Our fears over digital campaigning are misplaced
Don't believe all the hype and hysteria about Cambridge Analytica. The evidence just doesn't stack up, writes Kevin Cunningham
In the context of the Cambridge Analytica scandal, a panic has developed about the potential for digital campaigning to 'rig' the upcoming referendum on the Eighth Amendment.
There is a misplaced fear surrounding the efficiency of digital campaigning. However, the idea that Cambridge Analytica and their methods are a threat to democracy is not only extremely far-fetched but it is counter-productive to coming to terms with understanding the motivations of voters and current trends in Western democracies.
To explain why this problem is overstated, we must briefly explain this 'psychological warfare'. Cambridge Analytica's personality traits application amounted to surveying 300,000 people on Facebook about their personality, asking people to evaluate the accuracy of a number of statements such as "I don't talk a lot", "I seldom feel blue", and "I get upset easily". Responses to these questions enabled Cambridge to categorise people according to the 'Big 5' personality traits, namely Extroversion, Conscientiousness, Openness, Neuroticism, and Agreeableness.
Cambridge also requested access to what respondents had liked on Facebook and using this data they could infer a relationship between personality traits and what people had liked on Facebook.
Those that were more 'Agreeable' tended to like 'The Bible' while those that were 'Disagreeable' tended to like 'Rammstein' - a heavy metal rock band, for example. They would then be able to assume, for example, that anyone who liked Rammstein on Facebook would be more disagreeable and they would then target these people with ads in line with this personality trait.
While some have confidently asserted that Cambridge Analytica's work amounts to one of the most successful campaigns of digital persuasion yet, the evidence is contrary to this. In the context of declining class identities and partisan loyalty, the academic literature does point to the relevance of personality traits in dictating voting behaviour.
However, it is a leap to assume that one should target on the basis of personality questions. Asking people to objectively self-assess their own personality is difficult as self-perception is highly subjective and heavily depends on mood. As evidence shows, it is wildly unreliable.
Indeed, in 2017 I was able to help win a contract in the UK on the basis of conveying such evidence. Secondly, Facebook 'likes' age badly. The data is quite sparse as many people haven't liked anything on Facebook and the number of people who regularly use the tool is quite limited.
Even the academic paper on which this is based reveals that the relationship between Facebook and personality is weak.
Indeed, these tactics led to the company being fired from the Ted Cruz campaign after their 'Military Grade Psychological Weapon' turned out to be little better than tossing a coin.
In Oklahoma, more than half of those that they identified as being supporters of Ted Cruz turned out to favour other candidates. Furthermore, as good (or indeed bad in this case) as the targeting may be, the message itself is far more important.
Facebook data reveals that per-advertisement Hillary Clinton's campaign received higher levels of engagement. This, coupled with the fact that Hillary's campaign outspent the Trump campaign by a ratio in excess of 3 to 1, should attest to the assertions of many studies that the effects of digital Facebook advertisements are weak. Every election gives rise to a pride of gurus equipped for the new formula for political success: Saatchi and Saatchi benefited from Margaret Thatcher, Stan Greenberg benefited from Bill Clinton, Lynton Crosby from the UK Conservative Party, Dan Wagner from Barack Obama. Cambridge Analytica happened to be in the right place at the right time, riding a wave of populism that established firms wouldn't touch.
Democracy is still threatened, but by its traditional foe - big donors. The final price tag of the 2016 elections was $6.1bn. Clinton's largest donors were all associated with hedge funds seeking to influence public policy ensuring that should she get in she would do little to upset their interests.
It is implausible that the Clinton campaign would not have used behavioural, attitudinal and psychographic data in their message testing and deployment in the campaign. The only difference is that the Clinton campaign had the money to merely buy-in whatever data they wanted, outspending the Trump campaign by enormous margins. Almost all campaigns, political and commercial buy databases and conduct segmentation based on behavioural and psychographic variables. What CA did was perhaps petty theft up against a juggernaut that had the capital to buy the shop.
There are those who would prefer to believe that there is some conspiracy at the heart of any perceived malaise, rather than acknowledge or seek to understand other points of view. It is worth remembering the context of the election results currently under the microscope and they are not at all at variance of what should have been expected.
The evidence suggested that there was a particular malaise in towns and villages in western democracies suffering from depopulation in the aftermath of the global recession. There are deeply held concerns about the decline of local communities and what some have failed to understand is that these concerns in many cases outstrip economic concerns.
This rise of populism has continued to impact elections with no relationship to digital campaigning, including the most recent Italian general election.
While the theatrical skills of Donald Trump and indeed Nigel Farage have helped their cause, so too has castigating these people and their followers which merely ensures that they are viewed as an agent for change among those desperate for change.
- Dr Kevin Cunningham is Director of Ireland Thinks and former targeting and analysis manager for the British Labour Party