Sunday, January 18, 2015

Speaking Statistically and Worrying About Stuff

I must begin by making the disclaimer that I have never taken a statistics class.  However, I have tutored four students in college statistics and they all passed.  Would I advertise myself as a stats tutor?  No, but if I was trying to sell my services I could pitch it like this:  Stats Tutor-  100% of students pass.  Prospective students might feel 100% confident, but the tutor would be a whole lot less confident, about 99% less confident (and you should be asking:  100% confident in what and 99% less confident than what?)

How did I get on this horrible subject?  I was reading an article that stated that there was a device for those caught in avalanches that boasts a 97% survival rate. The device helps to keep you on the surface of the snow.   I was intrigued and followed a link that led to a blog dealing with this statistic.  What is interesting is that the survival rate without one of these devices is 81%!  That's right, 81% of those caught in avalanches survive without this nifty device. But if we take a moment to look even more closely, the way blogger Bruce Tremper did, we see that a lot of survival depends on what kind of terrain you are on when the avalanche hits.  What reduces survivability is "high consequence terrain" as Tremper calls it.  And for those people caught in an avalanche on this type of terrain the chances of survival plummet no matter what kind of gear you bring with you.  Why?  Because the avalanche isn't directly the cause of death; it sweeps you into a tree, dumps you off a cliff, slams you into a rock.  Thus the 97% survival rate device is not likely to make a difference.  (Tremper's article goes on to show other factors and fallacies in this 97% survival rate.)

Stats are an interesting thing.  They are based on math, but they are only as good as the material that is fed into the "machine" and the mind that creates or uses the statistic. It is easy to get sucked under by the avalanche of negativity when we read about survival rates, disease rates, accident rates, etc.  If we took the time to find out exactly how these stats were obtained, we might find that there is a lot less to fear than it looks when someone uses the stats to bolster their point of view.  Many of us will question the stats if it is obvious that someone seeks to profit from using them. But how often do we accept the stats when that profit is hidden, as in the case of medical stats that bolster the "need" for  more medical services and research?  The media uses statistics to gain more readers who want to be "informed" of the dangers lurking out there.  And the stats themselves can be highly suspect, as the example of the device to "save" you in the event of an avalanche and many, many others. 

In fact, it might be fair to say that if you don't know how the statistic was obtained, disregard it.  With all statistical evidence: get suspicious.  And if you think that the stat is accurate, think more deeply about what it means.

As Mark Twain famously said:  "There are lies, damned lies, and statistics."  But my personal favorite is on Brainy Quote by humorist Art Buchwald,
"The buffalo isn't as dangerous as everyone makes him out to be.  Statistics prove that in the United States, more Americans are killed in automobile accidents than are killed by buffalo." 
I couldn't find the actual numbers, but we can probably agree that the evidence is there to back up his statistical statement (even if the animal in question is a bison.) From this we could further deduce, (following the loose logic that so many use with statistical information):
Need transportation?  Buy a buffalo-- statistics prove that they are safer than cars! 
Next time you are in Yellowstone in late July or August (rutting season) and a herd of buffalo crosses your path- get out and walk.  You are safer near the buffalo and than in your car.

We need to think on these things before we "buy" the latest statistic offering "proof" that we should be worrying about something.

No comments:

Post a Comment