by RAHEEM KASSAM
Polls are deployed only when they might prove useful — that is, helpful to the powers that be in their question to maintain their position and influence.
— Christopher Hitchens, Harper’s Magazine, 1992
The latest rod by which to beat the President of the United States is a Washington Post/ABC News poll showing Mr. Trump’s approval ratings languishing at 36 per cent.
Americans across the board appear, on first take, to be well out of love with the new President. But the “small print” on the polls reveals something quite disturbing about the methodology of the group responsible.
The poll was performed by AbtAssociates — a swamp dweller-staffed research and policy shop in Cambridge, MA.
AbtAssociates board members include former Bob Dole, Bill Clinton, Barack Obama, Planned Parenthood, World Bank, and Deloitte staffers. In other words, it is a microcosm of the party of Davos.
But that in itself doesn’t rule AbtAssociates out from doing decent work. What might, however, is the underlying philosophy behind political polling, and the methodology they use to conduct it.
MADNESS IN THE METHOD
WaPo/ABC polling obviously predicted the U.S. election incorrectly. They’ll point to the “margin of error” and claim they were aiming to predict the popular vote, of course. But this still led to them presenting Mrs. Clinton as the presumptive winner. Them, and basically everyone else, which is unsurprising, given the way these things work.
The latest headlines — which themselves have seeded hundreds of articles in the press about the matter — are about President Trump’s unpopularity, born out of a poll of just 1,001 people. That’s an average of 20 people per state in the U.S. they hold up as “representative”.
The representative element comes from projecting this data out using pollster voodoo. This is what you’re really paying for when you commission a poll. Anyone can survey 1,001 people, but these guys claim to know the “right” people, and then be able to project their views out onto the whole nation.
Then we go into the data tables. Since this time last year they have over-represented Democrat voters in their studies. For this latest one, 35 per cent of their respondents were Democrats, 23 per cent were Republicans, 35 per cent were Independents (who in turn lean towards voting Democrat), six per cent said ‘Other’, and two per cent had no opinion on the matter.
Interviews, they add, were conducted in English and Spanish, and demographic questions they asked are not included in the data. I asked the Washington Post why, but at the time of publication I had received no response.
Additionally, they didn’t even choose to whom they were speaking. They used a process called Random Digit Dialling, which I wrote about on the run up to the Brexit referendum (another one the pollsters got wrong):
One polling industry insider told Breitbart London that [pollster] ORB uses “random digit dialing” which effectively dials random numbers until it makes a connection with a human being who might answer questions. The methodology is used by Pew’s Research Center, but in tiny sample sizes can lead to a less representative sample that relies more on post-poll weighting.
This method has been under fire for years, but is still used with scant criticism.
Finally, the pollster declares in the data tables: “Interviewers called landlines and cellular phone numbers, first requesting to speak with the youngest adult male or female at home. The final sample included 350 interviews completed on landlines and 651 interviews completed via cellular phones, including 404 interviews with adults in cell phone-only households”.
Young cell phone users made up the dominant part of the poll, and cell phone-only households made up nearly half the interviews. It should not take a political scientist to work out how this, combined with the Dem/Ind bent (70 per cent) of the respondents, makes the poll unrepresentative.
Write ups of the poll include information about approval ratings, the Russia collusion story, healthcare, and the economy. But some of what they found is conveniently left out of the “news” copy or buried.
A greater stress is placed on those who disapprove than those who approve. No surprise given the poll’s weighting. But of those who disapprove, what was their core issue?
Domestic policy? Foreign affairs? Economic theory? No, it was the “Inappropriate way [Trump] talks and acts”. full story