CommunityElections

A Prediction About Election Predictions

A week out from the election I have a prediction about the professional election predictors. They are going to be accurate because our federal elections have become extremely predictable. If you know absolutely nothing about the Senate races except for the results of the 2012 Presidential election, you could end up predicting between 80-97 percent of all Senate races correctly this year.

Based on just this information, you would predict the Republicans would end up with 51 seats. The only race you would likely be off is Maine, where in a small state the surprisingly popular long time incumbent Republican Senator Susan Collins substantially bucks the trend. It is also possible this analysis might end up wrong in Colorado, Iowa, North Carolina, Georgia and Kansas. Given that CO, NC, and IA were all presidential swing states in 2012 and Kansas is the rare example of an election without a Democrat, that is remarkably accurate. Overall, this basic analysis will get the vast majority of seats correct.

Look at the results of a recent single pollster predicting only a marginal improvement over this basic prediction that could have been made over two years ago. The only states where the Yougov polls diverges from the 2012 Presidential results are in Maine, North Carolina, and sort of Iowa where the race is tied.

A basic polling average, like RealClearPolitics’, is likely better but still produces only very minor refinement. Its results diverge from what is predicted by the 2012 election only in Maine, North Carolina, Colorado, Iowa, and Kansas.

Finally, even though 538 includes some extra data and calculations in their analysis based on “fundamentals,” their overall prediction is effectively identical to RCP. They only diverge from the 2012 election in Maine, North Carolina, Colorado, Iowa, Kansas, and sort of Georgia which they have at 50/50.

2012 Presidential Difference YouGov Difference RCP Difference 538
Democratic CO, DE, HI, IA, IL, MA, ME, MI, MN, NH, NJ, NM, OR, RI, VA NC NC, KS NC, KS, GA (tied)
Republican AK, AL, AR, GA, ID, KS, KY, LA, MS, MT, NC, NE, OK, OK, SC, SC, SD, TN, TX, WV, WY ME, IA (tied) ME, CO, IA, ME, CO, IA,

The fact that all the election predictors are likely to be proved accurate on Tuesday sadly says less about their skills and more about the state of our democracy. Both parties have worked together to create election rules that overwhelmingly favor only two political parties, leaving us with only a few truly competitive elections and very limited democratic accountability.

CommunityFDL Main Blog

A Prediction About Election Predictions

A week out from the election I have a prediction about the professional election predictors. They are going to be very accurate because our federal elections have become extremely predictable. If you know absolutely nothing about the Senate races except for the result of the 2012 Presidential election, you could end up predicting between 80-97 percent of all Senate races correctly this year.

Based on just this information, you would predict the Republicans would end up with 51 seats. The only race you would likely be really off is Maine, where in a small state the surprisingly popular long time incumbent Republican Senator Susan Collins substantially bucks the trend. It is also possible this analysis might end up wrong in Colorado, Iowa, North Carolina, Georgia and Kansas. Given that CO, NC, and IA were all presidential swing states in 2012 and Kansas is the rare example of an election without a Democrat, that is remarkably accurate. Overall, this super basic analysis will get the vast majority of seats right.

Look at the results of just a recent single pollster predicts only a marginal improvement over this incredible basic prediction that could have been made over two years ago. The only states where the Yougov polls diverges from the 2012 Presidential results are in Maine, North Carolina, and sort of Iowa where the race is tied.

A basic polling average, like RealClearPolitics’, it is likely better but still produces only very minor refinement. Its results diverges from what is predicted by the 2012 election only in Maine, North Carolina, Colorado, Iowa, and Kansas.

Finally, even though 538 includes some extra data and calculations in their analysis based on “fundamentals,” their overall prediction is effectively identical to RCP. They only diverge from the 2012 election in Maine, North Carolina, Colorado, Iowa, Kansas and sort of Georgia which they have at 50/50.

2012 Presidential Difference YouGov Difference RCP Difference 538
Democratic CO, DE, HI, IA, IL, MA, ME, MI, MN, NH, NJ, NM, OR, RI, VA NC NC, KS NC, KS, GA (tied)
Republican AK, AL, AR, GA, ID, KS, KY, LA, MS, MT, NC, NE, OK, OK, SC, SC, SD, TN, TX, WV, WY ME, IA (tied) ME, CO, IA, ME, CO, IA,

The fact that all the election predictors are likely going to be proved very accurate on Tuesday sadly says less about their skills and more about the state of our democracy. Both parties have worked together to create election rules that overwhelmingly favor only two political parties leaving us with an extremely small number of truly competitive elections and very limited democratic accountability.

 

Previous post

Bob Dold, Bruce Rauner, Rep. Paul Ryan all trying to run to the Left on the issues?

Next post

A Prediction About Election Predictions

Jon Walker

Jon Walker

Jonathan Walker grew up in New Jersey. He graduated from Wesleyan University in 2006. He is an expert on politics, health care and drug policy. He is also the author of After Legalization and Cobalt Slave, and a Futurist writer at http://pendinghorizon.com

18 Comments