Despite the (still) prevailing orthodoxy that election strategies are a matter of well-timed policy announcements, media management and “feeding the chooks”, and focus-grouped television ads, the fact is that personal contact is the most effective way to change someone’s voting intention.

A while back, I had a discussion with someone who posited that political campaigns made carefully calculated judgements about the mix of campaigning — how much television, direct mail, phone canvassing, door knocking and radio, etc — based on a formula to maximise impact and vote.

Unfortunately, endless research and insider accounts of elections suggest that the reality is more akin of alchemy than chemistry.

Nevertheless, on the other side, there is research that looks at what actually works to influence someone’s voting decisions.

The Australian literature is different and scarcer than US or UK research — largely in my view, due to compulsory voting. In voluntary voting nations, there is an incentive to determine why people vote. In Australia, we (think we) know why people vote: they’re forced to. Most academic research therefore looks at the meta-analysis: political communication, televised leader debates, economic trends, media appearances, polling (internal and public), demographic changes, strategic policy announcements, etc. What we don’t look at enough is why people vote for one candidate or another.

A substantial research on the effect of personal contact on affecting someone’s voting preference is from Kevin Arceneaux, an academic from Temple University in Philadelphia, whose 2007 article “I’m Asking for Your Support: The Effects of Personally Delivered Campaign Messages on Voting Decisions and Opinion Formation” looks beyond the normal “get out of the vote” research.

Arceneaux argues that political campaigns (in the US) not only have to encourage their existing supporters to go to the polls, but also “persuade others to support their candidate”. He notes that “The persuasive aspect of grassroots campaigning has certainly not been overlooked by political scientists” — something that I agree with.

His research relies on randomised field experiments rather than observational survey data. The downside of survey data is that it is difficult to rule out other influences on voter decisions than just personal contact (for example, advertising, media reports, direct mail, etc). Randomised field experiments allow for comparisons between treatment groups — and his in particular focus on voter decision, not “get out the vote”.

Arceneaux examines both personal, face-to-face contact (door-to-door canvassing) and phone canvassing (specifically calls made by paid phone bankers). The field experiments used trained canvassers who spoke to targeted voters with a carefully prepared script. They supported a Democratic candidate.

Both the door-to-door and commercial phone bank efforts increased support for the candidate. Those canvassed by the candidate increased their support by more than thirteen percent compared to the control group; being canvassed by a campaign worker were over eight percent more likely. Voters who received only phone-calls were over eleven percent more likely to support the candidate.

What is important about this research is not just that personal, face-to-face contact is effective in changing voting intention, but that mass media really does not influence voter choice much at all. To the extent that mass media advertising does influence voters, comes down to the concept of “priming”. That is, there is some evidence that advertising helps cause some considerations to become “more accessible” than others. However, this simply emphasises existing concerns or beliefs rather than changing them — and of course there is no guarantee that the “primed” consideration will have any bearing on the voter’s eventual choice.

Thus, there is no real evidence that television or radio ads can actually change someone’s beliefs or opinions about a candidate, although there is no doubt a lot of belief amongst political circles that it does. In fact, research demonstrates that most people “ignore” the mass media when it comes to forming their voting intention. Simply put, political advertising has a minimal effect on voter preference.

What Arceneaux’s research does is show that campaigning — in particular the personal contact, face-to-face contact, makes a substantial difference. The best kind of contact is from the candidate herself/himself. Interesting is that other research suggests that while phone banking has little impact on “get out the vote”, it does make a marked improvement on voter choice.

Usefully, Arceneaux also breaks down some of the finances of personal contact:

The field canvassing results provide a conservative estimate of the effect that canvassing has on candidate support with the campaign workers attracting one supporter for every 12 attempted contacts. If canvassers were paid $16 an hour and they attempted to contact 20 houses per hour (a more conservative number than Green and Gerber’s), the campaign would, on average, attract one supporter for every $9.60 spent on canvassing. The phone bank generated one supporter for every nine attempts. So, assuming $1.50 per completed call and a 50% contact rate, a commercial phone bank would attract one supporter for every $6.75… With the same assumptions about costs, a canvassing operation would mobilize one supporter for every $10.40 in labor cost and a phone bank would mobilize one supporter for every $10.50. In short, it appears that personally delivered campaign appeals can be a highly cost-effective way to attract and mobilize supporters in low salience elections.

As an end note to this research, I’d like to point to another paper by two academics Price and Lupfer, who show that canvassing has an effect on preventing defections of support. That is, it helps lock in support from “leaners” who are considering switching their support from one candidate to another. There is good reason to believe that focusing on these potential switchers is far more effective than trying to identify and convince “undecided”.