I’ve been feeling that weather forecasts have become less helpful recently (that’s a purely personal and subjective view, albeit influenced I suspect by the fact that the great British summer has not exactly materialised this year!). But what does appear to be real is that forecasters are talking more about probabilities – eg ‘this is the most likely scenario but it is possible that. . . . ‘
And that’s set me thinking about Daniel Kahenman’s Thinking Fast and Slow and how one of the things that our System 1 thinking struggles with is probabilities. So, does offering people information that says it will probably be X but could be Y actually help in making better decisions or not? I am not qualified to comment from any specific scientific angle, but my experience of getting people to make decisions, particularly difficult ones in an environment of change is that alternatives, particularly those that involve probabilities are less than helpful. So, in weather terms, it may be that more information is really less, in terms of value offered, even if at the end of the day the actual view offered is technically more accurate. Knowing that X is materially more probable than Y is not at all the same thing from a decision making perspective as X is right and Y is wrong. And yet, to a great extent that is how many are using the data when they hear it.
I think there are lessons here for using data wisely. The ability to draw conclusions from data and analytics often will throw up the probability of a trend, or an event – but for whom is that information helpful? One of the reasons sites like Trip Advisor work for people is that the element of probability is removed – we trust people like me to make the same decisions as us – indeed in many cases to make the decisions for us in practice. Am I advocating this? No I’m not, but I think as we use data more and more, and we throw up trends and probabilities we need to get a good deal better at dealing with those as probabilities, rather than the black and white world of yes and no. I am for example an advocate of scenario planning as much because of its ability to help people understand that there are many possible variations of the future – some more probable than others but none absolutely and categorically right.
And speaking as someone who for varied and unplanned reasons has studied statistics at multiple points in my life, I have to say that certainly the way I was educated about them was of no help in thinking about probabilities and certainties in very different ways. If there are ‘Lies, damned lies and statistics’ (popularised by Mark Twain but attributed to many originally) then what can we say about data? Where does that sit on the spectrum?