July 2019 newsletter: The trouble with fake news

16 Jul 2019

Michele Belot, Professor of Economics at the European University Institute argues that the problem posed by ‘fake news’ lies not in fake facts, which can be easily checked — as social media companies are promising increasingly to do —  but in the misleading interpretation of those facts. This is much harder to identify.

A few weeks ago, Italian Interior Minister Matteo Salvini revealed that crime has reduced sharply between 2018 and 2019. Il Giornale reported the following: ‘According to police statistics, crime had decreased by 15 per cent with murders down by 12.2 per cent and attempted murders down 16.2 per cent. Other crimes were down as well, some even more drastically such as sexual violence which had seen a 32.1 decrease and robberies which were down 20.9 per cent.’ Impressive. The facts are impressive, and the source (Police statistics) appears reliable. Of course, we know that crime statistics are not necessarily a true reflection of reality. For a crime to enter a statistic, it needs to be reported or identified by the police. If the police reduce monitoring, does not enforce laws or make it harder for people to report a crime, measured crime will go down. This is a well-known issue in the research in economics of crime. The sceptics among us will quickly realize that and, as a consequence, take this ‘news’ with a grain of caution.

The issue of fake news has been a prominent issue in public discussions for a few years now. It is the plague of social media, and now companies like Facebook have invested massively in ‘global fact-checking initiatives’. Fact checking is certainly a way forward. It is the obvious way to start: let us first check if the facts are actually correct. ‘Is the statistic correct’, or ‘did X really claim Y’.

The Italian crime statistics are an example of a ‘fact’ that is not necessarily fabricated — the statistics may well be true — but the ‘news’ might be incorrect insofar as the reality may not be what it suggests. Fact checking algorithms can tell us whether the statistic is true, but they won’t tell us whether crime really went down. One could do further investigations of course — a more sophisticated investigator or algorithm could look at reports of police numbers assigned to departments monitoring crime. Or one could choose to trust media sources that do have a tradition of investigating beyond the mere statistics.

Either way, the choice is either to trust someone else’s report or analysis, or to investigate oneself. The first requires the ability to evaluate someone’s credentials in making a claim, the second requires time and expertise. The digital revolution has made it much easier to access directly the source of information and to publicize the results of our own private investigations. In a few decades, the world has now become a world with millions of amateur investigators and journalists. In this world of freely available information, it should be easy for the truth to emerge quickly, one would think. But with the growing availability of information/opinions, it is perhaps more challenging than ever to identify reliable sources of information.

Coming back to Italy again, another topic that has made the headlines over the last couple of years is vaccination. The anti-vaccination movement has become stronger and in response to falling vaccination rates, Italy has made 10 vaccinations mandatory to attend school. Like other countries, Italy has also conducted large ‘information campaigns’ aimed at educating people and informing them that vaccines are effective and safe. But here is the catch. Parents who do not vaccinate are generally not ‘uninformed’. In fact, websites casting doubt on the effectiveness or the safety of vaccinations often portray themselves as advocating ‘informed decisions’ and provide a lot of information, including links to real scientific studies. True, there are also celebrities like Oprah Winfrey and Jenny Mccarthy who express opinions on the safety of vaccines, but quite a few active members of the ‘vaccine-sceptic’ community claim to have reasonable credentials to warrant being taken seriously. Websites pro- and anti-vaccines are also full of statistics, which highly educated people like. For example, vaccine sceptics will often show a graph that shows the steady and continuous fall of the number of deaths caused by a disease, which often preceded the introduction of the vaccine itself, casting doubt on their effectiveness. In contrast, websites advocating vaccination will display graphs showing the sharp fall in the number of diagnosed cases of many diseases following the introduction of vaccines. No need to know fancy econometric techniques to conclude that vaccines are effective in reducing incidence. Unfortunately, few websites will show you both graphs.

The challenge is that even when gathering a lot of information, it is hard to become a real expert. Take for example a study published a couple of years ago on the effects of Hepatitis B vaccination (HPV) on brain development in rodents. The study was published in 2016 in the journal Psychoneuroendocrinology and states the following: ‘This work reveals for the first time that early HBV vac cination induces impairments in behavior and hippocampal neurogenesis. This work provides innovative data supporting the long suspected potential association of HBV with certain neuropsychiatric disorders such as autism and multiple sclerosis.’ Now if you have time on your hands and know experts in neuroendocrinonology, you may want to investigate this further. The journal is a peer-reviewed journal with an impact factor of almost 5, which must mean this is a very good journal. Still when I asked colleagues in the neuroendocrinology department, they pointed out important weaknesses of the study and did not agree with the conclusion. Their arguments were sufficiently convincing that they convinced me. The study may well be internally valid, but it is the interpretation of the evidence that seems problematic.

Ultimately, it is clear that on many (most?) topics one runs quickly out of one’s depth, even if one has several degrees and has spent hours doing ‘research’ on the web. But still many of us feel empowered to access and evaluate information, and to express an opinion on topics we have no specific expertise in.

Even when the interpretation behind the facts seems straightforward and does not seem to require any expertise at all, there appears to be scope for misinterpretation. For example, one of Trump’s most outrageous acts during the 2016 Presidential campaign was to mock openly a disabled journalist named Serge Kovaleski, at one of his rallies. We all saw the video. We actually all saw the fact. This is obviously ‘true news’, because we saw first-hand what happened. Nevertheless, there are some Trump supporters who openly denied what seems undeniable. Ann Coulter, a conservative commentator and author of In Trump We Trust has defended Trump for his gesture. She claims, backing her claims with video-footage, that Trump made the exact same gesture when referring to people who were not disabled (including for example Ted Cruz). She also claims that Steve Kovaleski has a disability that prevents his hand from moving — one can indeed find videos of a man with a hand that appears completely rigid. If one would want to mock him, making an uncontrolled hand gesture may not be the most obvious thing to do. So here we are. We can spend time watching the other video footage, perhaps even double check that the video of Steve Kovaleski is really a video of Steve Kovaleski, and read the Washington Post response on why Ann Coulter does not prove anything. And all this ‘fact checking’ investment appears necessary for a fact that seemed really undeniable. The bottom line is that it seems hard to find ‘pure facts’ that have not been subject to somebody’s interpretation. Somebody who may have an agenda, or biased beliefs, perhaps even unknowingly so.

The challenge of fake news is not fake facts. Fake facts should be easy to tackle and correct, certainly with our modern technology. It is much more challenging to correct for misinterpretations. Nevertheless, the interpretation of facts is essential to stir policy and to hold politicians accountable. The public needs to know whether crime really went down, whether vaccines are effective and safe and whether Trump mocked a disabled person.

On a positive note, it is not clear that there is more misinformation now than there used to be. Before, wild speculations may have been easier to maintain. Now everyone can investigate what everyone else is saying. You may have heard from someone who knows someone who developed autism after being vaccinated. Now we can at least try to check aggregate statistics, read scientific studies or read interpretations from ‘experts’. Even if amateur investigations may not be the most efficient way of getting at the truth, it is probably the best way, despite the caveats. There is now wide competition in investigations, and people are more able than ever to have a critical eye on facts and information.