Wednesday, March 4, 2009

Random freaking adults, again

Big headline at MSNBC:
Poll: Obama's rating at all-time high
And Mark Murray's story begins:
After Barack Obama's first six weeks as president, the American public's attitudes about the two political parties couldn't be more different, the latest NBC News/Wall Street Journal poll finds.
Despite the country's struggling economy and vocal opposition to some of his policies, President Obama's favorability rating is at an all-time high. . . .
By comparison, the Republican Party — which resisted Obama's recently passed stimulus plan and has criticized the spending in his budget — finds its favorability at an all-time low. It also receives most of the blame for the current partisanship in Washington and trails the Democrats by nearly 30 percentage points on the question of which party could best lead the nation out of recession.
What Murray doesn't tell the reader until the 15th paragraph is who was polled to get these results:
While the poll — which was conducted of 1,007 adults from Feb. 26 to March 1, and which has an overall margin of error of plus-minus 3.1 percentage points — finds Obama in a strong position after his first six weeks as president, the same isn't true for Republicans.
Random adults! Not "registered voters." Certainly not "likely voters." Just random adults. Look at the poll details (PDF): No effort was expended to determine the voting habits of the respondents. There was no screen at all. Just whatever 1,007 people answered the phone and said they were 18 or old.

This is important. The pollster, Peter Hart, and the folks at NBC/WSJ who commissioned the poll are not so stupid that they don't understand the importance of the distinction between "random adults" and actual voters.

Anyone who knows anything about public opinion polling knows that non-voters are different than voters. Everything we know about the behaviors and attitudes of non-voters shows that these are people disengaged from, and ignorant of, the political process. Compared to voters, non-voters tend to have lower levels of income and education, to be less informed about politics and current events, to score lower on every measure of civic involvement.

And guess what else we know about non-voters? If they could be motivated toward engagement with the political process, they would overwhelmingly vote Democrat and support liberal economic policies. So when news organization commission random-adult polls about political questions, the results will always skew substantially leftward, giving the false impression that Democrats and liberal policies are more popular than they actually are among actual voters.

Nobody knows this more than Democratic Party political operatives, who expend massive resources to register the unregistered, and to get as many as possible of these unengaged and uninformed people to vote. If you go back and look at 2008 exit polls, you find that Obama got 73% of voters with annual incomes under $15,000, and his strongest educational cohort was high-school dropouts (63%). Yet even with a historic level of Democratic get-out-the-vote efforts in 2008, there were still many millions who could not be bothered to go to the polls on election day.

Non-voters are irrelevant to the political process. Taking a random-adult about political questions and turning it into a major news story is therefore a fundamentally dishonest action. Polling random adults is fine, if you're asking about Coke vs. Pepsi, McDonald's vs. Burger King. But to do a random-adult poll about politics -- without even endeavoring to find within the respondents a subsample of registered voters or likely voters -- is not honest journalism.

And the fact that Murray buried in the 15th paragraph the fact that the NBC/WSJ poll was of random adults tells you that he understands, and wishes to conceal, the essential bogusness of the result.

1 comment: