? Like it or not, Young will be close to 70%.
It’s the “Democratic Party”. I know that you support the Republic Party but you really should get your endings right.
I don’t think anyone thinks Young is ever going to pull those numbers again.
It’s Flashback Friday! Back to another poll that Hellenthal got completely wrong! http://amandacoyne.com/politics/new-poll-shows-a-tightening-in-senate-race-oil-tax-repeal-winning/
This is ridiculous. The sample size too small and the duration of the poll too long. Besides, its been on the street a couple of days. Furthermore, Sam Kito was never a legiislator. Usually your judgement and information is better than this.
It did. I changed it. Haven’t written a correction yet because I’m like doing 30 things at once. Arggh.
Parnell is 7 points better than Sullivan? Find that VERY hard to believe. Young with 55%? He had 64% against Cissna, 69% against Crawford.
Thanks. I thought the original said III (the son).
It was the father, not the III.
I think you mean former lobbyist turned legislator Sam Kito III