Monday, January 16, 2012

Science is never 100%

The new twist in Mass Lt Gov Murray's car crash is very relevant to forensic investigations of all kinds, especially in human factors.

For those of you unfamiliar with the story, here is a basic recap.  He was driving down the highway late at night and crashed into a tree.  He claimed that he was going about the speed limit and the car slipped on some black ice.  He had no problem releasing the black box recorder (like the ones in planes) that was in his government owned car.  When the data was analyzed, they found he had accelerated to 100 mph just before hitting the tree.  They concluded that he must have fallen asleep at the wheel.

From a human factors point of view, this makes sense.  If you are nodding off and don't realize it, then suddenly wake up as the car hits the grass, it would seem like you had slipped on ice.  It is also possible that slumping forward would cause your weight to push harder on the pedal and accelerate.  So his story made sense.  There were no alcohol or drugs in his system and the forensic science fit the story. 

The Lt Gov was convinced and admitted the possibility that he could have fallen asleep, although he didn't remember it.  The media was convinced too.  Case closed.

Then today, they reported a new analysis on the black box data suggesting he really did slip on ice.  He was going 75 (not the speed limit) just prior to the acceleration.  He acknowledges that is possible.  So now the case is not closed but confused.

All science makes conclusions based on less than 100% certainty.  The standard can be 95% for many things, or 99.999% if its life threatening or mission critical.  But nothing is ever 100%.  So it is not surprising that this new conclusion could arise from a more detailed analysis of the data.

But people outside of science are led to believe (by the media and our high school science books) that science is always 100% sure of everything.  The media was sure he had fallen asleep at the wheel. The driver was convinced by the black box analysis, even though he didn't remember falling asleep.  The public bought it too.  100%.

In one sense, it is good that the public thinks in absolutes.  When scientists generally accept a theory at the appropriate level of confidence, it is better to design public policy as if it was 100%.  When the public thinks there is uncertainty in the science, even .01%, they can engage our remarkable ability at willful blindness to clutch onto a more comforting conclusion (my religion hasn't been lying all these years, there is no such thing as evolution).  It is actually better that the media usually report science as if it were 100%.  Imagine if we weren't totally convinced about gravity.  Or if we still believed in the chance the earth was flat. 

But when a theory turns out to be wrong, all of science immediately becomes suspect in the public's mind.  Wait - you have been lying to us all these years?  Anti-depressants don't work any better than a placebo?  Mammograms are not beneficial to 40-something women with no family history of breast cancer?  So I guess I don't have to give up my Hummer because global warming could be false too.  Maybe smoking doesn't increase the risk of lung cancer.