Discussion: (1 comment)
Comments are closed.
Pollsters around the globe are under more scrutiny than ever before. While it will take time to investigate the polls’ strengths and weaknesses in the recent British general election, we now have an excellent summary of their performance in the US election. Last month, the American Association for Public Opinion Research released a very careful look at the polls’ performance in 2016 here. The report, prepared by a committee of well-respected US scholars and polling practitioners, is candid and exhaustive. This is important given the almost universally negative popular verdict on the polls after the election. We need to preserve one of the best tools we have to understand the public, and frank and honest assessments of its weaknesses and strengths are essential.
It is worth spending a minute on what the researchers did to give a sense of the scope of their work. The committee began its work in the spring of 2016, intent on examining the accuracy of the 2016 primary and general election pre-election polls and the different methodologies pollsters used. The committee reached out to 46 different polling organizations, half of whom responded to their requests for data and methodological information. Several pollsters provided microdata for analysis. The committee itself examined design features. In addition to this exhaustive analysis, they looked at the performance of three polling aggregators, 538, HuffPost Pollster, and RealClearPolitics, each of which commanded enormous attention during the election. The researchers were as meticulous as they were comprehensive. Here are their key findings about the polls’ and aggregators’ performance:
What went wrong? Why were some polls – and the expectations many people had based on them – so off the mark? The committee considers a number of different possibilities for the underestimation of Trump’s support. They affirmed some of commonly offered explanations, but rejected others.
This summary hits a few of the high notes from an exhaustive and lengthy report. It should be welcomed as a serious effort to understand what happened in 2016 in the US.
Comments are closed.
1789 Massachusetts Avenue, NW, Washington, DC 20036
© 2017 American Enterprise Institute