Lies, damned lies, statistics and partisan polls

Two partisan polls from both sides of the aisle were released this week. Both of the polls included Alaska issues, and the U.S. Senate race. Both are more interesting than credible, but perhaps open a window of light into the race that has so far released few numbers.

The first poll was from the left wing firm, Public Policy Polling, commonly called PPP, and was conducted earlier this month. That firm is no stranger to Alaska polling. Following the gun control debate in Congress, the organization released an improbable poll that said that U.S. Sen. Mark Begich’s vote against gun control resulted in a decline of support among Alaskans.

In this poll, PPP polled 850 registered voters including 442 Republican primary voters. The firm’s methodology has changed some. Included in the robo calls are opt-in internet surveys. But it still can’t manage to get its demographics right. This one, like the one on guns, over sampled both women and Democrats.

The second is from Harper Polling, a right-wing firm, which surveyed 677 likely voters Jan. 20-22. Harper also uses robo calling and also appears to oversample women and Democrats, and although it has a similar margin of error as the PPP poll, it comes up with very different numbers. The left leaning group has Begich on top and the right leaning polling group has two of the GOP contenders beating Begich. Surprised?

Here’s PPP’s numbers:

  • Begich beats former DNR Commissioner Dan Sullivan 41 to 37 percent.
  • Begich beats Lt. Gov. Mead Treadwell 43 percent to 37 percent
  • Begich beats Joe Miller 45 to 25 percent.
  • In a GOP primary, Sullivan leads with 30 percent to Treadwell’s 25 percent and Joe Miller’s 20 percent.

Harper Polling comes up with the following:

  • Treadwell beats Begich 47 to 41 percent.
  • Sullivan beats Begich 47 to 41 percent.
  • Begich beats Miller 51 to 32 percent.

All of which perhaps proves the adage that there are three kinds of lies: lies, damned lies, and statistics.

It’s also worth noting that nobody who’s watching this race believes that the majority of the public at this point can distinguish between Dan Sullivan the Senate candidate and Dan Sullivan, the well-known Anchorage mayor.

Also, local pollster Marc Hellenthal, who knows how to poll this state, has Begich doing pretty well in Alaska, particularly in Anchorage, which had been a problem for him.

Hellenthal wouldn’t release the numbers, but he said that Begich seems to “have solved his Anchorage problem.”

Contact Amanda Coyne at


8 thoughts on “Lies, damned lies, statistics and partisan polls

  1. Samuel Abney

    I’ve always thought that polling was pretty good up here. However, I’ve often wondered if they polled the sample’s likelihood for actually voting as well as candidate preference. My guess is probably not. Back in 2002 Fran Ulmer and F. Murkowski were neck-in-neck in the polls, but on election night Murky trounced Ulmer (I still haven’t gotten over that.). The same has been the case for many statewide elections since.

  2. xxoxxoox

    Enjoyed the data points of the polls. To Willis’ comments – – yes, polls if sampled correctly can and do capture a reading of the public’s opinion at a given moment in time. Polls are snap shot of time, not moving pictures. There have been many accurate polls conducted in Alaska over the years that have accurately predicted elections. If they didn’t work and weren’t reliable, I doubt that polling would have grown to the size of industry it has become.

  3. Lynn Willis

    Absent any intent of bias, are these pollsters able to truly sample the population of Alaska; 700,000 thousand people distributed over an area twice the size of Texas?

    (This is from a Wikipedia article) In 1936 the Literary Digest ,a established and respected national magazine, conducted a presidential election poll predicting Ralph Langdon the Governor of Kansas would overwhelmingly defeat Franklin Roosevelt. The prediction was based on a sample of 2.4 million results which is a significant number. Roosevelt then won 46 states.
    This was during the middle of the great depression. The polling sample was based on subscribers to the Digest, telephone listings and automobile registrations – three things that require disposable income which most people at the time did not have.

  4. Anne

    Polls are only as good as the pollster, their methodology and sample is. PPP ha been criticized for their samploing in Alaska on many occasions and even in the pages of I agree with the comment expressed by another reader that this race will be difficult to poll because of 2 Dan Sullivans in the race. It is also very early in the campaign cycle and few campaigns have even hit the airwaves which will really impact polling data. In simple terms, these polls are almost irrelevant in terms of the results today. I promise that those results will change significantly as the campaign progresses. Polls are only a snapshot of the moment. Here is what I think : AK is a redvoting state that tends tovote for republican conservatives with a sometimes ppulist bent as we’ve seen will Sara Palin and Joe Miller. The underlying demographics will make the senate race very difficult for Mark Begich. It is somewhat of an anomoly that he even won and was fortunate to have been running in a race with such extreme and unique circumstances that he did to win. Begich will have a very difficult race this time unless Joe Miller emerges as the republican nominee. Probably both Sullivan or tredwell can give Begich a race. Most likely Sullivan willby far be the strongest candidate as evidenced with his fundraising success and his fast appeal to republican party regulars. Tredwell has been very slow coming out of the box and hasn’t been able to get any traction. My bet in the general is Begich versus Sullivan with the republican nominee winning by 2 pts. This prediction and a dollar bill might get you a soda depending on where you buy it; nonetheless, here’s my prediction and reasoning behind it. This blog is my favorite source of political news and my favorite overall read every day. Keep it coming.

  5. Erica

    The polls will shift up and down and change several times before the election. For a newcomer, Dan Sullivan appears to have situated himself as the front runner in the GOP primary. He’s done a remarkable job raising campaign funds. Also, it apppears as I watch the race unfold that Sen. Begich wants to run against anyone but Sullivan which tells me he’s the one candidate Begich fears most.

  6. Mormon Voter

    Polling in AK this year is going to be a freaking nightmare given that there are two Dan Sullivan’s in the race. This challenge will impact the integrity of all the polls testing their names.

  7. Doporobo

    Robo polls leave a lot to be desired. Then, if the sample is weighted incorrectly the results tend to be even further skewed. The numbers may or may not be accurate. Comparing the polls and thoinking about them, I feel it is safe to say that it is going to be a real race. Further, as an incumbent Begich should really be concerned with his showing. As an incumbent he appears to be very weak. Will be a fun race to watch.

Comments are closed.