Applied Statistics - Mike Kellermann 

 This week, I will be giving the talk at the Applied Statistics Workshop; as they say, turnabout is fair play. The talk is entitled "Estimating Ideal Points in the British House of Commons."  I've blogged a bit about this project  here .  An abstract of the talk appears on the jump: 


 
Estimating the policy preferences of individual legislators is important for many studies of legislative and partisan politics.  Unfortunately, existing ideal point methods do not perform well when applied to legislatures characterized by strong party discipline and oppositional politics, such as the British House of Commons.  This project develops a new approach for estimating the preferences of British legislators, using Early Day Motions as an alternative data source.  Early Day Motions are petitions that allow MPs to express their opinions without being bound by party whips.  Unlike voting data, however, EDMs do not allow legislators to express opposition to a particular policy.  To deal with the differences between voting data and EDMs, I adapt existing Bayesian ideal point models to allow for the possibility (supported in the data) that some Members of Parliament are more likely to sign EDMs than others, regardless of policy content.  The estimates obtained have much greater face validity than previous attempts to estimate ideal points in the House of Commons, and have the usual benefits associated with Bayesian ideal point models, including natural estimates of uncertainty and the ability to calculate auxiliary quantities of interest directly from the posterior distribution.