Making predictions is hard!

This week’s Silver reading fits the current events a bit too eerily with its focus on disease epidemics and their past predictions.  I remember the 2009 swine flu fiasco.  We cut pork out of our diet for a few months, and I was made to stand in line at my high school for hours in the rain and cold to get the vaccine, until I was too miserable and demanded to go home.  The results? No one we knew had the swine flu, and I kept my flu-free record without any vaccination(even until now, *fingers crossed*).  I have always taken these “outbreaks” with a few large grains of salt, not because I don’t recognize them as credible threats, for no one can deny the havoc ebola is causing in Africa, but because I understand how the media likes to blow things out of proportion.  But somehow, cases after cases of misleading forecasts, the general public still eats up all that the media dishes out.

From reading about all these under- and over-forecasting of epidemics over the past few decades, I honestly feel that statisticians are have the shortest end of the stick.  They can make simple models, or multi-levels complex models, and still ends up having the wrong predictions and get bashed around for it.  It brings up an interesting debate of what happens when you know your model isn’t precisely right, would it be better to be conservative, and keep things calm for as long as you can, and changing your strategy as the event progresses? Or over-forecast, and over-prepare, possibly resulting in loss of millions of dollars and/or creating other panic problems, like in the 1976 swine flu scare?  It is a really hard decision to make when you know that so many people rely on it.

I really like what Silver quote of Geroge E.P. Box, “All models are wrong, but some models are useful.”  I think that we should adopt this way of thinking.  Rather than trying to come up with a solid number of how many people will die of a disease, what might be more important is using models to figure out the components of the disease, how it can be spread and how quickly, under certain circumstances.  It might be frustrating when all we want is to put a number on it, but in the long run, it would be much more useful to look at the trends rather than the values.

This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s