Tonight, as the election returns come in, I’ll be focused the most on three Senate races: Missouri, Arizona, and Nevada. These are the three contests that seem to be true toss-ups. Indiana appears to lean towards the incumbent Democrat, but if so, only barely. Thus, I’ll keep an eye on that race too.
If the Democrats sweep the three toss-up races (and win in Indiana), this will mean a net pick-up of two seats from these races. That pick-up will be offset partially if, as expected, the Republicans pick up a seat in North Dakota. Thus, if all other races go as expected, a Democratic sweep of the toss-up races would leave Republicans in control of the Senate, but only by virtue of the vice president’s tiebreaker vote.
If Republicans win all three toss-up races, then, other things being equal, they will extend their margin to 53-47. Throw in Indiana, and the count will be 54-46.
In theory, it’s unlikely that either party will win all of the toss-ups. The chances of winning three toss-ups are 1 in 8.
However, this assumes each toss-up race is an independent event — i.e., that the outcome of one race has no relationship to the outcome of the others. That’s true if one is tossing coins. It’s not a fair assumption in political races.
The label toss-up in the political context is earned by polling data showing the candidates neck-and-neck. Projections based on polling data depend on assumptions about turnout. If one party turns out more of its voters than expected, the race tilts its way and no longer is a toss-up (so too if the polling process systematically reaches too many voters who lean one way or if respondents are lying to pollsters because they want to hide a preference for one particular party).
In a given election year, turnout in one state is not independent of turnout in other states. Thus, for example, if anti-Trumpism is driving an unexpected turnout advantage for Democrats in Missouri, it probably is doing something similar in Arizona and Nevada (and Arizona and Nevada have demographic similarities). By the same token, if Trump’s mass rallies confer an unexpected advantage on Republican candidates in one state he’s visited, they probably do so in others.
We’ve seen elections in which the close races break overwhelmingly in favor of one party. The midterms in 1986 and 2006 are examples, if I recall correctly.
In a true wave election, the tide pulls in races that weren’t thought to be toss-ups. 1980 is a good example. An old friend recalled watching those election returns with a Republican incumbent who was stunned by the GOP wave and eager to find out who the heck Paula Hawkins was.
If 2018 is a wave election for the Democrats, they might well hold the North Dakota seat and could pick up Tennessee and maybe even Texas. Thus, we could wake up tomorrow and find the Dems headed for control of the Senate in January.
If 2018 is a wave year for Republicans, Florida, Montana, and maybe West Virginia become probable pick-ups. Several Midwest/Upper Midwest seats might also come into play.
Given my prediction record in recent years, I have no business forecasting today’s outcome. For what it’s worth, though, it looks to me like the Republicans will either hold their current Senate majority or extend it by one. I think the over-under in the Senate is Republicans +0.5 seats.