THE PROBLEM WAS THAT THE NATIONAL WEATHER SERVICE AVOIDED USING PROBABILITIES IN THEIR FORECASTS. Of course government agencies, including the National Weather Service, should keep the public up to date on what their models are predicting. But the cause of the weather service’s problem is that they did not use probabilities from the beginning in giving the public their forecasts. For example, suppose that the probability of 24 inches of snow was announced as 60%. If that probability became 30%, the public would already have known that the 24 inch prediction had been subject to change and would be more likely to accept the change. If some members of the public are not now aware that weather probabilities are subject to change, they should soon learn a better understanding than a simple “the weatherman is sometimes wrong”.
This is a hobby horse that I have been arguing for fifty years. (Here is a post from 2007 arguing that forecasts should be expressed in probabilities.) Telling the public what the probabilities are would have the obvious advantage that it would accurately inform the public of what is known at each point of time.
The major objection that weather agencies have used to avoid using probabilities is the one that the weather service advanced a few days ago—-the public won’t understand. This battle should be already won. The weather service gives forecasts of rain in %’s and time of day as a matter of routine. One of the ways that people have learned about probability is through weather forecasts. The weather services have become more specific about using numbers over the last 50 years and consumer understanding of what forecasts mean has accordingly increased.
The after effect of this week’s forecasting events will be that people will include in their mental appraisal of forecasts that they have been made in part with a view to changing the public’s behavior rather than presenting the best available predictions.