Discussion about this post

User's avatar
Merrill's avatar

Simon,

I have written and conducted numerous consumer polls throughout my consulting career. I really applaud your explanations about how to interpret any give poll result.

Also, the increased variability using crosstabs is a serious statistical issue journalists should be sensitive to.

There is an additional problem bias within any poll which results from how accurately the participant groups are created. A pollster can always tip the results one way or another by overweighting any particular group of participants. When I looked at the terrible NYTs Sienna Oct poll which has sadly become gospel across the journalist world, it was clearly overweighted with Republicans. Nate Cohn noted this in the polls footnotes and claimed the polling company applied various "smoothing" algorithms to the data in order to rebalance the results..of course, makes the results less valid. Then a week later, the NYTs printed a correction notice explaining that the initial participant groups were not overweighted with Republicans and no algorithms were applied to the final results..

So, why did the Times rush out poll results they knew were questionable at best??

Many of us who truly admire and deeply appreciate the great job Biden has been doing are gravely disappointed in the Times and have sent in many letters to express our disdain for such partisan behavior on the papers part.

We are working on ways to pressure the Times editorial board to get back into reality..if Trump is an existential threat to Democracy, a message the Times had benn trumpeting loud and clear, the Trump is a threat to the existence of the NYTs in equal measure.

Wake up NYT and get on the Biden bandwagon. It's your best chance at saving democracy in America.

Expand full comment
Philip Graham's avatar

Not so long a post . . . when it’s music to my ears!

Expand full comment
24 more comments...

No posts