Saturday, November 12, 2016

Death and election results

Many of my favorite people are in shock over the election results.  It is not unusual for some of them to tell me that I usually seem intelligent but I must too thick to grasp the meaning and significance of the way things turned out.  Why aren't I just sick, too?

I looked at FiveThirtyEight at least once a day for a week or two.  I got the idea that Nate Silver and his crew were at the forefront of what they do and I put my effort at foreseeing what was going to happen into following their statement.  Their chart showed Clinton with 60 to 70 points and Trump with 40 to 30.  They did have a chart of the point history and it showed that the two candidates drew very close to each other, within a few points, repeatedly but that over time, Clinton kept pulling back ahead.

I taught statistics for several decades and often had the job of explaining concepts of probability and risk.  The entire subject of statistical analysis for risk assessment and forecasting is about 100 years old.  Many thinkers have divided the history of mathematical prediction and forecasting into two parts: Anglo-American and Bayesian.  The Anglos took the route that a small enough probability can be ignored.  That brought up the question How small is small enough to be ignored?

Let's say that I realize that I could die today.  If I do die, I am going to be cremated and for that, I will be undressed.  So, if the likelihood is that I die today, I might want to save the effort of getting dressed, only to be undressed later.  But I don't feel as though I am near death, and given my intentions for the day, I am going to want to be dressed for social, legal and thermal reasons.  How likely is it that I am going to die?  The originators of the line of thinking called Anglo-American statistics decided to use 5%.  They thought that if the best evaluation of my dying today was 5% or less, I should go ahead and get dressed.  

What I have written in this last paragraph above has been examined very carefully by many thinkers and investigators.  Many of them concluded that Reverend Thomas Bayes (1702-1761) had an idea that could be used to try a different approach, which is basically to keep investigating and examining and gathering data until you decide to stop.

There is a line in "Oh God", starring John Denver and George Burns, in which God (Burns) informs the hero (Denver) that he, God, knows the past but that he does not know the future until it becomes past.  The degree of shock and surprise at the election results, no matter who you supported, is a measure of how little you and I, mere mortals aimed at eventual death, can foretell the future.  The critics of the Anglo-American approach say," Are you kidding?  A 5% chance can't be ignored. You will be wrong on average 1 time in every 20."  If you don't stay with averages, you could have a bad run where you are wrong, wrong and wrong again.  

Thus I say, keep your eyes open, stand up for what you believe, support our best principles and remember that you don't know, you don't know,you don't know what is going to happen.  Based on past experience, some of what happens will be terrible but some will be better than we ever thought possible.

Popular Posts

Follow @olderkirby