CA120: The math of the March primary

As we barrel toward the March 3 primary election, most eyes are on national and statewide polls showing a tight contest between four top contenders, with the latest Capitol Weekly polling showing Senator Bernie Sanders with a slight lead over Sen. Elizabeth Warren and Vice President Joe Biden, followed by Pete Buttigieg, the former mayor of Southbend, Ind.

But California’s Democratic primary isn’t won at the “national poll” level, or even at the “statewide poll” level – it’s won through a complicated, proportional, delegate-allocation procedure that affects  campaign strategy and could muddy the final election results.

So get ready for some numbers — a lot of numbers.

California will send 495 delegates to the National Democratic Convention this summer in Milwaukee, Wis. Of those, 416 of these will be pledged delegates (meaning that they are committed to a candidate), while 79 will be the unpledged “superdelegates,” consisting of the 30 Democratic National Committee members, 46 Democratic members of congress and the governor.

The bulk of those pledged delegates will be allocated at the congressional level, with four to seven delegates awarded in each of California’s 53 congressional districts, for a total of 272. In addition, there are an additional 144 delegates awarded based on the statewide results.

But, there’s a catch, and here’s where it gets tricky.

To win any delegates within a congressional district requires the candidate to obtain the support of a minimum of 15% of that district’s voters. 

And the state result is subject to the same rule: If only one candidate gets 15% statewide, that person could win all 144 statewide delegates, but if five candidates reach 15% they would divide the delegates among themselves.

Given the method of calculating delegates at the congressional district level, the “winner” might only get 20,000 votes more than the closest competitor but receive just one more delegate.  Or, if the winning contender was the only one to reach 15%, he or she could win by a mere 1% margin yet capture  100% of the delegates.

Here’s specifically how the rules are applied, and then a couple examples.

–1. Votes are cast and calculated at the congressional district level.

–2. Any candidate that reaches 15% is eligible to obtain delegates.

–3. Those who don’t reach 15% have their votes eliminated from the calculation and percentages are re-calculated for the allocation of delegates.  This is called the “qualified” vote.

–4. The re-calculated percentage is multiplied by the total number of delegates (anywhere from 4 to 7 based on the district) and whole numbers of delegates are awarded. (Example: if a candidate’s percentage of the vote times 5 delegates equals 1.1, they get 1 delegate.  If another candidate’s calculation yields 1.9, they also get one delegate.)

–5. After all the shares of delegates are allocated, any remaining delegates are allocated by one going to the highest number after the decimal point, until remaining delegates are all allocated. For example, if someone’s calculation was 1.8 and the other person’s was 2.7, then the extra delegate would go to the candidate with the “.8”.

Using our polling to date, we can come up with a mock election at a congressional level in order to play out this math. Click here to see a graphical depiction of delegate selection for each of the four candidates.

This is meant to tell us  how the votes can break down, and it draws from several months of polling. It isn’t intended to be a snapshot of the electorate today, but it illustrates what the kinds of data that pollsters examine.

For voters who selected candidates who have dropped out, we have reallocated votes based on second choices.  We also weighted the results to account for a lower natural response rate from Latinos in our polling to date.

Here are two examples from districts with over 100 completed responses:

CD 2 (110 Respondents)
The initial vote in this district shows the same top four candidates, along with voters down the ticket getting some votes. This is consistent with what we could expect in a final result.

The first step is to eliminate all the votes cast below the 15% threshold, which eliminates everyone but Sanders and Warren and creates the new share of “qualified vote.”

Now with the qualified vote, we can multiply those by the six delegates in this congressional district.  Multiplying .597 and .403 times the 6 delegates gives Warren 3.6 and Sanders 2.4. We truncate at the decimal for the initial allocation, giving 3 to Warren and 2 to Sanders, for a total of 5 delegates allocated.

We now have one remaining delegate, and this goes to the person with the highest value after the decimal, so, in this case it goes to Warren.

CD 5 (100 Respondents)
The initial vote shows three candidates above 15%, with Sanders just a couple points out of the delegate hunt.

We use the 15% threshold to focus on the three candidates with a qualified vote to earn delegates.

Now with the shares recalculated to just the qualified vote, we can multiply those by the 6 delegates in this congressional district.  Multiplying the recalculated vote share by the available delegates gives Warren 2.4, with Sanders and Buttigieg at 1.8 each. We truncate at the decimal for the initial allocation, giving 2 to Warren and 1 to Sanders and Buttigieg, for a total of 4 delegates allocated.

We now have two remaining delegates, and these go to the candidate with the highest value after the decimal. In this case, that means one each for Biden and Buttigeig.

What we see here is a Warren victory by 5-points, adjusted to 9-points after eliminating those who didn’t reach the 15% threshold, and still getting the same number of delegates as her two challengers who also hit the 15% threshold.

Interestingly, if we played this out a bit differently – say,  lowering Buttigieg’s initial vote percentage to just 15% he would still get two delegates, and if we increased his vote percentage to 29% he would get two delegates.

The greatest gain in this scenario, where three candidates make the threshold, comes from exceeding the 15%, with no gain for nearly doubling the vote total. 

Statewide Delegate Allocation
Using the polling data broken down to the district level we can also provide a breakdown of our last few months of polling and calculate delegate allocations. 

This, however, is not a projection – it is a way to understand how delegate allocation works.  The median sample size is only 51 respondents, so we have some districts with strange results.

With that caveat, you can view a visualization of these delegate allocations or the raw data and district level calculations.

The current polling for January shows Sanders in first place, but using surveys going back three months the leader is Warren, by just two points over Biden and three points over Sanders. 

As in the January poll, this three-month sample has Buttigieg lagging behind the top three, but with a high enough vote share to ensure he gets some delegates at the congressional level.

In this three-month sample, Buttigieg gets zero statewide delegates, and they are split up fairly evenly by the top three candidates.

The congressional level allocation provides Buttigieg with 24 delegates, but still puts him significantly behind Biden, Sanders and Warren, who each get around three times the number of congressionally allocated delegates.

When looking at the impact of the proportional delegation we can see how it distorts the real voter choices at both the congressional and statewide level in three distinct ways.

The first distortion is the elimination of vote share for any candidate that doesn’t reach the 15% threshold.  Based on this delegate calculator, 20% of California’s primary votes in the Democratic primary are tossed out because those candidates don’t reach the 15% threshold in the district or statewide level. 

The second is how the failure of Buttigieg in this model gets just two points away from the 15% threshold, and that costs him any chance to get delegates at the statewide level.  At the same point, the delegate allocation for the other three candidates are given a 10%-to-11% bonus – the failure of one candidate to make the threshold serves as a prize for those who do meet the threshold.

The third distortion is at the congressional level, where the allocation again hurts the lower-polling candidate, taking 4 percentage points off their statewide total due to he or she not reaching the 15% threshold in most districts.  At the same point, the other candidates earn a 5%-to-10% bump through the process of eliminating delegates from candidates under the 15% threshold and dividing the remainder among them.

It’s noteworthy that none of the lower polling candidates picked up even a single delegate at the congressional level. 

The closest was Michael Bloomberg who obtained 14% support in the 52^nd Congressional District.  If the delegate allocation was based on just the last two months of polling, Bloomberg gets two delegates in this district and picks up delegates in three other seats, for a total of 7 delegates from the congressional level.

Lessons for Campaigns
Campaigns heading to California might be largely focused on the early states, hoping to catch a wave of momentum heading in the Super Tuesday. 

But, looking to California there is a real need now to look strategically at the state and the delegate allocation rules to ensure a strong showing in March.

That 15% threshold is a killer.  Not achieving it statewide is going to hurt, but not reaching it in targeted congressional districts could hurt even more, particularly for a lower polling candidate.

Running up the score and getting more votes in a congressional district where you’ve already reached that 15% threshold could be wasted votes, and the energy being used to win big in one seat could be better used in a lower target district just breaking the threshold.

Getting to 15% also means that the small things matter more. There will be dozens of districts where a couple thousand votes determines a delegate or two.

How are campaigns going to deal with the early voters – those who consistently vote as soon as they get their ballots? 

These voters aren’t going to be swayed by the results of South Carolina or New Hampshire because they will have already voted. Campaigns can be banking their early supporters as soon as the ballots drop – inoculating themselves from any bad results in the other early states. Or they could be persuading those who are already flagged as early voters in the 2012 and 2016 primary elections.

And, as we have covered extensively, how are these campaigns converting their support among nonpartisan voters into real votes when these independent voters are having trouble getting the correct party primary ballots?

Finally, we have to expect that these calculations will take time – potentially weeks – for us to truly determine who won California, and by what margin.

We know that there will be potentially hundreds of thousands of independent voters re-registering as Democrats or replacing their ballots to get the Democratic presidential crossover ballots — potentially increasing the administrative tasks for counties. 

The state is implementing reforms, like a significant expansion of same-day-registration.

But that could extend the counting time even more, as same-day registrants must have their eligibility verified and counties must confirm that the voter didn’t cast a ballot in another county. And that has to be completed before those votes can be tallied.
—
Editor’s Note: Paul Mitchell, a regular contributor to Capitol Weekly, is the creator of the CA120 column, vice president of Political Data and owner of Redistricting Partners, a political strategy firm.