Today's important headline comes from The Hill, Polling Could be Missing Reality, Again Mark Penn.
Here's my take.
The actual work to get it right is complicated but the actual way of going about it really isn't. First, to address the 2016 stuff again. The polls were right. But you've got to know how to use them. Listeners in 2016 know that I was able to accurately project Clinton's win with the popular vote but Trump's win in the Electoral College. The basis for getting it right were the same polls that the rest of the media-choked on.
Here's what it takes to get it right. First of all, you've got to figure out what type of cycle an election resembles. This means looking at voter trends and if they are extremely predictable over time. For example, a midterm election with a Republican President equals a 92% chance of Democrats gaining seats. Second, you've got to figure out how undecided voters break traditionally in a similar cycle. Last but not least, you've got to account for how many people are saying they're voting for third party candidates in a given election. Accounting for how they really break in a similar election because more people always say they'll vote for a third-party candidate than actually do.
Going more towards the point of the story. The writer is right. If you look at contextual questions for the info that you want to find and use to fit an agenda you're screwed. It doesn't really mean anything in elections. Part of the reason pundits and media outlets blew it so badly was that kind of stuff. Remember, when Trump was the most disliked Presidential candidate ever on Election Day? How much did that matter in the polls? Millions of Americans who said they didn't like him voted for him. That's all that mattered. Same deal here. In fact, anyone who's talking about Trump's approval rating and talking about midterm election outcomes is already screwed. There's no connection between the two.
Unless it's the Great Depression, a botched impeachment effort or 9/11 the other party wins and it didn't matter how popular the president is. So, you can start there when trying to figure out if the folks you're hearing from have a clue, to begin with.
Photo by: Imgur