Today is the official start of the final stretch of the November 2014 campaign season which makes it a good day to take a peak out our status in polls and forecasts.
Josh Katz,
Wilson Andrews, and
Jeremy Bowers, of the
New York Times publish
Elections 2014: Make Your Own Senate Forecast. The
NYT Leo Statistical Election Forecasting Tool base case predicts Republicans have a 66% change of gaining a majority in the Senate, with the most likely outcome being that Republicans win 51 seats and Democrats 49.
I do not know what the percentages are by the states yet. They appear to be the confidence that the state is in the correct category, not vote margins.
Leo predicts Democrats will win NH84%, MI 75%.
Toss ups are CO57%, IA53%, AK52%, NC51%, LA60%
States expected to go to the Republicans are AR66%, GA82%, KY88%, WV99%
The NYT makes all of their assumptions and inputs available to us to change and invites us to make own assumptions and offers a special award to whomever makes the best prediction.
Below, we let you tweak Leo’s settings, explore the effect of different assumptions and create your own forecast. You can submit one set of predictions or a new forecast every day — or anything in between. After the elections are over, we’ll award the best forecaster the coveted Upshot Cup for Excellence in Election Forecasting.
The Upshot’s Senate forecasting model uses a combination of polling data and a fundamentals model that includes — among other things — candidate political experience, money raised, and state-level voting history. The poll aggregates include adjustments for house effects and a weighting scheme that weights more recent polls more heavily. “Nontraditional polls” include polls from organizations that use nontraditional sampling methods or are not conducted using live phone interviewers. ...
Entries will be scored by taking the product of the probabilities that the reader assigned to the actual outcome in each state. The entry with the highest score will win the Upshot Cup. In the event of a tie, the tiebreaker will be the entry that was received earliest.
It looks as if the NYT is making a play to become a hotspot of election polling and forecasting with this interesting new tool. I have not had a chance to study it closely so offer it here as a conversation starter on what we can do to get ourselves fired up for this important November election.
Sadly, Democratic voters have a poor record for turning out for off-cycle elections. I believe somewhere I read Republicans may have a 14% greater "participation rate" for these non-presidential elections.
What can we do to make this statistic and assumption obsolete? We have to find ways to fire up our base, which theoretically should be easy given the advantages we should have on issues.
How to do we connect with our voters and get them to the polls in November?