Weekend open thread: Flawed election polling edition

What’s on your mind this weekend? This is an open thread: all topics welcome.

Happy Mother’s Day to those who enjoy the holiday, and healing thoughts to those for whom today is a difficult reminder of bereavements or a less than ideal mother/child relationship. In past years, Bleeding Heartland has compiled Mother’s Day-related links here, here, here, and here.

My most substantive post about mothering was not related to the holiday: My case against Hanna Rosin’s case against breastfeeding.

Since Thursday I’ve been caught up in news about the May 7 general election in the United Kingdom. While polls predicted a few of the results, such as Scottish National Party gains in Scotland and devastation for the Liberal Democrats nearly everywhere, no one anticipated such a large popular vote lead for the Conservative Party, which gave the Tories enough seats to form a government without any coalition partners. As election day approached, it appears that polling firms were “herding” to avoid releasing a survey that could be an embarrassing outlier. Nate Silver discussed the phenomenon of pollsters “putting a thumb on the scale” after last year’s midterm election.

Damian Lyons Lowe, founder of the British polling firm Survation, admitted here that his company “chickened out” of publishing data from a telephone poll taken the day before the UK election, because “the results seemed so ‘out of line’ with all the polling conducted by ourselves and our peers.” That final Survation poll turned out to be close to predicting the popular vote share for the Tories and Labour.

Facing a similar situation last fall, the Des Moines Register’s pollster Ann Selzer stood by her final numbers for Iowa’s U.S. Senate race. That poll looked like an outlier to me and many others, but Selzer was wise not to chicken out or tweak the numbers to follow the herd.

So far this year, various Iowa polls on the presidential candidates in both parties have largely agreed with one another. Most recently, Quinnipiac found a statistically significant lead for Scott Walker and a “scramble for second place” on the Republican side and a huge gap between Hillary Clinton and the rest of the Democratic field. I’m curious to see whether survey findings from different firms will start to diverge as we get closer to the Iowa caucuses early next year.

 photo logo_zpsfkarz0ku.jpg

About the Author(s)

desmoinesdem

Comments