Tuesday, September 21, 2010

Andrew Malcolm on the Sarah Palin puzzle

Our friend Andrew Malcolm at the LA Times' Top of the Ticket seems puzzled by a mixed bag of survey findings. Some polls have indicated that a plurality judge Sarah Palin as unapproved of and unqualified for the presidency, while another poll, the results of which were released Monday, find that a majority say that they Gov. Palin thinks like they do:
In a modern media age of images and sound bites, especially with so many feeling disappointed by another prominent pol, that's a powerful position to hold, even if she doesn't seek another political office.

Such an identification with voters, especially Republicans, independents and women, may help explain Palin's success endorsing primary candidates this year; about three out of four of her picks have won their race, including most recently Christine O'Donnell in Delaware and Joe Miller of Alaska. Very few of President Obama's high-profile endorsements have won, although voters in one poll claimed his endorsement outweighs Palin's.

A new Rasmussen Poll 23 months before the Republican nominating convention for 2012 finds that a majority of Americans (52%) now say their own personal views are closer to the former governor who didn't finish her first term than they are to the former senator who didn't finish his first term. You know, her down-to-earth talk about commonsense conservatism and reining in Washington's runaway spending and arrogance.

Only 40% say their views are closer to Barack Obama's and 48% now see his political views as extreme.

All part of the puzzling and polarizing political package that is Palin. Even her ardent detractors can't silently dismiss her as a nobody might merit. They must vociferously denounce her, which in politics is actually a sign of respect, the louder the better.
This is a puzzle that has a solution. There is a key to putting the right pieces in their proper places in the puzzle. First, you have to understand how and why the pieces were designed.

Comparing the results from different pollsters is like comparing apples, oranges and... rotten apples. There is a disparity in polling results because those who conduct these surveys do not use the same methodology to collect their data and evaluate it. A few pollsters, Rasmussen included, poll likely voters, while many survey registered voters, and some poll all adults. Different polling organizations use a variety of sample sizes from which to draw their conclusions. We've seen samples as small as 400 and as high as 20,000, but a sample of 800 to 1,000 is most common. A poll's margin of error is directly related to the size of the sample. The larger the sample, the smaller sampling error.

Some polling firms, instead of using random samples, design their samples to reflect preconceived notions of what the electorate should look like. Several firms deliberately include considerably more Democrats than Republicans in their samples. They do this, they say, because there are more registered Democrats than there are Registered Republicans. Indeed there are, but how many more are there? Some pollsters are still using relative party registration proportions as they existed years ago, ignoring recent trends who have shown a decrease in the numbers of registered Democrats and increases in the number of registered Republicans. Once the data is collected, some pollsters then "weight" their samples by adjusting the relative proportions of Democrats, Republicans and independents for reasons they claim make their samples more representative of the wider electorate.

All of these ways of massaging data make us recall what dear old Dr. Phil Taylor, our Statistics professor in graduate business school was fond of saying, which is that you can use statistics to prove any point you wish to make, whether it is valid or not. This is why we prefer the methodology used by Scott Rasmussen, who polls only registered voters. Otherwise, his polling methods are completely random, and he doesn't manipulate the data after he's collected it.

Interestingly, we have found that Gov. Palin polls most poorly in Washington Post/ABC surveys. By some strange coincidence, those two media outlets have been among her most severe critics. Go figure...

- JP

No comments:

Post a Comment