Hype and Humanity in Evidence: Doubt Is Not Pleasant, but Certainty Is Absurd Voltaire

Hype and Humanity in Evidence: Doubt Is Not Pleasant, but Certainty Is Absurd Voltaire

Preamble

Michael Blastland is a well-known writer, author contributing to a variety of publishing houses. He is the author of books titled “The Tiger That Isn’t”, “The Number Game”, “The Hidden Half”, and many others. The webinar aims to initiate the talk with Mr. Michael Blastland’s hype and humility in evidence is the main theme of his book “The Hidden Half”, which is a detailed analysis of research on data and how data is to be treated. The author provided a summary of the book and discussed all key points of the book.

Key Messages

• Michael believes we humans are far more ignorant than we tend to admit, very deeply fundamentally are ignorant. Moreover, he states that we are bedevilled by hype, various self-driving biases, overconfidence often based on statistical naivety.
• He believes the problem before decision-makers are juggling uncertainties and working out how to manage a state of uncertainty when we are just not sure. He emphasizes the fact that we must accustom ourselves to saying that we can’t know this and how; therefore, we manage when we can’t know this.
• Through his personal experience being a journalist, he states that in journalism, one of the most taught points is that you must simplify and amplify.
• He then proceeds to give several examples to prove his point regarding the simplification and amplification in newspapers and research publishing institutions and how much uncertainty there is in those data and how much we rely on evidence.

Some Areas Where We See Simplification and Amplification

Air quality

• We are often told just how many people are dying because of air pollution around the world. Nobody has ever reported how many people have been saved by the trend of betterment in air quality in those trends
• The Times, the BBC, and The Guardian, along with King’s College London, have done a great deal of research regarding the seriousness of the bad air quality around the world and how many people die because of it, yet they seem always to fail to mention the change in trends towards the betterment, as in the improvements and how many lives are being saved because of these trends.
• This approach is a simplifying hyperbolic one that everything is terrible, and the implication is that it is getting worse, contrary to the truth.

Obesity

• Michael does believe there is a problem of obesity. However, the issue he has is with the way it is reported by newspapers and websites
• Most major newspapers and websites and even Public Health England exaggerate the actual values to make the statistics obtained through vast surveys and research more ‘attractive,’ i.e. 38% is an ‘ugly’ number, so they notch it up to 40 %.
• you can see the process of the elimination of any uncertainty, simplification, and the exaggeration of the probable trend
• He believes that if they are there to persuade, how far they are allowed to go when they are persuading. As a result, he believes that our information is corrupted in both the media and the research world.
• He does not think that the people that do it are bad nor are they are lying, but he does think it is misleading and thinks it is a result of an exaggerated sense of what we can plausibly say; what it is fair to say

Unemployment and GDP

• The BBC published that unemployment fell by 3000 to 1.44 million; however, 95 percent confidence interval is plus or minus is seventy-seven thousand. So the fall of 3000 could have been an increase of 74000, that is the degree of uncertainty around employment; it is never reported by mainstream media and is buried by the national statistics office.
• Often, the absolute change in unemployment is far smaller than the degree of uncertainty around it.
• Similar to all the points mentioned above, there is also uncertainty in the points being published regarding the GDP of the UK. He was told in an interview that if a given set of data suggests that the economy is growing by 1%, it could be growing at around 3%, or it could be behaving at around 5% – that is the span of uncertainty. You can’t be sure
• In statements published by banks, newspapers, and researchers. The uncertainty is not acknowledged, forecasting uncertainty is better than is more acknowledged, historical uncertainty, present-day uncertainty not nearly as much

Covid

• Michael sheds light on how much the data gathered regarding different aspects of COVID varied immensely.
• One research data might say each case infects two people in a week, or maybe each case infects three others in five days which slightly tweaks the parameters. Consequently, the results are that one scenario gives you 2000, the second gives you 10 million. That range is staggering, and we did not know whether we were facing the former or later scenarios.

He believes that some of the things we can do to improve our research and correct simplification are: we must endeavour to broaden our views by taking account of a wide range of facts and possibilities, and when we are done so to the utmost of our power we must still remember that from the very nature of things our ideas fall immeasurably short reality.

 Question and Answer Session

During the discussion, the questions were raised, which were answered in detail by Mr Michael Blastland. Questions included:

• What is your stance on evidence-based policy when policies are made based on hard science and good Econometrics? And how do you see the confirmation biases being so strong that we can’t be objective by data at all?
• Given these two things that more observations should help narrow the band and that we are capable of generating orders of magnitude more observations than before, would this not give us more confidence about statistics in future?
• Does the evidence generation change if the person doing this activity has a different requirement regarding researcher and policymaker? Does the evidence matter nowadays? And to persuade people what would be more effective?
• How do you think can revise the incentive structures to promote nuance and complexity? And are these incentive structures not responsible for increasing levels of political polarization around the world?
• What do you think, why are we so obsessed with certainty in social sciences while objective sciences move towards uncertainty?

Response

In response to these questions, Michael answers the first question that we all have to make decisions, and we do it based on the best available evidence, but we must do it with an acute sense of the fragility of that evidence. He further added that we ought to demand sensitivity analyses that show how much our estimate depends on particular characteristics in the model. Therefore, We ought to demand confidence intervals from the people who give us this kind of information, and we ought to ask how reliable is this data, and how reliable has it been in the past. We ought to demand our policymakers the capacity to be flexible when we discover that the numbers weren’t the way they seemed. He said that the best way is to treat data with a sense of fallibility. It doesn’t mean that we stop making decisions, we have to make decisions to run the country, but there is a world of difference between saying I know and I think, possibly, maybe, and the strategies that follow from these two attitudes are radically different.

While answering the second part of the question, Michael stated that there are some large regularities and some areas where data is conclusive or conclusive enough that we can act with quite a lot of confidence. He agrees with the bias problem and describes that it is not true of everyone and that some more flexible minds prefer uncertainty. He further believed that the biases would not be the determining factor if people are willing to say that they do not know, they are unsure and that their data is unreliable, etc.

Answering the 2^nd question, Michael said his confidence in statistics would increase a little but possibly not as much as others would feel.

On changing the evidence generation, Michael agreed and said that different places emphasize different kinds of interpretation for everyone. He further argues that people do every possible thing to achieve their goals, and if it does not work properly, they switch their way of doing it. To the second part of the question, Michael had some different viewpoints. He continued by saying it seems to be a social problem, and one should not always be persuasive because there lies a matter of trust and trustworthiness, and most of the time, people demand trust and trustworthiness. He further enhanced his viewpoint by saying that the data which would be provided to the people should be authentic and provided by a trustworthy person, and after that, it should be left to the individuals what are they going to make it from.

In response to the 4rth question, Michael argues that we have seen reactions to the replication problem in Europe and the US which has started to emphasize collective work rather than individual reputation. He further informed that a very interesting publishing model about to launch in the UK called octopus doesn’t give all the glory to one principal investigator. It tries to allow people to publish part work as the individual contributions to the paper, these papers have multiple authors, and different people have done different things on them.

In response to the social and objective science question, Michael responded that he is not sure about the whole working of the world between natural science and social science, as, in social science, he believed that the people could isolate the causes more or less in a peculiar way. He further added that he prefers the illusion of things being uncertain to things being certain in an odd way.