As predicted by Nate Sliver and his 538 blog, the President was re-elected by a comfortable margin in the Electoral College. Obama's 2.2% margin over Romney in the popular vote was sufficient for at least 303 EC votes, and probably 332, because his campaign's strategy focused on that metric.
I'm a long-term, diehard 538 blog fan, so I pretty much knew how things would turn yesterday. But I also knew there was uncertainty in the validity of polling methodology, and therefore was not terribly confident or comfortable until the reporting precincts in Ohio creeped upwards of 70%. In retrospect, I should have been assured in the outcome.
Mr. Silver's model was nearly perfect, calling all 50 states and missing the popular vote by fractions of a percent. It's interesting to think about how the success of Nate's poll aggregation, statistical model approach will impact future elections. Surely it will result in more personal fame and fortune for Nate Silver, but what will it mean for the media's reporting on future elections and, indeed, voter behavior?
Will TV's so-called news journalists get away with claiming an election is a "razor tight" toss-up when evidence shows a clear lead for one candidate? Will partisan pundits be able to claim their candidate is leading by cherry-picking polls that favor them, or will program hosts point out the invalidity of that approach?
How will it affect voters' behavior when they are able to "see into the future" by peeking at a website or mobile app? Will the leading candidate's followers become complacent, or energized? Will the underdog's troops be dispirited, or will they rise to the challenge? Does a proven forecasting model alter the ability to forecast?
To the latter questions, I expect there will be no measurable change in voter behavior. Aggregated polls, processed intelligently, will be able to accurately predict most future elections as they did this time. Unless contests are exceptionally close, suspense will not be part of our election experience.
To the former questions, I'm uncertain how different media coverage will be. I do anticipate a plethora of "polls-of-polls" to be up and running by the next midterm election. There is only one Nate Silver, so all but the highest bidding cable or broadcast station will be in need of their own polling model guru. (But since Nate is the only one who's proven himself, most will tweak their models to match his results.) Most stations will move toward a stance of predicting a winner, versus their current tendency to label races as "too close to call".
In the old paradigm, it was in everyone's interest for races to be suspenseful. TV programs need drama to build and hold their audience. Winning politicians want to keep their followers engaged and energized. Losers don't want their voters to be dispirited and disengaged. So a "razor-tight" race is a win-win-win.
What will it be like when the question is not, "who will win Ohio", but, "at what point in the ballot counting will the expected victory be verified"?
Congratulations to Nate Sliver. Last night, you changed election forecasting – and reporting – forever.
No comments:
Post a Comment