One of the greatest advantages of living in Melbourne, Australia, is the Australian Football, colloquially known as Footy (but pronounced ‘Foody’).
I was watching the game between the West Coast Eagles (WCE) and Melbourne (MELB) last Thursday when the TV commentator made the comment that there is a 20% chance of rain during the game. My teenage son, who was sitting next to me, thought for a second and made an observation that this does not make sense. It is either going to rain or it is not. And what does 20% chance of rain mean anyway.
Dealing with probabilities isn’t easy nor is it intuitive as it all depends on your interpretation. Let’s look at the options:
1. 20% chance of rain means light rain
2. 20% chance of rain means we can take an umbrella with gaps such that only 20% of the umbrella is covered.
3. 20% chance of rain means that only 20% of the stadium will get wet.
Clearly, none of the above is true. So how are we to interpret this piece of news?
Weather forecasts are scientific predictions based on what happened in the past during conditions that are similar to the current weather patterns.
When a meteorologist says there is a 20% chance of rain, he/she means that since we have been keeping weather data, 20% of the time that similar conditions existed, it rained and 80% of the time that similar conditions existed, it did not.
Even when there is a 100% chance of rain, it is possible (but highly unlikely) that it will not rain. This is because the 100% chance is based on what happened every other time in the past when we had the same or very similar weather conditions.
The above has profound implications to all predictions, even those based on past experience. Clearly, the more past observations your predictions are based on, the greater the probability that your future predictions will be close to being correct. With very large number of observations comes a near certainty – what some call ‘virtual certainty’. If your historical records are lacking or your sample set is small there is a possibility that even 100% certainty will be found to be incorrect and so your future predictions will need to be adjusted accordingly.
Think about it!