FiveThirtyEight

We’ve just launched FiveThirtyEight’s 2016 general election forecast, which projects how the 538 Electoral College votes could break down in the presidential election. The forecast will be continually updated through Election Day on Nov. 8. Here’s a bullet-point-style look at how it was built.


What’s new in the model since 2012?

Major themes and findings

Three versions of the model

Differences between polls-plus and polls-only

Differences between polls-only and now-cast

Four major steps

All versions of the model proceed through four major steps:


Step 1: Collect polls

Almost all state and national polls are included. If you don’t see a poll, it’s for one of these reasons:

banned pollster list

Which poll version do we use?

Sometimes, there are multiple versions of a poll. For example, results are listed among both likely voters and registered voters.

Calculating a weighted average

We calculate a weighted average in each state, where poll weights are based on three factors:

pollster ratings

Step 2: Adjust polls


There are five adjustments, listed here in the order in which the model applies them. (The trend line and house effects adjustments are generally the most important ones.)

Likely voter adjustment

Convention bounce adjustment

Omitted third-party adjustment

Trend line adjustment

loess regression

House effects adjustment


Step 3: Combine polls with other data

forecasting what will happen in the Electoral College

Adjusting the third-party vote

Allocating undecided voters

Projecting the national popular vote

National polls versus state polls

To recap, the model mostly uses state polls. But national polls can influence the forecast in some subtle ways:

Partisan voter index (PVI)

used by Cook Political Report

Calculating demographic regressions

Instead of using one regression model, we take three strategies, which range from more simple to complex, and blend them together. The reason for this is that the more complex methods (especially strategy 3) are subject to potential overfitting. Hedging the complicated methods with simpler methods produces a better result.

Census Bureau regionspolitical regions

Blending polls and regression

Calculating the economic index

2012nonfarm payrollsindustrial productionreal personal incomepersonal consumption expendituresconsumer price indexS&P 500

The “fundamentals” forecast

treats this factor as neutral

Blending polls and fundamentals

Lauderdale and Linzer

State elasticity scores

elasticCooperative Congressional Election Study

Step 4: Simulate the election

Three types of error

Each simulation accounts for three potential types of error and uncertainty:

National error

Demographic and regional error

State-specific error


Odds and ends

That’s basically it! But we’ll conclude with a few odds and ends:

Our distributions have fat tails

t-distribution

Handling the third-party vote

beta distribution

Third- and fourth-party ballot access

ballot access deadline

Tipping-point chance and voter power index

Errors and omissions?

the modelpolls@fivethirtyeight.com

Filed under