How to get politicians to think experimentally


Politicians engage in a variety of local campaign strategies that they think will help them get re-elected, but they don’t really know if the money they spend works or not. Peter John explains how experiments conducted by politicians and researchers working in partnership can be a useful way of finding out what actually works in local campaigning.

Experiments are becoming very common these days as public agencies turn to randomised controlled trials to evaluate public policies.  The Behavioural Insights Team, which was in the Cabinet Office, has blazed a trail by using trials to test for a range of innovations, such as getting income tax paid on time, collecting court fines, getting people into work, and improving education attainment, just to name a few of the recent applications. Despite this growing interest, one group of people doesn’t use experiments very much: the politicians, at least those in the UK.

Politicians want to get re-elected and they engage in a variety of local campaign strategies  with the aim of improving their chances of doing so such as leafleting, e-mailing, door knocking, using social media, and buying space in newspapers. But they don’t really know if the money they spend generates votes in the ballot box.  Politicians have been advised that they need to get information on the type of voters who support them so that they can target messages to them, or find their core or loyal voters to ensure that they turn out to vote.  But looking at election results after spending the money does not tell them whether their campaigning worked or whether they would have won anyway. In contrast, experiments use randomisation to provide a fair comparison between doing nothing and carrying out an intervention.

In the US, the argument for experiments is starting to get some traction. Isenberg’s The Victory Lab shows how both Republicans and Democrats are using experiments to test their messages.  But the methodology has not had so much impact on campaigning in the UK.  The exception is work by myself and Florian Foos, who have worked with local MPs rather than the central party organisations. We have found that it is possible to find individual politicians who understand the approach.  An example is a project that took place during the 2014 European election campaign in the constituency of North East Somerset. This research tested whether, as a result or getting leaflet or a door knock, supporters of the incumbent MP were more likely to turn out to vote than opponents. And we found that Conservative supporters were four percentage points more likely to turn out to vote where supporters from rival parties were just over six per cent less likely. What struck me was that the MP, Jacob Rees-Mogg, was genuinely curious about finding what actually worked with his local campaigning and in understanding how experiments work.  The constituency office was also excellent in running the intervention.

It should be pointed out that as researchers we are non-partisan and work with all parties, and Florian has worked with local Labour parties.  The key is a partnership between us and the MP: the local party can learn lessons about how to canvass and where to spend scarce funds; the researchers get the freedom to publish from the research.  The party works normally but with randomisation generating different messages that the local canvassers need to follow.  We designed the study in partnership with the local party campaign staff.  The data needs to be sent to us to analyse and then we write a short report with the results.

Florian and I are making contact with politicians for the 2016 elections in the London area. Our work shows that we can get politicians to think experimentally and that they find it useful.  We hope to interest more politicians in this line of work. In a future blog I will talk about other uses of experiments in British politics.

This post draws on research published in a paper titled ‘Parties are No Civic Charities: Campaigns, Demobilization and the Changing Composition of the Electorate’. It can be accessed here.

About the author

 Peter John is Professor of Political Science and Public Policy at UCL.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s