I believe we discussed this before, my big issue is how the media covers polls and forecasting models. If readers/viewers understood polling error, they could have a higher level of trust of models and polls. Rebuilding that trust would be good, especially for non-election polls.
While this would require a lot more work, I wonder if models could include a few Electoral College maps to illustrate polling error. There could be five different maps: the traditional model, two normal sized polling error maps (errors favoring each candidate), and two larger than normal sized polling error maps. I wonder how people would react if they saw the range of outcomes in a visual manner. If models were showing maps where Trump was winning and where Biden was winning big, would that help people understand polling error? For 2020, I would've wanted people see a map where Trump won all the states he won in 2016, plus NV and a map where Biden won 413 Electoral Votes with a TX win.
I found an article that Dave Wasserman wrote in 2016 very helpful. Going over five scenarios showing five different maps helped me visualize the range of outcomes. This is a possible solution to communicate more effectively.
You’ll Likely Be Reading One Of These 5 Articles The Day After The Election
Maps are a good idea! The issue is that at some point we risk giving readers _too much_ information and overwhelming their ability to process it. So we’re working on striking the right balance for next cycle.
Hi Elliott,
I believe we discussed this before, my big issue is how the media covers polls and forecasting models. If readers/viewers understood polling error, they could have a higher level of trust of models and polls. Rebuilding that trust would be good, especially for non-election polls.
While this would require a lot more work, I wonder if models could include a few Electoral College maps to illustrate polling error. There could be five different maps: the traditional model, two normal sized polling error maps (errors favoring each candidate), and two larger than normal sized polling error maps. I wonder how people would react if they saw the range of outcomes in a visual manner. If models were showing maps where Trump was winning and where Biden was winning big, would that help people understand polling error? For 2020, I would've wanted people see a map where Trump won all the states he won in 2016, plus NV and a map where Biden won 413 Electoral Votes with a TX win.
I found an article that Dave Wasserman wrote in 2016 very helpful. Going over five scenarios showing five different maps helped me visualize the range of outcomes. This is a possible solution to communicate more effectively.
You’ll Likely Be Reading One Of These 5 Articles The Day After The Election
https://fivethirtyeight.com/features/youll-likely-be-reading-one-of-these-5-articles-the-day-after-the-election/
We also need more of these articles.
Trump Is Just A Normal Polling Error Behind Clinton
https://fivethirtyeight.com/features/trump-is-just-a-normal-polling-error-behind-clinton/
Trump Can Still Win, But The Polls Would Have To Be Off By Way More Than In 2016
https://fivethirtyeight.com/features/trump-can-still-win-but-the-polls-would-have-to-be-off-by-way-more-than-in-2016/
I'm looking forward to reading your book!
-Elliot
Maps are a good idea! The issue is that at some point we risk giving readers _too much_ information and overwhelming their ability to process it. So we’re working on striking the right balance for next cycle.
Book, over here, please.