Behind the model

How we calculated how much your vote counts…

In How much does your vote count? we built a tool that helps you understand the power of your vote based on your age, gender, race or ethnicity and where you live. If you are wondering how the model works, here’s how we calculated voter power.

The president isn’t directly elected by the people. Instead, each state gets a certain number of electoral votes: one for each of that state’s Senators and Representatives. (Washington, DC also gets three votes.) Almost every state awards all their votes to the winning candidate, whether they win by millions of votes or just a few hundred. Nebraska and Maine award a vote to the winner of each congressional district, but those contests are similarly winner-take-all.

So to understand the power of an individual vote, we needed a way to put a number on the power of voters in each state. As part of their election model, Fivethirtyeight.com calculates a Voter Power Index, which they define as “the relative likelihood that an individual voter in a state will determine the Electoral College winner.” The index is based on polling data and fluctuates as the race heats up.

The actual probabilities that a single vote will decide the election are vanishingly small. For your state, it’s the probability that the electoral college will be deadlocked without your state’s electors, and that the voting in your state will be exactly tied but for one tie-breaking ballot, and —most importantly— that that ballot will be yours. In this way, Fivethirtyeight.com’s Power Index takes into account not only up-to-date polling data, but how many electoral votes each state has up for grabs. When you measure this value for each state, you get a relative sense of the influence of each state in the election as it stands now.

Once we knew the power of a vote in each state, we used demographic data for each state to estimate the electoral power of America’s voters by racial and ethnic groups.

And then we made one more adjustment: The electoral college is not the only thing that affects the power of a citizen’s voice. The only votes that are counted are those that are cast, so we looked at how turnout varies by race, age and gender in each state, and we added these to the model. (The racial and ethnic breakdowns in our model mirror those used in the Census Bureau’s report on 2012 turnout)

We’re defining turnout as the percentage of adult citizens who cast a vote, so this number takes into account all of the reasons people might not vote: not being registered, being a felon unable to register, not possessing sufficent identification in a state that has tightened restrictions, not being able to get time off to vote, or simply opting not to vote.

Relative turnout rates for age and gender are fairly consistent nationally. No matter what the state’s average turnout rate, women tend to turn out at slightly higher rates than men. Likewise older voters tend to turn out at consistently higher rates. Because these effects are consistent from state to state, we used a national average rather than state-by-state turnout rates for gender and for age.

It’s also important to note, as mentioned earlier, that the model assumes that turnout rates in 2016 will be the same or similar to turnout rates in 2012. However, it is likely that they may be different and the results of the model are estimates only.

The Census Bureau does not provide historic turnout rates by race in states where the population of certain groups is too small to be statistically significant. In those places, which include DC, and many of the other lower power states, we used national turnout averages as a proxy to estimate the turnout for each group.