What do the academics say? Contacting voters, especially in person, works

Welcome to another in my occasional series highlighting interesting findings from academic research. This time, it is a look back at the work of David Cutts, Ed Fieldhouse, Justin Fisher, Ron Johnston and Charles Pattie with data from the 2010 British Election Study on the impact of contacting voters.

Our analyses of the 2010 election, using data supplied to us by the individual candidates’ agents, found that the more people canvassing for a party in a constituency – both members and volunteer non-member supporters – the better its performance there. Canvassers having conversations with potential voters apparently wins them over.

Other evidence sustains that. The 2010 British Election Study questioned some 19,000 people at the start of the official campaign in late March-early April. It asked how they voted in 2005, whether any of the parties had contacted them in the preceding months, and how they intended to vote in May. It also asked how the contacts were made: was it by telephone, by a leaflet, by a meeting in the street, by a visit to their doorstep, by email, text, social media or what.

Of that large sample, we look here at the 4,294 who voted Labour in 2005: 2,441 intended voting Labour again, and a further 580 were leaning towards a Labour vote…

Those contacted by Labour during the preceding months in one of five ways (very few were contacted by email, text, or social media) were much more likely to intend voting Labour again than those ignored by the party…

Just getting a leaflet increased the percentage intending to vote Labour again from 48 to 57 per cent; 79 per cent of the small number who got an email were going to remain loyal, compared to 48 per cent of those who didn’t; and there was a 15 percentage points difference in loyalty between those who did and didn’t receive a home visit.

The British Election Study interviewed those individuals again immediately after the election, asking them if they voted, if so how, and whether the parties had been in contact during the last month of the campaign. With those data we can see whether contact with the 2,590 individuals who intended to vote Labour when the campaign started a month earlier, plus a further 640 who were leaning towards a Labour vote (not all of whom voted Labour in 2005), made a difference…

It did. Most of those who intended to vote Labour did so, but there was a difference of up to 14 percentage points between those contacted by the party during the campaign and those who were not: 97 per cent of those visited at home voted Labour compared to 83 per cent of those not visited. Labour lost around 16 per cent of those who committed to it in March-April but whom it failed to contact during the heat of the campaign.

Some 40 per cent of those leaning towards Labour changed their mind during the campaign, but many fewer if the party contacted them then. Home visits were especially helpful in shoring up potential support. Only 66 received one, but 79 per cent of them turned out for Labour, compared to 58 per cent who were not visited. Even delivery of a leaflet helped: 65 per cent of those who received one decided that they would vote Labour, as against 56 per cent who didn’t.

We also looked at those undecided who to vote for when the official campaign started. Again, contact mattered: those with whom the party’s candidates and canvassers engaged during the next few weeks were much more likely to vote Labour than those who received no contact. Among the undecided, 40 per cent who received a home visit from Labour voted for its candidate, for example, compared to only 19 per cent of those not visited.

A crucial caveat to all this is that parties don’t contact all voters nor do they contact them at random. Rather, they target their contacts. Hence it’s not just a case that a voter is more likely to back a party that delivers a letter, it’s also the case that a party is more likely to deliver a letter to someone who is likely to back it. With emails there’s even more caveats because the group of voters who give their email address to a party and who don’t opt out subsequently is doubly self-selecting.

Even so, the evidence here is of a significant impact, albeit one where the figures need to be tempered by their wider context.

When canvassing, however, I hope it goes better than this…

You can read the other posts in the What do the academics say? series here.