NationStates

A simpler, what-you-see-is-what-you-get version of the model. It assumes current polls reflect the best forecast for November, although with a lot of uncertainty. Polls-plus combines polls with an economic index; polls-only does not. Polls-plus will include a convention bounce adjustment; polls-only will not. Polls-plus starts by assuming that likely voter polls are better for Republicans; polls-only makes no such assumption. Both models revise this assumption as more data becomes available. Polls-plus subtracts points from third-party candidates early in the race, while polls-only does not. Both models employ a regression that is based on demographics and past voting history. But polls-only weights the regression less and places less emphasis on past voting history. Polls-only accounts for more uncertainty than polls-plus. Polls-plus and polls-only will tend to converge as the election approaches. The polls-only and polls-plus models discount polls taken just after the conventions, whereas the now-cast will work to quickly capture the convention bounce.

All versions of the model proceed through four major steps:

Step 1: Collect, weight and average polls.
Step 2: Adjust polls.
Step 3: Combine polls with demographic and (in the case of polls-plus) economic data.
Step 4: Account for uncertainty and simulate the election thousands of times.

Almost all state and national polls are included. If you don’t see a poll, it’s for one of these reasons:
The poll is very new, and we haven’t had a chance to add it yet. The poll was conducted directly on behalf of one of the presidential campaigns or an affiliated PAC or super PAC. The pollster is on FiveThirtyEight’s banned pollster list (pollsters we think may be faking data or engaging in other blatantly unethical conduct). Sometimes, there are multiple versions of a poll. For example, results are listed among both likely voters and registered voters. We prioritize polls as follows: likely voters > registered voters > all adults. If there are versions with and without Gary Johnson, we use the version with Johnson.

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

FiveThirtyEight’s pollster ratings, which are based on a firm’s track record and methodological standards. Sample size. A larger sample helps, but there are diminishing returns. Recency, which is less important early in the campaign and becomes more important later on. When a firm polls a state multiple times, the most recent poll gets more weight, but the older polls aren’t discarded entirely. The trend line adjustment (see Step 2) also helps to put more emphasis on recent data.

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 (in only the polls-plus model)
Omitted third-party candidate adjustment
Trend line adjustment
House effects adjustment

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.

Strategy 1: Pure PVI. This works by taking the national popular vote estimate and adding a state’s PVI to it. For example, if Clinton’s up 4 percentage points nationally, and a state’s PVI is Democrat +5 percent, she’d be projected to win the state by 9 percentage points.

Strategy 2: Regional regression. The adjusted polling average in each state is regressed on PVI and on dummy variables indicating which major region the state is in (Northeast, South, Midwest, West). Historically, this is the most effective approach. It can capture major changes in voting patterns from one election to the next, but doesn’t suffer from much overfitting. Regions are based on a combination of Census Bureau regions and political regions, as defined by FiveThirtyEight. Where they differ, a state is considered split between two regions. For example, Maryland is considered half Southern and half Northeastern by the model.

Strategy 3: Demographic regression. We regress the adjusted polling average in each state on PVI and several other variables, mostly related to race and religion, that are pertinent in this year’s election. (This includes the percentages of voters who are black, Hispanic, Asian, non-Christian, evangelical Christian, Mormon and college graduates. It also includes an economic index for each state, showing change over the last 12 months.)

The regression employs a technique designed to remove spurious variables. Still, it’s vulnerable to some degree of overfitting. That means the regional regression gets the most weight. Polls-only and now-cast do not use pure PVI, and instead use a 70/30 blend of strategies 2 and 3. So far, our regressions suggest that the electorate is slightly less polarized than in 2008 or 2012. Red states aren’t quite as red, and blue states aren’t quite as blue.

The adjusted polling average in each state is combined with the regression. The regression estimate gets more weight early in the race and when there’s less polling. The regression gets 100 percent of the weight when there’s no polling in a state. The polling average can get as much as 95 percent of the weight late in the race in a state with abundant polling. Polls-only and now-cast give slightly less weight to the regression than polls-plus does. As a final step, the regression is recalibrated so that the overall national popular vote is unchanged. If a candidate gains ground in one state because of the regression, the model will necessarily have her lose ground in another. In other words, the purpose of the regression models is not to say the country’s demographics inherently favor Trump or Clinton. Instead, it’s to create a more realistic distribution of the projected vote across each state, especially in states with limited polling. We don’t want to have Clinton winning Kansas based on a single poll there, for instance, while she’s badly losing Nebraska.

We usually run at least 20,000 simulations for each version of our model each day. That’s a lot, but it still produces a small amount of sampling error. You shouldn’t worry too much when win probabilities change by less than a percentage point. We simulate the vote by congressional district in Maine and Nebraska, which award one electoral vote to the winner of each district and two electoral votes statewide. Where available, district-level polling is used in these forecasts. If no candidate receives a majority of electoral votes, the model assigns the election to Trump half the time (because Republicans are very likely to control a majority of congressional delegations if the election is close) and to the winner of the popular vote the other half of the time

[/url]