The Democratic nomination race is chaotic. There’s a crowded field, ever-shifting polls, spirited debates, political elites failing to unite behind one candidate, and Sen. Bernie Sanders, a self-described democratic socialist, emerging as the front-runner after the first three contests. So who will be the nominee, and when will we know?

Political scientists who study presidential nominations typically have a firm grasp of the factors enabling a candidate to emerge victorious. Historically, the candidate who wins the nomination is the person who has raised the most money and is leading in the polls headed into the nomination season.

But this year, that understanding has been disrupted by one person: former New York mayor Mike Bloomberg. Bloomberg’s unprecedented leap into the race makes it difficult to predict how the coming weeks will unfold.

Usually, the nomination process is an ongoing feedback loop: Early success leads to more success

The current U.S. nomination process involves a sequence of primaries and caucuses that take place over months. This creates an interconnected and cyclical relationship between success in the early contests, fundraising, media attention, polling numbers and success in later contests. While the early states don’t award many delegates, they can offer candidates momentum: Candidates who seem to be on an upward trajectory attract favorable media attention and additional voter support, which boosts them further.

In his book “Before the Convention,” political scientist John Aldrich shows that a candidate who performs well in an early state will see an uptick on several important dimensions: favorable media coverage, the ability to raise money, and his or her standing in the polls. All this should lead to success in subsequent primaries — continuing the cycle over the course of the race.

Bloomberg disrupts the loop

Bloomberg has disrupted this anticipated cycle in several ways. The former mayor did not even announce his bid until November 2019. By not competing in the first four contests, Bloomberg has removed the possibility of generating momentum from the early states and bringing in the resources that we expect to follow.

We saw Bloomberg rise in the polls after Iowa and New Hampshire. This is seemingly incongruous, as research shows that the results of New Hampshire help predict the future primary vote. But that boost can be explained by several factors: former vice president Joe Biden’s poor performance in those first two states; the Democratic Party electorate’s emphasis on electability and defeating President Trump in November; and the search for the most-electable “moderate” candidate. But it still doesn’t fit with political science expectations. Only once since 1976 has a candidate become the nominee without winning Iowa or New Hampshire: Bill Clinton in 1992.

Bloomberg already has the money, and he’s spending it on advertising

Unlike his competitors, Bloomberg doesn’t need the flood of resources that momentum brings to a candidate. Because he is self-financing his campaign, he isn’t concerned with raising money from rank-and-file voters and big donors.

Through record spending, Bloomberg is bombarding voters with paid advertisements and campaign mailers, having already spent more than $233 million on advertising. This also reduces his reliance on free media coverage, since he can get his name and message out to voters on his own.

Will that barrage matter? Recent work analyzing advertising in the 2016 presidential election suggests that the impact and effectiveness of advertising depends on voters’ knowledge of the candidates and how the media covers the campaign. Other research has found that advertising can indeed persuade. Because Bloomberg’s approach is entirely new, we just don’t know what the possible persuasive effects could be when one candidate is outspending all of his competitors so overwhelmingly. It’s never happened before.

How much do the debates matter?

Only recently have most voters had an opportunity to see Bloomberg unscripted, outside his own carefully polished advertising. He wasn’t in any of the first eight debates, for several reasons: First, he didn’t jump in until late, and second, once he did he was barred by the Democratic National Committee’s rules. Candidates qualified for these debates by meeting specific thresholds for raising funds from small donors and reaching a certain level in the polling. Because Bloomberg didn’t solicit donations at all, he remained ineligible.

In January, the DNC announced that to qualify for the ninth debate, held in Nevada on Feb. 19, a candidate would need to either have secured a delegate or surpassed a polling threshold. The removal of the fundraising qualification and the switch in selection criteria language from “and” to “or” meant that Bloomberg could (and did) qualify for the debate stage.

Political scientist Julia Azari recently reviewed the research on the effect of debates in primaries. Debates can affect what voters learn about the candidates; their impressions of the candidates; and their assessments of who’s electable. How the media covers debates can also influence voters. All that is likely bad news for Bloomberg. His first debate performance was widely interpreted as dreadful, though he was judged to have performed a bit better in the most recent debate.

What will happen next?

Biden’s performance in the South Carolina primary on Saturday, three days before 14 states vote on Super Tuesday, will be critical for his ability to remain in the race, as well as Bloomberg’s prospects. If Biden can regain some momentum and revive his campaign, he will probably persuade some moderate voters in Super Tuesday states to move away from Bloomberg.

If Biden falters, that will further open the door for Bloomberg’s pitch that he’s the most electable candidate, especially given his seemingly limitless campaign spending. On Super Tuesday, we will have the first test of whether unparalleled numbers of advertisements and an unprecedented decision to skip the first four states can defeat the dynamics political scientists expect to unfold.

Caitlin Jewitt is an assistant professor of political science at Virginia Tech and author of “The Primary Rules: Parties, Voters, and Presidential Nominations” (University of Michigan Press, 2019).