A beginner’s misfortune of going viral

I have to admit, none of my research had gone viral before. I work on things that can potentially save billions of pounds to consumers by identifying how unscrupulous business behaviour or misaligned government intervention can harm people. Despite its relevance, the general public has never been particularly interested in my work.

On Wednesday evening I posted something on my blog that changed all of this. I stumbled upon an interesting correlation between the proportion of obese people and the proportion of Leave voters in the same geographical area. What made this really interesting was the finding that the correlation remained even after accounting for regional differences in income, qualification, or employment. I did not jump to conclusions, I reported the data, provided sources so that anyone could check my findings, and concluded that more work would be needed to identify what is behind this correlation.

Throughout the night my blog came to life and started its viral journey on Twitter. Within hours 1000s of people looked at my blog (I normally have around 5-10 downloads a day). My employer’s press office called that they would be interested in issuing a short brief, which I obviously did not object. By the afternoon, the local newspaper (EDP) published a piece with the sensationalist (and slanderous) title: “If you voted Leave you’re probably fat – study by UEA academic”. This was soon picked up and even further twisted by the Daily Express, and then all mayhem kicked off. A storm of abusive comments and messages followed, people posted my contact details highlighting my nationality under the Express article, leading to more racist slurs…

5 days later, things seem to be blowing over. Although a few obscure websites still dishonestly report that I labelled more than 17 million people ‘fat’, it is now significantly quieter.

Apart from the obvious lesson about the morality of the press I gained valuable experience from this. Some of these seem too obvious to report but even though I had been aware of all of these obvious things I still didn’t anticipate how true they are:

• As an academic I like to think that in the grand scheme of things I am an impartial spectator. By giving access to my data to allow reproducibility, I need to remain impartial and objective because if I’m not, people can easily find out. But once my work leaves the realms of academia allowing reproducibility means nothing. It’s blatantly obvious that most people do not want to reproduce my analysis. Suddenly I am only as impartial as anyone likes to believe it.
• Empirical evidence is unlikely to convince anybody that the opposite of their own belief is true. This is a recurrent discussion topic among my academic peers, yet, I naively assumed it didn’t apply to simple bits of evidence, especially when the interpretation is left open for the reader.
• For most people relationships are deterministic. Stochasticity, or randomness is a concept that is hard to interpret for most. When I said that areas with high obesity rates are typically areas with high Leave vote rates, to most people it means all areas with high obesity rates voted leave. This should not have surprised me. As an economist dealing with lawyers, I have noted this many times in the past, but it had never struck me with such intensity.
• I largely underestimated the power of availability heuristics: people rate personal anecdotal evidence higher than distant data-based evidence. “It didn’t happen to me so it cannot be true.” Again, this is something I had known, but I naively thought that sharing the data for replication would cancel this effect out. I can confirm, up to today nobody has downloaded the data I attached.