Skip to main content
ABC News
We Re-Ordered The Entire Democratic Primary Calendar To Better Represent The Party’s Voters

Few things stay the same from one presidential primary season to another, but one remains constant: Iowa holds its caucuses first, and New Hampshire follows with its first-in-the-nation primary. But kicking off the primary season with these two states might not make sense for Democrats. These are two of the whitest states in the country, which makes their populations a poor reflection of the increasingly diverse Democratic Party.

But for the sake of argument, let’s say Democrats could order the primary calendar any way they wanted. How should they do it? One way would be to order primaries1 by how similar a state’s Democratic electorate is to the party’s nationwide voter base. The early states play a key role in winnowing the candidate field, and a state electorate that looks more like the party as a whole may vote in a way that better reflects the opinion of Democrats across the country.

To sort states by how much they resemble the larger party, I looked at the race, ethnicity and education levels of Democratic voters in each state using the Cooperative Congressional Election Study, a survey of more than 50,000 people conducted by YouGov in conjunction with Harvard University.2 The CCES asks respondents who they voted for in the general election, so to estimate a state’s potential Democratic electorate, I included anyone who voted for Hillary Clinton in 2016, plus anyone else who identified as a Democrat. From there, I broke the Democratic electorate into five groups: white voters with no college degree, white voters with a college degree, African-American voters, Hispanic voters, and “everyone else.”3 (I broke white voters into two groups because education is a particularly meaningful distinction among white Democrats — and white voters overall.) I then looked at how different each state’s demographic makeup was from that of the national Democratic Party electorate. This allowed me to sort states by which ones best reflected the party.4

And as you can see in the table below, Illinois is the state whose population comes closest to being a cross section of Democratic voters. So under this hypothetical where Democrats prioritize states that best reflect their party, Illinois would go first in the nominating process, and Iowa and New Hampshire would move toward the back of the line. Now, if this calendar followed the current setup where four “carve-out” states vote by themselves at the start of the primary process, the three states after Illinois would be New Jersey, New York and Florida. Just after the first four would be Nevada, which currently goes third, reflecting the fact that there has been some effort to increase diversity at the start of the real presidential primary calendar.

This would be a pretty big shake-up to the primary calendar, as the top four states in this ranking do not have a tradition of voting early in the nominating process — although Florida has had some success at moving up in line. Heck, New Jersey has usually voted toward the end of the process, in June.5

A primary calendar that better reflects the Democratic Party

States by how similar their 2016 Democratic electorate is to the U.S. Democratic electorate in terms of voters’ race, ethnicity and education, where lower scores mean more similar

“Other” includes people who identified as Asian, Native American, Middle Eastern, mixed or other.

The Democratic electorate includes anyone who voted for Hillary Clinton in 2016 and anyone who didn’t vote for Clinton but identified as a Democrat.

Similarity is determined by Euclidean distance, where a distance of 0 means the items are identical and higher scores mean more dissimilarity.


So what would this mean for the primary process if these states went first? Well, reordering the primary could significantly alter the path to victory, which could favor different candidates than the current process does. A candidate who matches up well with Iowa caucus-goers might not be as successful in appealing to Illinois primary voters.

Another notable consequence is that states with smaller populations would no longer be at the front of the line, as the four states that look the most like the national Democratic Party rank among the 11 most populous states. The first four states in the current system — Iowa, New Hampshire, Nevada and South Carolina — are all relatively small or in the middle of the pack in terms of population. One argument for having small states kick off the process is that they allow a greater emphasis on retail politics as opposed to the expensive ad campaigns required to reach big swaths of voters in major media markets. Because it’s fundamentally harder to meet face-to-face with any measurable percentage of voters in, say, Illinois (population 12.7 million) than in Iowa (population 3.2 million), this new primary calendar would increase the importance of costly media exposure, which would probably benefit candidates who can raise a lot of money early on.

Putting more populous states first also means that a lot more delegates would be up for grabs sooner. While the four states actually leading off in 2020 collectively have around 150 convention delegates who will be pledged to candidates according to how they do in those states’ primaries and caucuses, the first four states in this hypothetical would have around 700 such delegates available out of close to 4,000 total pledged delegates.6 But it’s hard to say whether this would slow or quicken the nomination process.

On the one hand, early contests tend to result in fragmented votes because more candidates are competing, so it’s possible that the Democratic delegate allocation rules could hand a larger number of delegates from these big states to just a few top contenders. If that happened, it could quickly winnow a large field down to a small number of candidates. But states have been trying to cram themselves into the early stages of the nomination process for years, a process known as “front-loading” that also makes a lot of delegates available early on, so perhaps having more delegate-rich states right at the start of the calendar wouldn’t notably speed up the winnowing process. In fact, if three or four candidates garnered a fair number of delegates at the start, they might feel encouraged to fight on until late in the primary season — or all the way to the convention — which could make it harder for one candidate to win a majority of delegates and drag the winnowing process out rather than speed it up.

As you work farther into primary season, the differences between the calendar we’ve come up with and the still unfinalized 2020 calendar become less stark. The next 10 states in our rankings offer a wide variety of geographical and size diversity, much like the current Super Tuesday batch, and the order of states would matter less if multiple states vote on the same day, as is often the case. The main difference at this point is where some of the current carve-out states fall in our rearranging of the calendar. Iowa, New Hampshire and South Carolina would all go much later because they aren’t that representative of the national Democratic Party as a whole, but punting these three states to the bottom of the list could risk serious political fallout for the party, in part because all three states would likely fight to maintain their privileged positions.

In particular, South Carolina’s move to late in the calendar highlights an issue with this ranking that could cause a lot of blowback in the party. African-American voters account for about 20 percent of all Democrats nationwide, and they are particularly concentrated in the South. Along with South Carolina, the Deep South electorate in Alabama, Georgia, Louisiana and Mississippi all fall to the bottom of the rankings. But in the 2016 presidential primary, all of these states had voted by the second Tuesday in March, putting them toward the front of the calendar. African-Americans form a key voting bloc in the party, and moving heavily black states so far back in the calendar could anger some of those voters.

It’s worth noting that this ranking technique gives a big advantage to states with diverse populations within their borders, but another approach could involve grouping states to produce representative electorates across state lines. Because multiple states vote on some primary dates, a group of states that aren’t very representative of the national Democratic electorate on their own could combine to form a voter pool that looks much like the party as a whole. In other words, in lieu of ranking the states individually, the states could be organized so that each primary date is fairly representative. And the way the 2020 calendar is shaping up, the slate of states going after the carve-out contests will be quite diverse.

Another reason our primary calendar might fall short is that it doesn’t take into consideration what general election voters might look like. The overall electorate is whiter than the Democratic Party, so determining the order of states by how representative they are of the whole national electorate may make more sense. After all, the ultimate goal for each party and its candidate is to win in November — it’s not enough to just win the party’s nomination. Our hypothetical calendar also doesn’t account for other considerations that might be worth digging into, such as the ideology of a state’s electorate.

If this thought experiment has shown anything, it’s that any change to the primary calendar would involve trade-offs for Democrats. In other words, there’s no perfect primary calendar — but it’s still worth discussing whether the current one is worth keeping.

Additional contributions by Laura Bronner.

FiveThirtyEight’s 2020 draft: Episode 2


  1. And caucuses, though primaries have far higher turnout.

  2. I used the recommended validated voter weighting.

  3. The “everyone else” category includes those who identified as Asian, Native American, Middle Eastern, multiracial or “other.”

  4. Using the Euclidean distance formula, which calculates the distance between two points — in this case, between each state and the U.S. as a whole. Considering five groups for each state means that the points in question are five-dimensional, but calculating the distance between them works the same way.

  5. The only recent exception was 2008, when the state moved its presidential primary to February as a part of Super Tuesday.

  6. All delegate figures are unofficial at this point, and these numbers do not include superdelegates, who will only come into play if no candidate wins a majority of pledged delegates by the end of the primary season.

Geoffrey Skelley is a senior elections analyst at FiveThirtyEight.