News

CA120: In the primary, mixed signals from early turnout

SAN FRANCISCO, CA - February 15, 2020: Voting Season Starts for California Democrats - Kevin McGovern, Shutterstock

Voters in California have been casting ballots for nearly a month.  In total, more than 2.5 million voters have returned ballots, and in-person voting has begun in 15 counties.

You can track the ballot returns here or sign up for a daily update here.

The big question is what these early numbers mean for turnout.  This is a more complex question than it seems – high early turnout can sometimes be a sign of increased voter enthusiasm and carry through to Election Day. But it can just as easily be that voters are simply casting their ballots earlier for some reason, and those early votes are “cannibalizing” the Election Day turnout numbers.

There are a litany of reasons why we should be skeptical about drawing broad conclusions from early vote numbers.  The most important is how mechanical changes in our election system drive changes in early returns.

The biggest change is how many voters are getting mailed ballots. We have gone from 10 million ballots mailed in 2016 to 16 million mailed this year – a 60% increase.  This has been due to two factors: more voters electing to be permanent Vote By Mail and the growth of Vote Center counties, 14 of which mailed ballots to all their voters, plus L.A. County which mailed ballots to an additional 20% of their voters under a modified version of the law.

As we saw this in Vote Center Counties of Sacramento and San Mateo in 2018, these new vote by mail voters were more likely to go to the polls or drop off their ballots on Tuesday or the final weekend.  Voters have their patterns and they don’t change.  So, increasing the number of vote-by-mail naturally shifts the average ballot return date to later in the cycle.

On top of this, as we have already reported, more than 3.7 million independents were mailed ballots with no presidential candidates.  The campaigns are contacting these voters to bring their ballots to the polls and exchange them, or request new mailed ballots. We have been looking at this, and among those independent voters without the presidential contest on their ballot, 60% say they want to vote in the Democratic primary; of those, 60% say they will go to the polls with their ballots.  If this were to happen we could see a million voters walking into polling places to exchange their ballots.

With these caveats explained, the fact is that we can’t help ourselves.  So here is some analysis of the early vote data and what they could mean.

Turnout is up, and down
Turnout is up, based on raw numbers. There have been 2.6 million ballots cast as of Wednesday night, which is slightly higher than the 2.4 million votes cast by this point in 2016.

However, based on a percentage of mailed ballots that have been returned, it is down. The turnout function is a fraction, and the denominator – those who have been mailed a ballot – has expanded by 60%. Ballot returns have not increased at the same rate.

As a share of voters who were mailed ballots, turnout is currently at 16% statewide, compared to 23% at this same point in 2016.  Underlying these numbers we can see some signals as to what is going on.

Within age groupings, there is some variation, but still higher raw turnout across all groups, even while the rate of turnout this year is lower.

Voters with Ballots

2016 Primary 2020 Primary
Votes Turnout Votes Turnout
18-34 250,374 10% 308,625 7%
35-54 486,365 16% 543,608 11%
55-64 506,697 28% 507,374 19%
65+ 1,129,362 41% 1,232,401 31%

This is a bit more dramatic among voters with Democratic ballots.  The raw number of voters over 35 is actually down, along with the percentages. This could point to something different about either the changes these voters have experienced with increasingly becoming vote by mail or their desire to wait until later in the election cycle.

Voters with Democratic Ballots

2016 Primary 2020 Primary
Votes Turnout Votes Turnout
18-34 163,030 13% 174,513 8%
35-54 269,558 20% 258,792 11%
55-64 269,183 31% 216,523 18%
65+ 583,445 44% 557,491 28%

Looking at other demographics, we can see some of the same patterns: raw turnout higher across the board, but lower turnout rate so far.  The turnout rate for voters who are flagged by Political Data as historically early voters is the highest among all demographics analyzed.

Voters with Ballots

2016 Primary 2020 Primary
Votes Turnout Votes Turnout
Latino 304,084 15% 376,081 10%
Asian 196,258 21% 250,379 15%
African American 28,132 17% 58,710 12%
White 1,861,430 26% 1,909,075 19%
Homeowner 1,545,918 28% 1,611,754 20%
Renter 329,652 19% 345,459 13%
Early Voter 1,080,601 49% 1,131,947 43%

Looking at this same data for voters who have Democratic ballots, we see fewer white voters returning ballots, but among every other population we see the pattern of higher raw turnout, lower turnout rate.

Voters with Democratic Ballots

2016 Primary 2020 Primary
Votes Turnout Votes Turnout
Latino 218,809 18% 240,152 11%
Asian 102,450 25% 113,327 15%
African American 24,962 20% 42,754 12%
White 948,344 30% 811,727 19%
Homeowner 786,383 31% 687,309 19%
Renter 209,169 22% 202,894 14%
Early Voter 554,654 54% 535,882 41%

In analyzing what is going on, it is best to break out the factors into their mechanical and organic pieces.

Mechanical causes of reduced turnout
As outlined above, there are plenty of reasons why we should expect a reduced turnout rate that have nothing to do with voter enthusiasm, but are driven by the changes in the election system.

In 2016 there were one million voters who received ballots in the mail who had previously voted at the polls, but never by mail.  This year that number skyrocketed to 2.2 million. So, a more than doubling of the number of potential first-time by-mail voters.  It’s not surprising that the rate of ballot returns hasn’t kept up. In 2016 only 100,000 of these voters had returned their by-mail ballot by now, the 2020 figure is just 140,000 – an increase, but in percentage time a drop from an 11% turnout rate to a 7% turnout rate.

We know the counties which have converted under the Voters Choice Act are mailing ballots to all voters and introducing this by-mail voting to many people who didn’t request it. After the 2018 elections we saw in Sacramento and San Mateo – the two largest Vote Center counties that year, that these voters did turnout in higher numbers, but they didn’t all vote early.

The lower participation thus far is reflected in the 17% ballot return rate in these counties, compared to 25% in 2016.  But we don’t believe this is a permanent deficit – these numbers should close as we get to Election Day.

These mechanical causes each suggest that turnout could be coming, but later for segments of the electorate that have developed patterns of poll voting, and the transition to by-mail voting might take a bit longer to cause them to change their voting habits.

Organic causes of reduced turnout
This is what most political consultants, candidates, and the media will point to.  They say voters aren’t motivated, or they are holding on to their ballots to wait for the results of the early primary states.

I have been fairly skeptical about this argument since we know from the data that most voters don’t think like political consultants, candidates and the media.  They are more tied to their traditional voting patterns. And we do see that those who have voted early in past election are much higher turnout – nearly 50% already – but we can also see a segment of highly engaged voters who haven’t voted yet.

The population that appears to be holding back from voting are the highest turnout voters – those we know are going to vote because they have cast ballots in five of the last five elections – including the very low turnout 2014 elections.

In 2016, when the election was a binary choice between Hillary Clinton and Bernie Sanders, 54% of Democratic 5 of 5 voters had returned their ballots.  This year that number has dropped to 36%.  The reduced rated of returns is strongest among white older Democrats who have perfect voting records.

These are the voters who appear to be playing a game of kitchen-table consultant, waiting to see how candidates perform in South Carolina, or waiting to see if someone will drop out.

These organic factors strongly suggest that we are headed to a higher turnout than 2016, once these voters participate.

If the early turnout deficit was greatest among infrequent voters, we might say that is a clear sign of low enthusiasm and lower eventual turnout.  But, when the voters who are holding back are the types we fully expect to eventually participate, we have to look at the rest of the data and conclude that turnout in 2020 will be higher than 2016, despite the fact that the current turnout rate is down.

Voting will likely commence at a rapid pace through the weekend and until Election Day.  Keep track at https://twitter.com/CA_120as we continue to track the vote and inital results.

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. 


Support for Capitol Weekly is Provided by: