This week it’s the UK General Election, and lots of other countries either have elections coming soon or have recently voted. Lots of pollsters and political scientists have been attempting to predict the outcomes - but how successful will they be?
In this Studies Show election special, Tom and Stuart discuss the various quirks and downsides of opinion polls, and ask how scientific political science really is.
The Studies Show is sponsored by Works in Progress magazine - the best place online to find beautifully-written essays about human progress. How can we learn from the past so that we can solve problems quicker in future? How can we apply this kind of mindset to subjects as diverse as science, medicine, technology, architecture, and infrastructure? Get some great ideas at worksinprogress.co.
Show notes
* Ben Ansell’s book Why Politics Fails
* The polls that got Brexit wrong (but where online polling did better)
* The “Lizardman Constant”
* Stuart’s 2023 i article on whether it’s really true that 25% of British people think COVID was a “hoax”
* Recent-ish paper by Andrew Gelman on Multilevel Regression and Poststratification (MRP)
* Examples of recent MRPs from the UK (and one from the US from 2020)
* The surprising utility of just using “uniform swing”
* The very embarrassing 2010 “psychoticism” mixup between conservatism and liberalism - which even has its own Wikipedia page
* Article on the replication crisis in political science
* 2017 article with examples of where political bias might’ve affected political science
* The Michael LaCour case, where a political scientist fabricated an entire canvassing study and got it published in Science
* Weirdly, even though the study was fake, the principle behind it does seem to be correct
Credits
The Studies Show is produced by Julian Mayers at Yada Yada Productions. We’re grateful to Prof. Ben Ansell for talking to us about polling. Any errors are our own.