The System Isn’t ‘Rigged’ Against Sanders

A week ago, New York Daily News columnist and Bernie Sanders supporter Shaun King tweeted the following about the Democratic caucuses in Washington, which took place in late March:

https://twitter.com/ShaunKing/status/733357861808529408

Whether King intended it or not, he implied that caucuses — which often require hours of participation and mean lower turnout — are representative of what would happen if a larger electorate had its say. Well, a funny thing happened in Washington on Tuesday: The state held a mail-in, beauty-contest primary — so voting was easy, but no delegates were at stake. (The Associated Press has declared Hillary Clinton the winner.) The results are still being finalized, but Clinton leads by about 6 percentage points with more than 700,000 votes counted. Sanders won the Washington caucuses, which had 230,000 participants, by 46 percentage points.

So, turnout was much higher in the Washington primary than in the caucuses, and Clinton did much better. Something similar happened in Nebraska, where Clinton lost the early March caucuses by 14 percentage points and won the early May primary, in which no delegates were awarded, by 7 points.

Nebraska and Washington are part of a pattern. As Sanders fans claim that the Democratic primary system is rigged against their candidate and that Sanders wins when turnout is higher, they fail to point out that Sanders has benefited tremendously from low-turnout caucuses. Indeed, if all the caucuses were primaries, Clinton would be winning the Democratic nomination by an even wider margin than she is now.

Let’s start out with the real-world numbers. Here are the delegate and vote totals by contest, including caucuses and primaries, so far:

Counting only caucuses, Sanders has won 63 percent of the vote, 64 percent of the delegates and 11 of the 16 contests. In doing so, he has earned 341 elected delegates, compared with Clinton’s 195 delegates, for a margin of 146 delegates. These caucuses have had approximately¹ 1.1 million participants. As a point of comparison, turnout in the caucuses has been only about 13 percent of the total number of votes President Obama got in the 2012 presidential election in these states.²

Sanders has done far worse in the states that have held primaries. Counting just primaries, including Tuesday’s in Washington,³ Sanders has won only 42 percent of the vote, 42 percent of delegates and 10 of the 34 statewide contests.⁴ Clinton earned 1,576 elected delegates, compared with Sanders’s 1,158, for a margin of 418. The turnout in these contests has been far higher than in the caucuses, with a little more than 24 million votes cast. That’s about 49 percent of the total number of votes Obama got in the 2012 election in these states.⁵

Now, it is fair to point out that the caucuses have taken place in states that are demographically different than the primary states. Caucus states in 2016 are overwhelmingly white and overwhelmingly rural compared with primary states. Still, these differences don’t come close to explaining the differences in results between the caucuses and primaries so far. We can look to Nebraska and Washington as two examples of the disparity. Of course, one could argue that because no delegates were up for grabs in those states’ primaries, the campaigns didn’t really compete for residents’ votes and therefore those contests aren’t representative of what a truly competitive primary would look like there. Fortunately, because the vote in the Democratic primary has largely broken down along demographic lines, we can use statistical models to approximate what would happen if states that held caucuses had held primaries instead.

At various times, we’ve tried using demographics to model the vote in the Democratic nomination contest so far. The model considers each 2016 contest and controls for (i) the black and Hispanic share of the Democratic vote in that state in the 2008 general election, (ii) whether that primary or caucus is “open” to independent voters unaffiliated with a political party, and (iii) the margin in national primary polls at the time the contest is held. This model estimates that holding caucuses instead of primaries is a massive advantage for Sanders. In fact, Clinton would do about 20 to 25 percentage points better relative to Sanders if a state changed from a caucus to a primary, the model estimates.

Here’s how we project each caucus would have gone if a primary had been held instead:⁶

Sanders fans have claimed that because caucuses have lower turnout the current national caucus and primary vote underrates how well Sanders is doing. In fact, the opposite is true. When we switch all caucuses over to primaries, Sanders actually does worse. Clinton’s lead in the popular vote would grow from 2.9 to 3.3 million votes. Moreover, her edge in elected delegates would expand significantly.⁷ Instead of her current lead of 272 elected delegates, Clinton would be ahead by 424.⁸ Some states that were won by Sanders in caucuses, including Colorado and Minnesota, would be won by Clinton in primaries, according to our calculations.

In fact, counting the 537 superdelegates The Associated Press currently gives Clinton, she would likely have 2,384 total delegates if every state had held a primary. That’s one more than necessary to clinch the nomination.

But what would happen if every state held a primary that was open to independent voters? Independent voters, after all, have been among Sanders’s strongest groups, and Sanders supporters have consistently cited closed contests as evidence the game is rigged. We can rerun the same regression as above but estimate what would happen if all the primaries are open to unaffiliated voters.

Clinton’s margin in the national popular vote shrinks to about 8 percentage points (from 12). That’s because opening a primary to independent voters shrinks Clinton’s margin in a state by about 10 percentage points on average, according to the model. Sanders would also project to win Connecticut and Kentucky, which he lost in the real world when they held closed primaries.

Still, this wouldn’t make all that much difference. Just 11 states⁹ held closed primaries, so the national vote is mostly reflective of a process open to unaffiliated voters. Indeed, Clinton has won 14 primaries¹⁰ open to independent voters, while Sanders has won nine.

In fact, if all states held primaries open to independents — instead of closed primaries, or caucuses of any kind — Clinton might have a larger lead in elected delegates than she does now. The model indicates that Clinton would have a lead of 294 elected delegates, compared with the 272 she holds now. That’s not a huge difference, but it means that Clinton has been hurt at least as much by caucuses as Sanders has been hurt by closed primaries.

Listen to the latest episode of the FiveThirtyEight politics podcast.

What would happen if the primary system conformed to each candidate’s best-case scenario? (All closed primaries for Clinton and all caucuses open to independent voters for Sanders.) If every state held a closed primary, Clinton would beat Sanders by 19 percentage points and have a 654 elected delegate advantage, we estimate. If, however, each state held an open caucus, Sanders would beat Clinton by 22 percentage points nationwide and have a 496 elected delegate lead. Of course, neither of those scenarios would happen.

Realistically, if you throw everything together, the math suggests that Sanders doesn’t have much to complain about. If the Democratic nomination were open to as many Democrats as possible — through closed primaries — Clinton would be dominating Sanders. And if the nomination were open to as many voters as possible — through open primaries — she’d still be winning.