Behind the Fox News Polls
Yesterday, President Trump tweeted out against a recent poll released by Fox News which shows him trailing in battleground states against his 2020 presidential contender Joe Biden. Fortunately, on 4th August 2020 we had the pleasure of hosting Chris Anderson and Matthew Shelter of Beacon Research – the firm that produces the Fox News polls.
Amongst the many things we discussed related to US politics with Chris and Matthew, we also quizzed them on their Fox News polling methodology. As the pair explained in this brief video clip, public political polls (like Fox News) undergo a far more rigorous question selection process than their private clients would ordinarily demand.
Chris Anderson and Matthew Shelter of Beacon Research discuss the Fox News polling methodology.
Summary of Fox News Pollster Roundtable on 4th August 2020
Today we had the pleasure of hosting a private virtual roundtable discussion with Chris Anderson and Matthew Shelter of Beacon Research. Beacon Research are currently assigned an A rating by Nate Silver FiveThirtyEight which puts them in an elite class of US pollsters.
Highlights of the discussion will shortly be released for our OnePass members. Here are the main takeaways from our discussion:
- Trump probability of winning the election is currently 40% (+/- 5%). This is broadly inline with other pollsters.
- GOP Senate majority stands at ~25% probability – a level we are unlikely to drop below from here.
- New voter registrations are down significantly since 2018 (midterms) which usually weighs overwhelming towards the young and Democrats.
- Typically 2-3% of mail-in ballots are disqualified due to voter errors. This disproportionately hurts Biden since Democrats represent the vast majority if this class of voters.
- Staging a live Presidential debate will more likely favor Biden. The Trump campaign has overplayed the narrative of an incoherent elderly adversary that can't talk.
- Karel Bass of California is a good outside bet for Biden's VP. Criminal justice reform is an increasingly important consideration in voter surveys.
- Elizabeth Warren less likely a choice for VP than others predict as it would play into the hands of the Trump Campaign plan to attack her.
- Hard to find a hidden Trump vote and there is more evidence suggest one for Biden.
- Many precautionary steps have been taken in 2020 by pollsters to ensure the mistakes of 2016 are not repeated.
- Election models such as the Economist's which use a number of key economic indicators to predict the outcome of US Presidential Elections are less useful in an era of polarized politics – when candidates are both capped and floored in their approval ratings.
- Contesting the outcome of the elections is a rabbithole discussion… perhaps for another occasion.
Our Summary
Karel Bass as an outside chance for Biden's VP came as a surprise to us. Similarly, the tactical considerations for not opting for Elizabeth Warren despite her strong credentials for the VP role. Other views expressed were broadly inline with consensus. Chris and Matthew gave careful consideration in their responses backed by rigorous (proprietary) data analysis. One would struggle to (scientifically) go against their analysis and polling data. The pair each have over 30 years expertize in election polling.
A broader observation on election polls (in general) we made is continued emphasis on identifying undecided voters, and the motivators for their voting preferences. Less is the focus in understanding the drivers of voter turnout. This creates large sources of uncertainty in predicting the final outcome of elections when there are systemic age/demographic biases.
