I SUPPOSE THE THING ABOUT STATISTICS is that you have to have some idea of the context in which they appear if they are to have any real impact.
For instance, it is an almost daily occurrence to hear that Facebook is sliding away in popularity but a recent survey showed that, in quarter 4 of 2016, 75% of male internet users were on Facebook as well as 83% of female internet users.
“So what?” you might ask; is that meaningful and does it indicate any significant shift in loyalty and usage?
Perhaps that’s the wrong question. This same article, published online in Social Media, showed that 22% of the world’s total population is on Facebook, so the scale and reach remains phenomenal.
Maybe the question I should be asking is: are people using it differently or are they cherry-picking social media sites and platforms for different purposes? We know that female internet users are more likely to use Instagram than men in a ratio of 38:26% and so we might easily conclude that female users use it differently from men but, even if that’s the case, we also know that 32% of teenagers consider Instagram to be their most important social network and that most Instagram users are between 18-29 years old, even though Snapchat reaches more than 40% of 18-34 year olds, so it’s clear that there’s no easy answer.
What we do know is that overall usage of social media is vast and varied – just 10,000 YouTube videos have generated more than 1 billion views and over 400 million snaps are shared on Snapchat every day. With overall usage still growing across all age groups, we need to better understand how people are using the internet and social media if we wish to better harness this tidal wave of behavioural change.
A recent study from King’s College London has challenged the idea that people are passive or helpless consumers of media in a publication which analysed online media use in a group of 8,500 teenage twins.
By comparing identical twins who share all the same DNA with nonidentical twins who share half of it, the study investigated the relative contribution of genes and environment to individual differences in engagement with online media ranging from education to games and social sites such as Facebook.
Results are expressed as “hereditability” scores that show the percentage of difference in individuals which can be attributed to inherited genetic factors rather than simply environmental effects.
Hereditability was 39% for gaming, 37% for entertainment sites, 34% for education and 24% for social networking. Researchers claim that these are larger effects than could be demonstrated by gender differences in media use and raise interesting questions about “filter bubbles” in social media which expose people only to information which already supports their point of view while sheltering them from conflicting or opposing arguments.
Echo chambers
We are all familiar with the idea of “echo chambers” in social media use where information is automatically tailored to the viewpoint of participants by their selection of a given platform within a networking experience.
This seems little different from the idea with which most of us grew up, that different newspapers “supported” a different point of view.
I have always believed that The Telegraph was a right-wing paper while The Guardian took the opposite polarity – and one’s choice of newspaper was a reasonable indicator for our political standpoint.
Within the confines of academic research, the idea that inherited genetic rather than environmental factors might affect an individual’s social media usage is fascinating, but how might this pan out in the real world?
The recent divisive US electoral process gives us something of an insight into how people are already harnessing this. While the party heritage of the electoral candidates remained constant, the issues on which the electorate cast their votes were miles off the campaign messages of both parties.
Intriguingly, Mr Trump’s back office chose to use a proprietary technique known as “pyschographs” which attempts to influence consumers, and in this case voters, with targeted messaging designed around psychological profiling based on market data.
The element of psychology is provided by persuading immense numbers of people to complete surveys on their chosen social networks and then compares that information with data it has harvested about other aspects of voters’ lives so that it creates psychological patterns. This enables the provider to “extrapolate backwards” to create targeted messaging that resonates with the fears and aspirations of millions of individuals.
The company behind this, Cambridge Analytica (CA), claims to have between 4,000 and 5,000 data points on every adult in the US and was used by the Republican party in 50 regional campaigns during 2016, seeking data which could be used to “flip” voters to support Mr Trump or to persuade Democrat voters to stay at home and not vote.
Achieving a far improved reading of voters’ intentions, particularly in swing states before the vote, would appear to have been an effective use of our growing understanding of how best to use our relationship with social media to affect behaviour.
Changing our understanding
Of course, we might say that such techniques would never work here although CA also advised the “Leave” campaign in its successful bid to persuade the UK to leave the EU. What this unequivocally shows is that data science is changing our understanding of digital privacy and the sanctity of democracy in ways that most of us did not know and may not fully understand when we do.
In the same way that we may need to require a kill-switch in robots before AI runs away with itself, the political landscape will never be the same again and any disquiet we may already feel about the state of affairs around the world may increase as environmental differences in accessibility and availability of social media diminish and digital ringmasters more widely use technology to control our actions.
