Anchoring Bias and Ensembles

Anchoring bias causes us to rely heavily on the first piece of information that we encounter.  I think of it a lot when I am interpreting model guidance, especially ensembles.   

Here's an example, which I draw from the new 82-member Utah Snow Ensemble.  One of the first products I often pull up is the plume for Alta-Collins.  I find it impossible not to immediately look at the snowiest ensemble member, which for last night's run is producing about 28.5" of snow from 3" of water.  

Like a drug, my heart rate doubles and I feel the adrenalin rush.

Then I pull up the four-panel total snowfall forecasts through 240 hours.  Again, my eye moves immediately to the lower right, which is the maximum snowfall.  Beautiful!  28.5"!

The problem is that such amounts are extreme outliers and highly improbable. What you don't see in a first glance at the plume is that more than 90% of the ensemble members aren't producing any snowfall.  That would probably be the best anchor for a forecast.  

Fight your primal urges when looking these forecasts, especially if your eye is drawn to the maximum.