After over one year of hate-filled rants from Donald Trump, the fiasco that was the Republican convention in Cleveland, the on-going scandals involving Donald Trump, Trump’s refusal to disclose his taxes, three presidential debates, and recent Republican threats to throw a tantrum for the next two to four years if they don’t win, there is little more that can be said about why the only choice in this election is to vote for the Democratic ticket. The continued loss of rationality and respect for facts in the Republican Party is a long-term problem that needs to be fixed because democracy requires, at least, two viable alternatives to work. But this year, the choice is clear. Even if you think that a Democratic candidate for a particular office is less than perfect, those candidates are still way better than what the Republican Party is offering. While there is still more to be done over the next three days to get every Democratic voter to the polls, Tuesday night is now looming ever closer. So, for the next several days, some thoughts about what to look for on Tuesday night. While the remaining posts in this series will take a chronological look at Tuesday night, this post is more about the basics and the mechanics.
For the media, there are two main tools for calling the election. While these tools have changed slightly over time, the fundamentals have basically stayed the same. The first tool is the “exit” poll. The second tool is the unofficial vote count.
Exit polls are just what the name suggests — individuals outside selected precincts taking a random sample of voters as they leave the polls to find out who they voted for and why. The groups doing the polling — at one time there was a unified poll conducted by all major media, but the various networks have gone back to doing their own polls — have identified certain precincts that are representative of various voting blocks in the state thereby giving them a chance to figure from turnout and responses at those precincts the likely vote in the state. The better efforts — recognizing that almost half of the voters in a state actually cast early votes — combine the exit poll with a poll of those who cast early votes so that the sample includes both those who voted on election day and those who voted before election day.
Because they are polls, exit polls suffer from some of the same problems as pre-election polls. While generally accurate, they can occasionally be off due to sampling error. Additionally, even without gross sampling error, some slight differences are to be expected. In other words, exit polls (particular when they include early voters) can indicate whether a particular race in a particular state is likely to be close or a landslide. When the exit polls suggest a close race, the media are going to want to wait and see a larger segment of precincts counted before projecting a winner. When the exit polls suggest a landslide, the media are likely to project a winner with few precincts counted. (I use the term project because that is all that the media actually does. The media does not declare a winner. The winner is only declared when the official results are released.)
The other half of the media equation is the unofficial vote count. The election night count is an “unofficial” count for a variety of reasons. First, it is incomplete. Depending on state law, there will be provisional ballots cast on election day and some absentee ballots — even though mailed before election day — will not arrive until several days later. Additionally, depending on the numbers of provisional and absentee ballots that arrive at the election authority, it may take days or weeks to verify which ballots should be counted and to count them. Second, the election night count, in some states, is tentative. These states use a hand count of selected precincts after the election to assure that the machine count on election night was accurate. If that hand count shows a problem, the hand count is expanded to cover all races and all precincts. For both reasons, the count on election night is likely to change before the final results are certified by state election authority.
With these caveats, however, the tentative counts are useful in determining when a race can be called. While it is possible for the average person to get an idea of the state of the race from the count, most networks rely on past experiences and data bases. Even more so in recent years as voters have become even more geographically sorted, precincts and counties have a history of voting a certain way. Election experts expect a certain county (or for larger counties, a certain part of the county) to vote a certain way. Before things got even more precise, this tendency helped campaigns choose where to focus their get-out-the-vote efforts. Today, it tells the media whether the race is going to be close. Simply put, data from past elections create rough models of how candidates should be performing in various locations in a close race, in a landslide loss, and in a landslide win. If the exit polls are forecasting an easy win for the Democrats in a given state (say Delaware) and the first couple of precincts show that Trump is underperforming the typical Republican results in Wilmington and in rural parts of the state, it is easy to project Delaware with only a tiny percent of the vote in. On the other hand, if the exit polls in North Carolina show a close race and the first couple of precincts show results in line with past close races (regardless of which parts of the state come in first), the media knows that they have to wait for more to project North Carolina.
For the states that exit polls and the first couple of precincts reflect a close race, it becomes easier for the average person to figure out who is going to win as the votes come in. Even if your state does not post county-by-county (or township by township) results, the major networks do. After around 40% or so of the vote is in, it is possible to get a rough estimate of who is winning in each county and how many votes are outstanding from those counties. (The media may have slightly finer tools for figuring out which precincts are missing and how those precincts differ from the precincts reported.) Based on those numbers (and knowing the size and location of the counties that have yet to report any results), it is possible to do a scratchpad estimate of how many votes are outstanding for each candidate. For example, if one candidate is leading by 500 votes after 50% of the votes are in from a given county, a rough estimate of the final result is that the candidate will pick up another 500 votes from that county. While it is possible that the rough estimate is off, this rough estimate is a good way of telling if there are really enough votes still outstanding in favorable parts of the state for the trailing candidate to win. In some cases (where the early reporting precincts are overwhelmingly from the part of the state that favors the leading candidate), it is even possible to tell that the trailing candidate is likely to win when the remaining precincts come in. Of course, the ability to do this type of estimate improves as fewer precincts are left. Depending on the state and the closeness of the race, it is somewhere at the 75+% of the vote counted that the “remaining” precinct method tells whether it is necessary to wait for more results or if it is possible to call the race because the gap is too large for the trailing candidate to recover.
Of course, the goal for now is to avoid the need for voters in the key state to wait until 2 in the morning to find out who won. That requires getting every democratic voter to the polls. So good luck in your state over the next 72 hours.