Sean Trende joins DoubleLine’s Jeffrey Sherman and Samuel Lau to discuss, among other topics, the use of aggregate polling to forecast election results, the challenges in creating accurate sample sizes in the era of the cellphone and internet, how personal biases impact not only how data models are interpreted but how the models are built and what COVID-19 might mean for November’s election process. Mr. Trende is the Senior Elections Analyst for news website RealClearPolitics. The podcast was recorded June 18, 2020.

Mr. Trende says the aggregation approach used by RealClearPolitics in its tracking of political polling is an advantage over single-sample polling, and the news site’s averaging methods promote transparency. He notes the difficulty of creating accurate polls when “everybody lies about voting and going to church.”

Mr. Sherman raises the issue of how to accurately capture public political sentiment considering the difficulty pollsters must have in getting people to respond to an unknown caller ringing their cellphones. This leads to a discussion among Mssrs. Trende, Sherman and Lau about the challenges of modeling in the worlds of politics and finance with a comparison made to recent questions about unemployment data. Mr. Trende says models using online polling still have a way to go.

Mr. Sherman wonders what the blown predictions that “remain” would carry the day in the Brexit vote means for the reliability of voter polling. Mr. Trende responds that the polls were accurate, and it was the analysts who blew the call, revealing their political biases. They talk about the problem of confirmation bias in modeling and reporting an era in which news consumers can filter out information they find disagreeable. Mr. Trende found this out personally in 2016 when he was savaged online after writing an article about data pointing to a possible Trump victory in November.

Mr. Trende sees Democratic candidate Joe Biden with a big lead today in the national popular vote, but there is still time for President Trump to rebound in the electoral vote by Election Day.