Campaigners used big data sets and Facebook to hyper-target voters in the Brexit and US presidential elections to surprising effect. Could the same techniques tip the balance in New Zealand’s election? Bernard Hickey reports.
Politicians and sometimes shadowy figures behind the scenes used the power of big data to target and convince just enough voters through their Facebook news feeds to shock the pundits and the pollsters alike in the Brexit and Trump election results.
Campaigners in Britain and the United States are reputed to have used billions of data points to build pyschographic profiles of millions of voters so they could send individualised messages to get them either to vote for their candidate, or not to vote at all. These techniques rely on building huge banks of data from electoral rolls, financial records, spending records and the liking, sharing and posting records of Facebook users to understand exactly who a voter is, where they live, what they do every day, what their political views are, and, most importantly, how likely they are to change their vote.
These campaigners were then able to send individualised messages to voters in the counties, states and regions that Trump and Brexit needed to win, and just enough to tip the balance. It is the equivalent in political campaigning terms of reaching into a voter’s brain, understanding what message will hit the right pleasure or pain centre that will trigger the correct vote or no vote at all, if that is what is needed.
That’s the beautiful theory. There is plenty of debate still about whether it was actually used to much effect in practice in last year’s elections in Britain and America. The poster child for these techniques, a company called Cambridge Analytica, boasted that it had built psychographic profiles on hundreds of millions of voters that could be used by Brexit and Trump campaigners in tandem with Facebook to get their votes. These techniques were very costly and accessible only to the richest campaigners, but it has raised questions about whether the wealthiest campaigners with the most expensive data scientists could effectively screw the scrum in an election to further their interests. Cambridge Analytica is owned by a data science billionaire and Trump supporter called Robert Mercer, who in turn was the sponsor of Steve Bannon and the funder of alt-right news network Breitbart.
But could it happen here? And are these tools and techniques already being used in this election?
“They may know the inside of my soul and everything about me, but it’s still not helping them to persuade me.”
The short answer is New Zealand’s political parties are starting to use some information gleaned from various sources to try to reach voters on Facebook, but these efforts are on a much smaller and less targeted scale than in America or Britain, and campaigners think they won’t be the magic bullet that can win an election. Not yet at least.
This week Victoria University hosted a panel of digital strategists attached to the three biggest parties, who discussed the potential for big data to be used and how effective it might be in this campaign ahead of the September 23 election.
Danyl Mclauchlan is a Victoria University scientist and novelist who is helping the Greens with their campaign. He said these techniques could be effective at the margin, but he was sceptical about how many people could be convinced through their Facebook feeds and google searches.
“If you take off your ad blocker and look at the actual ads that Google and Facebook are serving up, they are for products and they are for political parties, but the algorithms don’t seem to be very accurate,” Mclauchlan said.
“They are not selling me anything I want to buy. They’re not telling me anything about political parties that I find remotely persuasive. They may know the inside of my soul and everything about me, but it’s still not helping them to persuade me.”
However, he cautioned voters in the audience to be careful about how much information they give away by using Facebook quizzes or commenting and posting on political issues.
“Our time and our private information is actually the product that these companies are monetising and selling to political parties and market research companies,” he said.
David Farrar, a pollster for the National Party, said social media and data analytics were useful components of a campaign, but they were only secondary or tertiary factors in how well a party does.
“The most important factors for political success are brand, competence, unity, likability, the country’s mood and even, yes, policies do count. If you don’t have the basics right no amount of Facebook likes are going to help you greatly,” Farrar said.
“But if you do get the basics right in social media, the data can be useful tools towards growing your vote and targeting voters.”
Farrar referred to how the Obama campaign had assigned a percentage score on how likely each eligible voter was to vote, and then to vote for Obama.
“I don’t think New Zealand has got to quite that level. The fact that I get letters from Grant Robertson asking me to vote for him suggests targeting in New Zealand hasn’t got there yet,” he said.
“Data costs money. Colin Craig spent $4 million and Kim Dotcom spent $5 million and got basically nowhere.”
Rob Salmond, a pollster and strategist who works with the Labour Party, interjected to say he would make a point of crossing Farrar off Labour’s list.
Farrar also pointed to the heavy and ineffective spending on data and targeting in the 2014 election by the Conservative Party and the Internet Party – both of whom failed to get into Parliament.
“If you look at the last election it wasn’t just about data. Data costs money. Colin Craig spent $4 million and Dotcom spent $5 million and got basically nowhere. So I wouldn’t be too worried about it. Yes it is more sophisticated and all that, but the only reason it works is if people are somewhat open to hearing those messages.”
Salmond was also sceptical of the ability of political parties to swing an election through targeting voters on social media.
“The data work, certainly in New Zealand, is much much lower level than we would like to give ourselves credit for. And part of that is the data laws in New Zealand, which are a lot more restrictive than in some places like the US, and some of it honestly is about the budgets that people have to play with, which are tiny compared to what they are overseas,” Salmond said.
“Even though Cambridge Analytica in the US didn’t manage to get their full psychographic profiling — they had it in theory but not in practice, which isn’t very useful in an election — they were gathering all manner of troves of data about millions of people, and I don’t think New Zealand political parties can afford to do that with the resources that we have, so you can sleep slightly easier in your beds about that,” he said.
However, he also cautioned voters to be careful with how they used social media to be politically engaged, given all that data was being captured and could be used in future.
Auckland University political scientist Jennifer Lees-Marshment agreed that many voters were not that gullible that they could be convinced by their news feed alone.
“The only reservation I have about that is that it is usually the first company or political party to use a new tool in a certain way tends to gain a bit of an advantage, but pretty much everyone else will then catch up,” she said.
Farrar said New Zealand’s capability to use data for political campaigning was lower because there were just two to three million social media users here, versus 270 million in the United States.
“But certainly the capability is there,” he said.
Farrar pointed out a voter’s electorate profile could be linked to a Facebook profile if the political party had an email address to go with it.
“If you have given a political party your email address then they can link that to your Facebook profile, and even if they haven’t, just having your suburb and your name and date of birth on Facebook, that may be enough to link you to your electoral profile,” he said.
So the conclusion is political marketing on social media is still in its infancy in New Zealand in 2017, but given the pace of the development of big data sets and the falling cost of using these tools, the elections in 2020 and 2023 are more likely to be fought out in the news feeds than in the newspapers or television debates.