November 7, 2012

The Return of the Oracle of Delphi

Posted in Social Commentary tagged , , , , at 11:00 am by Doug Brockway

In Ancient Greece, well through the Roman Empire, leaders from around the Mediterranean World would travel to Delphi for a private session with the Oracle who would predict their futures.

In 21st Century America the oracle is shadowed by the pollsters and none bears a stronger social import than Nate Silver, he of the 538 Blog.  I happen to think the world of his work and his predictive skills.  That said, it seems to me the public has a very shallow understanding of what he does and what it means.

On the Monday before the election I listened to Ron Insana of CNBC dismiss Silver as “averaging” polls.  He doesn’t.  He uses poll data , along with other data about past elections, their polls by day, and the history of those polls as predictors by year, as input to a model that does calculations and runs scenarios showing how the competing candidates can reach the needed 270 electoral college votes.  That Insana, or Hannity or any number of other people confuse a weighted, historical model with an average suggests they are incompetent.

Silver never once says that any given scenario is the one we’ll experience when we actually votes.  He merely says how many scenarios there are, of what historical relevance, that suggest one candidate or another will succeed.  He’s suggesting the odds that one person or the other will win.  He’s also clear, the actual vote is all.

There’s a lot of mis-understanding and, by default mis-reporting of statistical information in campaign coverage.  During the return coverage I saw a report that said since the democrats had won senate seats in ME and CT that “makes it harder for the republicans to win an East Coast senate seat.”  If you’re going to use stats, take a stats class…. 1st, there is no causality one state to another in the same way a coin flip is 50/50 each time regardless of past flips.  2nd, some states report earlier than others, which does not cause the later REPORTING states to have voted one way or the other.  The analysts and reporters are confusing the fact that they had conclusive return data from CT and ME before VA’s returns were in with a causal relationship from one state to another.

But, people believe that because the Oracle Speaks the Oracle drives events.  This morning, on Facebook, I saw the following comments re: Silver. “If Obama wins, i’m buying Nate Silver’s book (not just taking it out of the library. THAT guy’s my hero.”  Fair enough.  Buy the book.  He didn’t CAUSE Obama to win.  He merely indicated how likely it was based on what the polls were saying.  Along the same lines in the same stream I saw, ” He deserves a Nobel, Pulitzer and Barrington Library Teen Room prize.”  Maybe the Pulitzer.  I dunno’ from the Barrington Library.  There’s no Nobel for statistical analysis.

There’s other nuttiness.  The president was shown by one set of analysts to have certain advantages regarding the voting trends for women, for women of a certain age, for women who are unmarried.  I saw nothing about for women who prefer Pinochle over Cribbage.  But, the reporting is of the “women are telling us” ilk.  They aren’t, any more than corporations are people with feelings who like to dance on Saturday night.  The aggregation of the actions of women have an impact and represent a trend.  Inside that aggregation are many points of view almost none of which, individually, match up to political analysts pigeon holes.

I happen to be happy that Obama won and that some real clowns did not win senate seats for the Tea Party.  But the Oracle didn’t make it happen.  Voters with ballots did.