Cybersecurity · Information operations · Information Warfare

Facebook ‘dark ads’ can swing political opinions, research shows


Monday 31 July 2017 05.07 EDT Last modified on Monday 31 July 2017 08.41 EDT

Using “psychographic” profiles of individual voters generated from publicly stated interests really does work, according to new research presented at the Def Con hacking conference in Las Vegas, Nevada.

Ads targeted using profiles generated from individual voters’ stated interests are more successful in shifting attitudes according to Online Privacy Foundation

Alex Hern in Las Vegas @alexhern

The controversial practice allows groups to hone their messages to match the personality types of their targets during political campaigning, and is being used by firms including Cambridge Analytica and AggregateIQ to better target voters with political advertising with so-called “dark ads”.

“Before the referendum results, the concern we had was that people’s biases were being manipulated, either intentionally or unintentionally,” said Chris Sumner, research director and co-founder of the not-for-profit Online Privacy Foundation, who led the research. “Now we’ve seen this [research], I’m as concerned as I was before.

“It’s not a surprise, it’s what we expected to see. People on one side, whichever side happens to be winning at the time, are going to say ‘no, it’s not a problem’, while people who have just lost are going to see it as a big problem.”

Psychographic profiling classifies people into personality types using data from social networks such as Facebook. Sumner’s research focused on replicating some of the key findings of psychographic research by crafting adverts specifically targeted at certain personality types. Using publicly available data to ensure that the adverts were seen by the right people at the right time, Sumner tested how effective such targeting can be.

Would-be voters were sorted into two groups of people, those with high and low authoritarian tendencies, using a mixture of age, gender, location and interest targeting: younger women score low on authoritarian rankings, while older men score highly.

Geographically, Sumner selected five local areas which have been found in previous research to have low authoritarian attitudes – Cambridge, Liverpool, Manchester, Edinburgh and Hackney – and seven with high – Basildon, Chelmsford, Dudley, Thurrock, Mansfield, Rotherham and Swindon.

But the most important factor was the use of publicly stated interests to derive a psychographic profile of the recipients. Using information Facebook already knows about its users, Sumner created a high-authoritarian group of people the site has identified as being interested in conservatism and the Daily Mail, and a low-authoritarian group interested in liberalism and the Guardian.

To test the groups were accurately sorted, they were asked whether they agreed with the statement “with regards to internet privacy: if you’ve done nothing wrong, you have nothing to fear”. When randomly sorted, 38% of people agreed with the statement; but once all the psychographic signifiers were incorporated, the low authoritarian group fell to just 25% agreement, while the highly authoritarian group rose to 61%.

Knowing the psychographic profiles of the two groups is more useful than simply being able to accurately guess what positions they already hold; it can also be used to craft messages to specifically target those groups, to more effectively shift their opinions. Sumner created four such adverts, two aimed at increasing support for internet surveillance and two aimed at decreasing it, each targeted to a low or high authoritarian group.

For example, the highly authoritarian group’s anti-surveillance advert used the slogan “They fought for your freedom. Don’t give it away!”, over an image of the D-Day landings, while the low authoritarian group’s pro-surveillance message was “Crime doesn’t stop where the internet starts: say YES to state surveillance”.

Sure enough, the targeted adverts did significantly better. The high-authoritarian group was significantly more likely to share a promoted post aimed at them than a similar one aimed at their opposites, while the low authoritarian group ranked the advert aimed at them as considerably more persuasive than the advert that wasn’t.

Psychographic targeting is allowed on Facebook, and the company advertises the platform to politicians as the perfect way to “persuade voters” and “influence online and offline outcomes”.

But the ability for campaigns to perfectly target different messages to different groups has been described by some as a concern for democracy itself, allowing politicians to appeal to the worst side of voters in an almost undiscoverable manner.

Source: https://www.theguardian.com/technology/2017/jul/31/facebook-dark-ads-can-swing-opinions-politics-research-shows

Advertisements

2 thoughts on “Facebook ‘dark ads’ can swing political opinions, research shows

Comments are closed.