Title Do partisan campaigning activities have “downstream effects” on voter turnout at subsequent elections? Evidence from a UK-based partisan GOTV experiment.
Post date 11/13/2017
C1 Background and Explanation of Rationale

This study seeks to examine whether the mobilisation effects in one GOTV field experiment ‘spill over’ to another election. I conducted a partisan GOTV experiment during the May 2017 local elections in the UK (“election 1”), and found that leaflets and canvass visits boost turnout. In this analysis, I seek to test whether exposure to these treatments had "downstream" effects and also boosted turnout at the general election the following month (“election 2"). In addition, I seek to assess whether this effect differs according to a pre-existing covariate measuring whether voters are registered to vote by post, or in-person.
As far as I am aware, this experiment represents one of the few temporal spillover partisan field experiments conducted in the UK.

C2 What are the hypotheses to be tested?

I found that partisan campaign treatment boosted turnout at election 1 by 3.7 percentage points (ITT, covariate-adjusted). The purpose of this study is to analyse whether – and how much – of these effects ‘spilled over’ to the general election the following month (“election 2”).

I anticipate that the downstream effect at election 2 will be around half of the original effects at election 1. As such, I expect that exposure to partisan campaign activities at election 1 will increase voter turnout at election 2 by around 1.8 percentage points (H1). I also anticipate that the spillover effect of the treatment will be higher among non-postal voters. supporters of that party (H2).

C3 How will these hypotheses be tested? *

I will test my hypotheses using the same estimator that was used to analyse the turnout effects in Election 1. I used a multivariate linear regression of turnout on assigned treatment, and controlled for covariates that were available from the local party voter database (sex, party support (based on previous canvass analysis), ward, postal voter registration, previous turnout, whether the individual lives in a single occupancy household, and age group). The same covariate-adjusted regression model will be run with turnout at election 2 as the dependent variable, rather than turnout at election 1.

C4 Country United Kingdom
C5 Scale (# of Units) 6,525
C6 Was a power analysis conducted prior to data collection? No
C7 Has this research received Insitutional Review Board (IRB) or ethics committee approval? Yes
C8 IRB Number not provided by authors
C9 Date of IRB Approval not provided by authors
C10 Will the intervention be implemented by the researcher or a third party? Researchers
C11 Did any of the research team receive remuneration from the implementing agency for taking part in this research? No
C12 If relevant, is there an advance agreement with the implementation group that all results can be published? Yes
C13 JEL Classification(s) not provided by authors