There are two major factors that drive the reporting of results on election night.  First, the U.S. is one of the few countries with a significant East-West width.  This fact, combined with state autonomy, means that, unlike a Germany or United Kingdom, we have staggered poll closing times (ranging from 6 p.m. EST in parts of Kentucky and Indiana to 1 a.m. EST in parts of Alaska).  Second, even with recent improvement in vote counting technology, there is (even with the same state) delays in reporting results that lead to precinct results being released throughout the evening due to:  1) processing all the people who were in line to vote at the official poll closing time; 2) getting the electronic vote counting devices from the individual precincts to the county/parish/township counting center; 3) downloading all of those devices into the counting center’s computer (obviously more precincts in urban counties = longer to download all of the data); and 4) reporting those results to the media and the state election authority.

Given that it takes hours to get near full counts (and days or weeks to get full counts), the news media uses “cheats” to project races as early as possible.  The two main cheats are somewhat related.  First, at least for state-wide races in state’s expected to be crucial, the media conducts exit polls at key precincts.  (These precincts are chosen to provide enough of all key demographic groups based on past voting history, along with weighting formulas based on past history adjusted by reweighting based on actual turnout.)  Second, the media relies on past history as far as how the parties have performed in counties and precincts in the past.  (The media has the advantage of having all of the relevant data pre-digested.)  For both “cheats,”   the question is how the early reporting precincts differ from what is expected.  If the exit polls show the Republicans “underperforming” in rural precincts by three percent, and the early precincts show a similar result in those precincts those results “confirms” that the exit polls are close.  Similarly, in a D+5 state, if the early results show that Democratic candidate is doing 5% better than the norm for those precincts in that state, that is a pretty good sign that the Democratic candidate is going to win.   Because most average people lack the media’s ease of access to this data, we are sort of in the position of having to reverse engineer things.

For the most part, there is no need to pay close attention before 9:00 p.m. EST.  Nine states (ten if you count Florida which is mostly closed at 7:00 p.m. EST) are closed before 8:00 p.m. EST.  And, for the reason noted above, it takes about an hour before a decent share of precincts start reporting.  (In some states, early vote results get released pretty quickly after the polls close, but you still need enough time to get a concept of how many people voted on election day and how much election day results seem to differ from early voting).  The 2016 election gave us a good clue on what we should be looking for — particularly given that we are looking at 435 individual house districts, 35 Senate seats, and 36 governor’s races.  In 2016, at the start of the evening, there were a significant number of states that were close enough that the media waited before calling.  However, as the evening progressed, the lean Republican states were being called for Trump while the lean Democratic states stayed to close to call. Continue Reading...