Now that we're only a few days away from the first actual voting that will take place in the 2016 presidential contest, we're at a good point to stop and think through the possible outcomes that might unfold in the first few bouts, what they might mean for the eventual outcome, and what evidence they might provide to political scientists who geekily see this as a nationwide exercise in hypothesis testing with lots of media coverage.
The Iowa caucuses will be held on Monday, February 1, and for the first time since Americans started getting riled up about this election (in 2013), actual breathing voters will cast actual recorded preferences for candidates. And it will mean something. But what will it mean?
There has been a lot of chatter about how polls may not matter that much, but political science research tells us two important things about polls and predictiveness. First, polls are more predictive the closer they are to Election Day. This means that today's polls are more meaningful than those from last fall.
Second, especially at this late stage, trajectory is significant. A candidate who has been lagging the entire season but is on the upswing late in the game can often be expected to maintain that trajectory and overcome an opponent. With these two findings in mind, let's look at the latest polls from Iowa.
On the Republican side, Donald Trump has a significant lead (more than 7 percentage points, which is a lot in this game). Both Ben Carson and Ted Cruz appear to have peaked and now have downward trajectories. Marco Rubio is in a far trailing third place but has been slowly and steadily climbing through the season. These numbers suggest that Trump may win Iowa on Monday.
When you add the information that Trump is also now leading in the prediction markets, his victory on Monday seems more likely. When we add in the numbers from New Hampshire, whose primary is on Tuesday, February 9, we see a good chance of Trump winning these first two contests.
The Democrats show a different story entirely. After having had a strong lead for many months, Hillary Clinton is now in a statistical dead heat with Bernie Sanders in Iowa.
Looking at Sanders's trajectory suggests that he may come out on top on Monday in Iowa. Add to this the New Hampshire polls, which give Sanders a whopping 14-point lead, and it seems entirely plausible for Sanders to win both of the first two Democratic contests.
The first thing to keep in mind about gaming out these scenarios is that much like the Federal Reserve's Open Market Committee, the fallout from the event is often more about whether expectations have been met than about the outcome itself. If the economy anticipates that the Fed will raise interest rates, and it does, then it's a non-event because the market has already corrected for this expectation. The actual event of raising rates is anticlimatic. But if the Fed raises rates when the market does not expect it, chaos may ensue, if only briefly.
We can think of the Iowa caucus the same way. The market is expecting Trump to win, and maybe Sanders too (but the latter is much less clear). If this expected outcome does not occur — if Trump, for example, comes in second or even third place on Monday — we can expect a big market adjustment. His numbers in New Hampshire and nationally would likely take a big hit. But if Trump wins on Monday, not much is going to change, and we'll march ahead to the next contest.
What implications can we draw from a Trump/Sanders win in both Iowa and New Hampshire? Trump and Sanders are unusual candidates. Both are populists with elements of ideological extremism. Both have significant institutional barriers to overcome on a path to their party's nomination. Trump has been denounced by leading members of the party whose nomination he seeks. Sanders lags significantly behind Clinton in fundraising, though he may have matched her in organization.
This conundrum left me wondering: Have we ever had a case where a non-incumbent candidate won both the Iowa Caucus and the New Hampshire primary but failed to go on to win the nomination? Does winning both of the first two contests, in other words, predict the outcome?
Surprisingly, we have to go all the way back to 1972 to find a case to falsify the above claim (hat tip to professor David Peterson for this gem of data). In 1972, the Democrats who competed for their party's nomination included Hubert Humphrey, George McGovern, George Wallace, Edmund Muskie, and Eugene McCarthy (plus some others).
McGovern was the nominee that year, and lost resoundingly to incumbent Richard Nixon in the general election. In the early primaries, though, Muskie won Iowa (January 24), Arizona (January 29), and New Hampshire (March 7). McGovern didn't win a primary until the Wisconsin primary on April 4.
The Democrats had an unusual process in 1972 for a lot of reasons. After the riots that occurred at the 1968 Democratic National Convention in Chicago, the party reformed the nomination process to allow more voices and fewer party elite directives. Ted Kennedy had been the establishment favorite for some time, but his presidential hopes were squandered when the public learned he fled the scene of a single-car crash in Chappaquiddick, Massachusetts, that resulted in the death of his longtime aide Mary Jo Kopechne. The party struggled in the face of these upsets and bungled its nomination process that year.
On the other hand, in 2000 Al Gore won the Iowa caucus and the New Hampshire primary and went on to be the nominee, and in 2004 John Kerry did it. Both of these candidates were certainly more "establishment" than Trump or Sanders, and the 1972 case may be a closer historical guide to current events.
If Trump and Sanders win Iowa and New Hampshire, 2016 may be like what Democrats experienced in 2000 and 2004, and we may see these candidates as the nominees. But that outcome is far from certain and still seems unlikely in the face of the enormous institutional obstacles they face in their own parties. It's historically unusual to observe the scenario where a candidate wins Iowa and New Hampshire but does not secure the nomination; it's therefore astonishing that we may see exactly this in both of the major parties this year.
If the polls are wrong (and they are more likely to be wrong in Trump's case than in Sanders's), we may see big adjustments happen quickly, and the tides may change (I'm looking at you, Marco). If Trump becomes the Republican nominee, it's likely to upset an already tumultuous party.
This is all still hypothesis testing for many of us, and most evidence still points to the parties pulling together and deciding this thing for more established candidates, but prognosticating is still fun. I expect we'll see Trump and Sanders do well in the early contests in the end not be the candidates who cross the finish line first.
*An earlier version of this post incorrectly described Sanders's lead in NH as 7 points (updated 1/29/2016 2:35pm)