In the run-up to the 2014 European and Council elections the UK Labour Party was contemplating different strategies of how to deal with difficult issues, particularly immigration. In the short term, would it be more effective to ‘move the conversation on’ to issues more favorable? Should it confront the difficult issue head on in line with the preconceived perception of the party? Or should it appeal to the concerns that people were most exercised about? Luckily, the party was ready to collaborate with researchers to test if issue-handling reputations influenced the effectiveness of ground-campaign tactics. We will present the results from the field experiment that we conducted in collaboration with UK Labour at the American Political Science Association Conference on Thursday, 4 September, in San Francisco and at the Elections, Public Opinion and Parties Conference in Cardiff on Saturday, 12 September, in Cardiff.
The questions Labour strategists were asking were related to different assumptions about how political campaigns influence voters. The purpose of an election campaign is to affect voter behavior – either by persuading electors to support a party or by mobilizing supporters to turn out and vote. Existing research indicated that it was potentially difficult to alter an elector’s opinion on a given topic in the course of a short campaign, but that campaigns could considerably influence which issues an elector may consider to be important when casting his or her ballot. Theoretically, it follows that if a campaign managed to increase the relative salience of more favorable issues that support for the party should increase. However, the influence of issue-ownership on campaign effects had never been fully evaluated in a real-world campaign experiment.
Working in collaboration with the Exeter Constituency Labour Party and the Labour Targeting and Analysis Unit, we devised a randomized field experiment to look at this question. Our experiment constitutes the first time issue-ownership theory has been evaluated in a field setting using random assignment to subject different tactics to a fair test. We randomly assigned whether voters in the Exeter Parliamentary Constituency received a Labour leaflet, and whether this leaflet emphasized a valence issue owned by the Labour Party (the NHS) or a valence issue most associated with the Conservative Party (policing). The government had recently been criticized for cutting both nurses and NHS frontline staff, as well as the number of police officers and damaging frontline services due to budget cuts. Both leaflets displayed similar media reports presenting figure on how the government was failing at policing and healthcare.
Our results displayed in Figures 1 and 2 suggest that even if leaflets highlighted obvious government failures on policing and healthcare, they did not induce favorable short-term changes in voters’ attitudes towards which party would better handle these very issues. However, as Figure 1 shows, both leaflets clearly affected which issue voters viewed as the most important when they made their electoral decision.
Our post-intervention telephone survey shows that the NHS became significantly more important to voters residing in streets that were randomly assigned to receive an NHS leaflet than for voters in streets that did not receive any leaflet. Surprisingly, even being sent a crime leaflet increased the salience of the NHS, albeit to a lesser extent than the NHS leaflet did, probably because the Labour brand is widely associated with the NHS, a clear sign of issue ownership.
In contrast, the policing leaflet did not affect the salience of the policing issue, nor did it positively affect issue-handling considerations. If at all, our best guess is that highlighting the issue of police cuts affected Labour’s issue ratings and vote intentions negatively.
These results fit expectations based on issue ownership theory. If you want to triangulate to change voters’ attitudes, do it early. Change does not occur over night, and last minute campaigning on your opponent’s home turf won’t do a party any favors with the electorate, even, and that’s crucial, if your criticism is based on substance.
Moreover, our best guess is that the NHS leaflet increased turnout in the nine electoral wards targeted by the Labour leaflet campaign by 2.9 percentage points compared to the control group and by 3.o percent compared to the crime leaflet (both differences are significantly different from zero with p<.10), mostly encouraging undecided voters to come out to vote, and even appearing to discourage Conservative voters to turn out. These results are in line with experimental research that two of us conducted in the same election with another campaign showing that leaflets can indeed have a differential effect on turnout, increasing turnout among supporters and decreasing turnout among self-identified supporters of rival parties.
The result of this experiment helped to inform Labour’s General Election campaign tactics. As per the document that was leaked to the media, the party on the ground aimed to emphasize favorable issues , and Labour decided to invest heavily in NHS leaflets in target seats. Would the election have turned out differently if Labour had opted to emphasize Conservative failures on the deficit and immigration more strongly in the last days of the short campaign? Although our experiment cannot fully appraise this counterfactual, the results do not appear to point in this direction.