Today in an impressively long and graph filled piece at Huffington Post , Mark Blumenthal dissects certain aspects of the Gallup Poll. After a short preamble he begins the piece thusly:
Over the past few years, however, polling junkies have noticed something curious: Gallup’s polls have produced results that appear slightly but consistently more negative to President Obama than those produced by other firms.
The Huffington Post has conducted an independent analysis that confirms the phenomenon and points to a likely explanation. The problem lies in the way that Gallup handles the racial composition of its samples, and the findings highlight significant issues with how polls are developed and conducted today.
First off, Blumenthal indulges in hum-buggery when he references “polling junkies” who supposedly have “noticed something curious”. Just who are those polling junkies and when was the first sighting of same? I haven’t heard a peep about this supposed phenomenon from the various “polling junkies” I follow . While I certainly don’t possess an encyclopedic knowledge of the know universe of “polling junkies”, absent more information I suspect that Mr. Blumenthal is overstating things by a large margin.
Secondly notice the last line of the quote above:
The problem lies in the way that Gallup handles the racial composition of its samples, and the findings highlight significant issues with how polls are developed and conducted today.
Its telling that in Paragraph Four of a roughly sixty paragraph article, Blumenthal breathlessly informs the reader that there is a “problem” with how “Gallup handles the racial composition of its samples”. He never actually explains why the “problem” is a problem. He goes into meticulous detail on the “problem” and why the “problem” might explain why Gallup’s polling results are different than other polls, but he NEVER explains why this “problem” is actually a problem. The reader is left to figure that out on their own and the nearly inescapable conclusion is that the “problem” is only a problem when viewed through the eyes of an Obama partisan.
Jay Cost does a great job here of responding to Blumenthal’s effort, but these two points really tell the tale:
Blumenthal’s criticism of Gallup is directed at its poll of adults, and specifically for having white adults comprise roughly 71 percent of its poll rather than roughly 68 percent. But what is not mentioned is that whites make up a larger share of the electorate than 68 or even 71 percent.
Even in 2008, when Obama generated a 27 percent increase in the black vote, whites still accounted for 74 percent of all voters. This means that, when it comes to measuring the electoral relevance of Obama’s job approval rating, the Gallup poll is still “skewed” toward the Democrats. It just happens to be less “skewed” than other polls.
This is rich. Gallup isn’t twisting the numbers ENOUGH in Obama’s favor so clearly Huffington Post must investigate!
The laughable aspect of this tempest in a liberal teapot is that the proof is in the pudding, and when it comes to political polls “the pudding” is each election day. In 2008 Barack Obama received 52.9% to John McCain’s 45.6% or just under a 7.5% margin. Gallup’s final tracking poll (taken 10-31-08 through 11-2-08) of Likely Voters showed Obama up 55-44 or an 11% margin. Note that Gallup’s “skew” here was a 3.5% OVER estimation of Obama’s margin. You may recall that Gallup actually ran two sets of polls in 2008. One of likely voters and another of registered voters. The final numbers in the Registered Voter tracking poll were Obama 53% McCain 40% or a 13% margin for Obama resulting in a Pro-Obama skew of 5.5%.
Given these facts, and the utter lack of evidence to the contrary, it appears that Blumenthal and Huffington Post are attacking Gallup’s methodology only because they don’t like the pollster’s results and not because the methodology is actually wrong. I think its fair to conclude that liberals are currently moving through the “denial” stage with “anger” next up. Given how angry they are in general the next stage should be a real wingding.