Monday, 14 July 2014

Improving Voter Turnout

I don't have a strong view on police and crime commissioners. But looking at a recent article on the West Midlands by election I was intrigued to notice that "the Home Office has chosen to spend an extra £1 million on a pilot programme sending publicity booklets to every household in the region in an attempt to drum up interest in one of the coalition’s flagship policing policies."

Wearing my 'what works hat' I am spending a lot of time thinking about how to evaluate different kinds of policy so I wondered what the Home Office was trying to achieve with these booklets. Jim Waterson (who authored the original article) kindly sent me a link to the legislation that explains all:

"This trial will allow the Government to evaluate whether the delivery of election booklets to residential premises significantly raises voter awareness about the candidates standing in a PCC election."

This sounds potentially interesting - after all we'd like to know whether providing information to households improves awareness, perhaps even increasing their chances of voting. In order to figure that out we'd need to provide information to some households and not others and follow up to see whether households that got information had better awareness, voted more, etc. A second best alternative might be to have some areas receive leaflets others not. We would, of course, need to randomise this because selecting specific households or areas might give some candidates an advantage over others. However, with a big enough election randomising the provision of information shouldn't favour one candidate over another, but should increase turnout in the group getting leaflets relative to those that don't (if providing information helps).

Unfortunately, it appears that the 'pilot' in the West Midlands by-election won't allow us to do any of this because the leaflets are going to all residential addresses. Without a comparison group it will be impossible to figure out whether providing information has an effect on voter awareness or increases turnout. Even before and after comparisons for a sample of households won't tell us anything much given that coverage of the election should increase voter awareness over time.

I have no idea whether the electoral commission would allow randomisation of election information (there's a case that they should). Either way, in the absence of a suitable comparison group, it's difficult to see how the £1m pilot in the West Midlands can tell us anything useful.

No comments: