To handicap the race properly, you have to know who votes when.
To handicap the race properly, you have to know who votes when.
I f you want to call the game, you have to know the playing field. The most important element of any election is the voters. The playing field of voters in a national presidential primary is unlike any other American election, because each state and territory votes not only separately, but also in sequence — not all at once, and not all under the same rules.
Last March, before Joe Biden had even officially entered the race, I looked at the Democratic primary calendar and what it could tell us. We’ve had a lot of water under the bridge since then — states moved their primaries or changed how they vote, delegate allocations were changed, major contenders left the race, and nearly a year of poll data gave us a clearer (but far from certain) sense of whom the leaders are. Let’s take a look at what the map is telling us as we enter the final weekend before the Iowa caucus.
The Field
First, here is where the poll averages stand at this writing for the national polls and the first four states that vote in February:
Barring a last-minute surge in Iowa by Amy Klobuchar, a Minnesotan who has banked her whole campaign on her neighboring state, this is a four-horse race that increasingly looks like it could quickly become a two-horse race between Biden and Bernie Sanders. But funny things can happen at the last minute in Iowa. The most stunning late surge was in the 2012 Republican caucus, when Rick Santorum won after being in sixth place and single digits in the polling averages as late as a week before the vote.
In December 1975, a month before Jimmy Carter won the Iowa caucus with 27 percent of the vote, a nationwide Gallup poll showed Hubert Humphrey in first place at 30 percent, George Wallace at 20 percent, Henry “Scoop” Jackson at 10 percent, and Birch Bayh at 5 percent. Some 29 percent of Democrats said they would back Ted Kennedy if he ran. Carter wasn’t even on the radar. Carter was in better shape in the Des Moines Register’s Iowa polling, but his victory still totally overturned the race. National poll leaders in January lost the Democratic nomination in 2008, 2004, 1992, 1988, and 1972. Polling has gotten more sophisticated since then, but large fields and sequential primaries make it a lot less reliable than general-election polling.
Iowa is particularly unsettled in this year’s Democratic race because of the way the 15 percent threshold interacts with the caucus process. Unlike the 2016 Republican race, and even many past Democratic primaries, there are no winner-take-all Democratic primaries this year. Various states have different ways of dividing up delegates — some statewide, some on a district-by-district basis — but many have a 15 percent or similar threshold that prevents minor candidates from gathering any delegates. And Iowa’s caucus rules have a particular wrinkle: In each individual polling place, after the original votes are counted, all the candidates below 15 percent are eliminated and their supporters must switch to one of the remaining candidates (or band together to make one of the under-15% candidates viable) if they want their votes counted. That means that even a candidate who wins the statewide popular vote may be effectively wiped off the ballot in some polling stations. Second choices could decide Iowa.
In the longer term, the proportional allocation of delegates and the elimination from first-ballot voting of most of the “superdelegates” appointed in prior Democratic races greatly increases the chances of a contested or brokered convention at which the party establishment is no longer able to dictate the outcome. A nomination divided at the convention is one of those quadrennial fantasies of political journalists, like the incumbent firing the vice president or a new nominee picking a running mate from the opposing party: often discussed, never done since 1976. It’s still likely that the front-runner will be clear enough by June that his or her rivals can be bought off with some concessions well before the primary, but this time the possibility is less fanciful than usual.
The Calendar
Now to the calendar. Below is a handy in-one-place chart (apologies, as usual, to our color-blind readers) showing the date each state votes, the method of voting, the delegates awarded (total delegates and the number who are pledged on the first ballot, in most states around 80 to 85 percent of the total), and the cumulative first-ballot delegate count, all color-coded by the state’s racial demographics (more on which below). A total of 4,750 delegates are awarded (all but five of them by the results of primaries), but only 3,979 of those (84 percent) are pledged on the first ballot, and unpledged delegates are not allowed to vote until the second ballot, if there is one. So to take the nomination on the first ballot, a candidate needs 1,990 delegates; to win on subsequent ballots, 2,376.
Super Tuesday is no longer the almost exclusively Southern event that it was in the past, and California — in past years the final major primary — is now its biggest prize by far. Nate Cohn at the New York Times has analyzed how the calendar is no longer quite as front-loaded with states dominated by black voters as it was in 2008 or 2016. If they can make it that far, Bernie Sanders, Elizabeth Warren, and Amy Klobuchar all have their home states voting on Super Tuesday, in addition to some unusual early entrants such as Utah. While two of the four February states are caucuses (Iowa and Nevada), there are only two caucuses among the 16 Super Tuesday contests, and they’re the smallest ones: Democrats Abroad and American Samoa.
Also, regardless of how big an event Super Tuesday is (with a third of all delegates on the line in one day), more than half of the delegates will still be on the table going into the St. Patrick’s Day primaries in Florida, Illinois, Ohio, and Arizona. The back end of the race has many more closed primaries, which work against nontraditional candidates (such as Sanders) who depend on drawing voters who are not registered Democrats. Caucuses get even rarer in the second half of the calendar: After March 14, only Wyoming, Guam, and the Virgin Islands hold caucuses.
Next, consider the demographics of the calendar, using the most recently available Democratic-primary exit polls. Iowa and New Hampshire have come under increasing fire for being racially unrepresentative of the Democratic electorate — which they are, though they are swiftly followed by Nevada and South Carolina. The 2016 exit polls showed that more black women voted in the South Carolina Democratic primary than white voters of both genders put together. Nobody has yet cracked Biden’s majority support with these voters, but he’s also not pulling the 90 percent share that Barack Obama drew in 2008. Nevada starts to bring Hispanic voters into the mix — a very large factor in the Super Tuesday races in California and Texas, with likely a third of the vote in each state. Significantly, after Louisiana on April 4, there is only one state (Maryland) where past exit polls show a likely electorate above 30 percent African American. If Biden, who still leads heavily among black voters, is going to rely on those voters to carry him to victory, he’ll have to deliver the knockout punch before the campaign moves to the Northeast and the Great Plains in April and May.
Pacing is also important, especially if Iowa is won by Buttigieg or Klobuchar, neither of whom has much in the way of support beyond New Hampshire. There are eight days between Iowa and New Hampshire, eleven more until Nevada, and another week until South Carolina. After that, however, there will be little time to organize, fundraise, and advertise before half the delegates are awarded in a two-week span. That, too, gives an edge to Biden and Sanders, both of whom have extensive name recognition, donor bases, and organization all over the country.
Grouping the states by their share of white vs. non-white voters, there are 13 contests (eleven states plus D.C. and Puerto Rico) where non-white voters are likely to be the majority. (I have not bothered categorizing the smaller territories, which between them award 64 delegates, but the same is likely true in some of those.) These 13 contests collectively will choose a third of the delegates.
Of course, “non-white” voters are hardly a monolithic voting bloc. Some states have lots of African-American voters and few Hispanic voters, or vice versa. Some, such as New Mexico and Hawaii, have large Native American/Native Hawaiian electorates. Polling has shown, especially among black voters, a growing generation gap: younger black voters tend to be more educated and more open to Sanders and Warren, while older black voters are overwhelmingly pro-Biden.
The share of delegates awarded by states with large non-white electorates rises to two-thirds when you add in the next group, the twelve states where white voters are a likely majority of Democratic primary voters, but non-white voters are expected to make up over 30 percent of the vote. These two groups of states also have a heavier ratio of female to male voters.
The in-between states are largely in the Midwest:
Finally, let’s look at the states that are very heavily white, which at least in the Democratic primary are mostly small states that do not deliver a lot of delegates. This list includes Iowa and New Hampshire but is otherwise very heavily concentrated late in the calendar.
Of the states with 2016 or 2008 exit polls available, we can also see the divide between primaries that are heavily female and those that are not. Ten contests had electorates that were at least 60 percent female in the most recently available data.
The more balanced states, a decided minority:
Age and Identity
How much do the racial and gender divides matter? Voter demographics are always important. But two things that have thus far not been a huge factor in this race — at least not in the way I and many others had expected — are identity politics and candidate age. By “identity politics,” I mean its most explicit manifestation: voters selecting a candidate by reference to the candidate’s race, gender, sexual orientation, and the like. The Democrats started with a huge and very diverse field, and we’ve heard a great deal from candidates, candidate surrogates, and progressive commentators explicitly arguing for the need to choose a nominee who is not white, male, old, and/or straight. I’ve collected a running thread of examples of these arguments on Twitter. Given the preferences of Democratic voters in the 2018 election cycle’s primaries and in the 2008 and 2016 presidential primaries, it was not unreasonable to expect this to be a factor. Yet, here we are, on the eve of the voting, and the two leading candidates are a pair of white men born in 1941 and 1942. Aggregating their standing in the polls over the early states, we see over 80 percent of the voters supporting white candidates, almost 70 percent supporting male candidates, and over 60 percent supporting candidates over the age of 70.
One explanation that has been offered for this — reporters in the field have heard it from Democratic voters, and some polling backs this up — is not abandonment of identity politics but something quite different: voter fear that only a white male candidate will be “electable” and appeal to voters who backed Donald Trump. This strikes me as Democratic voters’ projecting their own prejudices onto the thinking of swing voters; after all, it was not so long ago that Barack Obama won two popular majorities, something Trump did not do. Hillary Clinton was a deeply flawed candidate for reasons personal to her, but she still got more votes than Obama in the 2008 primaries and more votes than Trump in the 2016 general election. Even among Republican-primary voters, the two Hispanic candidates (Ted Cruz and Marco Rubio) combined to get more votes in the contested primaries and caucuses through Indiana than Trump did. It is certainly possible that Biden and Sanders are actually the most electable candidates — Biden runs better in head-to-head polls than any other Democrat, due partly to name recognition, and Sanders can reach some people who are alienated from the usual two-party system — but it seems unwise for voters to bank on their race and gender for such assessments.
As for age, Democrats have not elected a new president over the age of 56 since James Buchanan (1856). Democratic voters tend to fall in love with the new, fresh face — but the recent fade of Pete Buttigieg in Iowa and his failure to gain traction outside the first two states suggests that the contest may well come down to three candidates over 70, two of whom are as old as Donald Trump will be four years from now. And Trump himself will be older than any previous major-party nominee.
If Biden and Sanders are the choices, identity politics and candidate age will favor neither of them. If anything, polling data so far indicate that the more interesting voter-turnout questions will be about age and education more than race or gender.
2016 and 2008
Another way of looking at the calendar is to divide up the states by who won them the last time around. If we mark the states Hillary Clinton carried twice in blue, the states Bernie Sanders won in 2016 in red, and the states that went for the winner in each of the last two elections in brown, we see a pattern of Bernie support in New England and the West, and the decisive swing bloc in the South. The field is different now, however, and so are the states.
The states Hillary won twice are mostly a mishmash of states with large, urban Hispanic populations and states with large, rural white working-class populations (or in some cases both). Under the 2020 delegate allocations, they deliver just nine delegates short of the total needed to win the nomination.
The states that swung from Obama 2008 to Hillary 2016 control only half as many delegates, are concentrated in the South, and include a high proportion of open primaries where anyone can vote.
The most interesting fact about the 17 Obama-Bernie states is that only three of them will hold caucuses this year, compared with more than half in the past. Dominating the caucuses with progressive activists was a major element of the Obama 2008 and Bernie 2016 strategy, and the party establishment has sought to prevent that this time around.
Only six states swung from Hillary to Bernie in 2016, and not one of them has a closed primary.
Adding up all these different elements of the calendar, it’s possible to group the 57 contests. About 95.7 percent of delegates will be awarded by primaries, not caucuses. Over 60 percent will be awarded in March. More than half will be awarded in states where women are likely to be at least 57 percent of the voters, and 44.4 percent will be awarded in closed primaries, compared with 26.7 percent in fully open primaries.
The landscapes of the first four states remain familiar, and they will get the first say in whom the real contenders are. But after South Carolina, the calendar and the map are new, and they could make this race less predictable than in years past. Buckle up.
This article has been corrected to note that candidates eliminated by the 15% threshold in Iowa may still clear that threshold if they can draw more support in the re-caucus.