The good Lord willing and the creek dont rise the caveats of forecasting
When it comes to forecasting, Ive definitely got an outside looking in perspective. But sometimes thats a good vantage point for taking a good hard look at what can prove to be simple problems.
In my experience, people can handle almost anything that they are aware is possibility. And isnt that what forecastings all about preparing our audience for the possible realities so they can adequately plan and make decisions.But often forecasts are taken as reality, rather than a prediction of possibilities. This is well characterized in a recent article from the Grand Forks Herald .
For those of you who may not follow US weather, some of the worst flooding in more than a decade has recently plagued North Dakota along its Red River. The Herald points out that current practices of forecasting ranges, rather than a single number, for river crests is a marked improvement since the last major flooding in this region in 1997. Then, forecasters predicted a river crest of 49 feet and local residents took the number to be gospel, and were surprised (sometimes dangerously so) and angry with forecasters when the river actually crested at 54 feet.
To their credit, the National Weather Service learned from the experience and when forecasting crests last week offered ranges first 39-41 feet and then 41-43 feet. Although the prediction changed mid-stream (pardon the pun), the ranges instilled in residents a bit of the uncertainty that comes when trying to predict the wraths that Mother Nature unfurls.
The Herald urges weather forecasters to make sure forecasts dont imply certainty. But its a good bit of cautionary advice for us all. When we forecast, our audiences dont have the same perspective we do on all of the variables that play into making our predictions. They only see what seems to be a polished, packaged number that they may mistakenly believe is a given.
Part of delivering a good forecast comes in communicating in a way that our intended audience is sure to understand the level of confidence we have in the numbers. And in the case of river crests and a plethora of other data we relay to our constituents that may mean building the element of uncertainty right into the numbers themselves by offering a range or other caveats that stay attached to our predictions.
Its a bit like when I was a small child and asked my father every morning as he left for work, Youll be home for supper, right, Daddy? His answer was always: The good Lord willing and the creek dont rise. He meant, of course, that he had every intention of being home for dinner, but that life is full of uncertainties none of us can predict. The good folks in North Dakota know a thing or two about that.
A seasoned pharma forecaster once told me that its not necessarily that a forecast is good or bad, but more that the way in which we explain it to the intended audience is either effective or not. And sometimes I think we definitely neglect to adequately communicate the caveats, forgetting that they may ultimately prove to be the most integral part of the forecast.
Just a little food for thought - and my two cents.