Lots of studies use weather as an instrumental variable, but the core assumption of IV implies that an instrument should only affect one variable. I develop a full methodology for how to find and review possible exclusion restriction violations.

Regularization can reduce the problem of extreme inference but inappropriate use can lead to reopening backdoor paths and lead to highly biased estimates.

Do election results change because people switch parties or because different people show up to the polls? Across 104 election pairs, this papers shows that vote switching is the overwhelming driver of electoral change.

MPs often cited opinion in their constituencies to justify their votes on Brexit. Our analysis shows MPs Brexit stance made a minimal difference to incumbent success. A survey experiment on MPs shows that MPs have broadly accurate perceptions of this minimal impact.

The British Election Study team explain the origins of the dramatic events of the 2015 and 2017 UK General Elections and show that they were unprecedentedly volatile compared to all elections since 1964.