Tuesday 17 September 2024

Should We Trust The Anecdotal Or The Statistical?

 

You’ve probably heard of the wisdom of crowds – the notion that, when it comes to decision-making and prediction, large groups of people are often collectively more accurate than individual experts. That is, with 100 people guessing, the average guess of, say, the weight of a cow is likely to be closer than the guess of a single expert (see blog post here). And you probably know that in statistical analyses, individual anecdotal accounts are generally less reliable than statistical data collations. The anecdote “I knew someone who smoked all his life and lived to 95” is a less good way to evaluate the life expectancy of heavy smokers than statistical analysis of a set of heavy smokers against non-smokers. And you’ve probably also worked out that the news and media are not robustly reliable channels for distilling the truth when compared with statistics. The media is biased, but the statistics report more accurately (although not perfectly) on facts.

Given the foregoing, what is the best way to get to the truth of a matter? Statistics are generally more reliable than individual anecdotes and crowd-based opinions, but at one level there is no truth quite as powerful as first-person truth. On the other hand, given that there is no such thing as an average person, and that the first person perspective cannot easily be representative of some hypothetical social mean, some of our first person perspective is bound to mislead us into thinking we reflect wider societal views or preferences. For example, if you’re convinced you live in a patriarchy, or under a right wing government, or in a Christian country, or under oppression, or in a country with great opportunity, then conformation biases might exacerbate those beliefs against wider counter-indicators.

Here’s where I think this leaves us. In some cases, the first-hand experience knows best, but other times it should give way to the wisdom of the wider consensus. And where we defer to the wider consensus, we should first do so through hard statistical data (that can be rigorously demonstrated), not skewed media narratives, which depart further from the full truth with every passing year.


/>