Search This Blog

Divided We Stand

Divided We Stand
New book about the 2020 election.

Tuesday, April 17, 2012

"Rules" of Election Forecasting

Andrew Kohut writes at The New York Times:
With the focus now fully on the campaign between Barack Obama and Mitt Romney, commentary about the issues, the voting blocs and the strategies of the two campaigns dominates political discourse. But having polled on the last 10 presidential elections, I’m struck by a meta-question about each candidate’s viability that may trump all else. Obama and Romney both carry so much political baggage that one or the other will have to defy modern political history to win in November.
For Mitt Romney, the history buster would be that no candidate in the modern polling era with personal favorability ratings as low as his has ever won the presidency. With a 29 percent rating, he is far and away the least personally popular presidential candidate since pollsters have been regularly checking favorability ratings.
Barack Obama is much more personally popular than Romney, but his problem is that no incumbent president has ever won re-election with unemployment rates as high as they are likely to be in November. The modest decline in the unemployment rate in the Labor Department’s March report – to 8.2 percent — adds to the challenge he faces. In recent times, only Ronald Reagan struggled with a comparably high unemployment rate (8 percent) at the start of his re-election year. However, it declined steadily over the course of 1984 and it stood at 7.2 percent by Election Day.
Also at the Times, Nate Silver offers a useful reminder:
A final and more general point is that there have been only 16 presidential elections since World War II. That simply isn’t a lot of data, and overly specific conclusions from them, like “no recent president has been re-elected with an unemployment rate over 8.0 percent” or “no recent incumbent has lost when he did not face a primary challenge,” are often not very meaningful in practice and will generally not carry much predictive weight.