Thursday, July 10, 2014

Polling in Governor and Senate Races

Polls 16 weeks before an election and 14 weeks before early voting do not predict an election outcome. And, the number of recent polls in Colorado, especially the governor’s race, are few, but using the Real Clear Politics average show Senator Udall up by one percent and Governor Hickenlooper up by four percent.

The forecasting power of polls does tend to increase with the size of the lead and time before voting. In both of Colorado’s major races, using the New York Times polling model, the conventional view is affirmed. Udall has a micro advantage in a race too close to call (55% likely to win) and Hickenlooper still has the advantage (89% likely to win, which means that with a 4% lead 16 weeks out, a candidate has an 89% chance to win). Of course, the most recent poll that places Hickenlooper in a dead even race raises the possibility that the next round of polls could change the forecast.

The New York Times forecasting model has control of the U.S. Senate in a near statistical tie, with the Republicans having a 54 percent chance of winning at least six seats necessary for control.

The major forecasts tend to show the Colorado Senate seat a toss-up, but at least two – Rothenberg and Sabato – still have it leaning Democrat.
Forecasting began last fall and early this year, primarily examining historical election data and national polling results, such as President Obama’s approval. This spring pundits have added whatever local poll data was available. Some also rate the finances of each candidate and some indicators of campaign and candidate quality. Cook and Rothenberg tend to be handicappers and examine each race individually, with subjective judgment involved. Sabato both handicapping and forecasts; i.e., uses an algorithm.

No comments: