UPDATE (Nov. 28, 3:41 p.m., 2018): On Wednesday, Kings County reported its final batch of ballots, and Democrat TJ Cox won them 52 percent to 48 percent. That’s significant because, as we wrote below, Kings County was Republican Rep. David Valadao’s best — and perhaps only — shot at netting a significant number of votes. Barring something unforeseen, Democrats have won this election.
The California 21st District is a bit of an enigma. Initially, it seemed like a prime Democratic pickup opportunity in 2018, being three-quarters Hispanic and having voted for Hillary Clinton by 16 percentage points in 2016. But three-term Republican Rep. David Valadao has proven adept at keeping the focus on local issues, such as water access, in this heavily agricultural district in the Central Valley, and the race soon slipped off most pundits’ radar screens. On election night, Valadao jumped out to a 7-point lead, and both the Associated Press and television networks quickly projected him as the winner.
That now looks a bit hasty. On Monday night, TJ Cox — Valadao’s Democratic challenger — took a 436-vote lead (0.4 percentage points) after weeks of counting the district’s remaining ballots, which tend to lean Democratic in California. As a result, many media outlets have retracted their calls of the race. California’s 21st District is now the last unresolved House race of 2018 — if Cox wins, Democrats’ net gain will hit 40 seats; if Valadao prevails, it will stay at 39. For our part, we think Cox is the favorite based on our analysis of the outstanding vote.
We arrived at that conclusion by estimating how many Cox votes and how many Valadao votes remain to be counted — and whether there are enough Valadao votes to make up his deficit. At first glance, that seems very possible — Valadao currently leads in three of the district’s four counties. But those leads are largely due to that initial election-night count,1 which gave Valadao a 7-point lead overall. Since then, Cox has been doing much better in later vote dumps in all four counties.
The CA-21 votes counted since Election Day have leaned more Democratic
Initial vote-share margin in the California 21st compared to vote-share margin since Nov. 7
County | Margin Before Nov. 7* | Margin Since Nov. 7* | Overall Margin |
---|---|---|---|
Fresno | R+8 | D+7 | R+2 |
Kern | D+13 | D+34 | D+22 |
Kings | R+30 | R+16 | R+25 |
Tulare | R+14 | R+3 | R+9 |
Crucially, Fresno County has gone from a source of Valadao support to a place where Cox is now picking up votes. Meanwhile, Tulare County’s initial Republican lean has significantly diminished. Based on the vote-share margin for ballots counted after Nov. 7, Kings County is the only place where Valadao can safely count on picking up the majority of votes from here on out.
Given that, plus how many uncounted votes are left in each county, it’s hard to see a path for Valadao to regain the lead. The California secretary of state helpfully provides the number of “unprocessed” ballots in each county, and as of midday Tuesday, 26,659 ballots remained to be processed in the four counties that make up the California 21st: 15,200 in Fresno, 2,919 in Kern, 1,577 in Kings and 6,9632 in Tulare. But, unfortunately, there’s a catch: Only parts of Fresno, Kern and Tulare fall into the 21st District, which means we can’t just sum up the remaining ballots in these four counties to arrive at the outstanding vote total.
Luckily, there are two fairly simple ways to estimate what fraction of each county’s uncounted ballots were cast in the 21st. The first is to simply assume that the outstanding ballots within each county are distributed in each congressional district proportionally by population (e.g., 26 percent of Kern County voters live in the 21st District, so 26 percent of the remaining uncounted ballots are from there too). This assumption does have its limitations: Maybe, for instance, a county is counting its ballots in some specific order (Fresno County, say, could be sifting through all the 21st District ballots first before moving on to the less competitive 4th, 16th and 22nd districts). That said, we know what share of voters in each county live in each district because California breaks down registered voters by congressional district, so the math is pretty straightforward.
Using this method, we estimate there are 5,109 uncounted ballots in the 21st District. Fresno County contains a plurality of the remaining votes, while Kern County is mostly tapped out. If these estimates are roughly accurate, that would be mildly good news for Cox, since he has done moderately well in recent vote dumps from Fresno County.
Scenario 1: The remaining vote is distributed proportionally
Estimated number of ballots to be counted in the California 21st, by county, as of Nov. 27
County | Unprocessed Ballots | Share of county’s registered voters in CA-21 | Est. number of Ballots Remaining in CA-21 |
---|---|---|---|
Fresno | 15,200 | 16% | 2,496 |
Kern | 2,919 | 26 | 745 |
Kings | 1,577 | 100 | 1,577 |
Tulare | 6,963 | 4 | 291 |
If you’re not convinced that the remaining vote is distributed proportionally, there’s a second approach: Peg each county to a certain fraction of 2016 turnout — say, 89 percent. That’s realistic because we already know this to be true in at least one of the 21st district’s counties: Kings.3 Unfortunately, this method also requires a big assumption: How do we know that each county experienced the same presidential-to-midterm dropoff in turnout? We don’t, and in fact, we know they didn’t, since it’s mathematically impossible right now for Kern County to reach 89 percent (even if all of the uncounted ballots in Kern are in the 21st District, that would add up to only 85 percent of 2016 turnout). But to keep things simple, let’s assume that 89 percent number will hold for the parts of Fresno and Tulare in the 21st District as well. In contrast to the first method, that suggests that we have almost no ballots left to count in Fresno, with a plurality in Kern. This would still bode well for Cox, however, as he has been blowing Valadao away in late updates from Kern.
Scenario 2: Voter turnout can help estimate what’s left
Estimated number of ballots to be counted in the California 21st assuming 2018 turnout reaches 89 percent of 2016 turnout in each county*
Votes counted in CA-21 | ||||
---|---|---|---|---|
County | 2018 (So Far) | 2016 total | Est. 2018 Total | Est. Ballots Remaining |
Fresno | 39,925 | 45,135 | 40,198 | 273 |
Kern* | 38,791 | 48,955 | 41,710 | 2,919 |
Kings | 28,693 | 33,988 | 30,270 | 1,577 |
Tulare | 3,499 | 4,330 | 3,856 | 357 |
Which set of estimates is more accurate? We don’t know, but either scenario is good for Cox. If he does indeed score the come-from-behind win, it would be Democrats’ seventh red-to-blue flip in California alone — more than one-sixth of their nationwide pickups. The district may not matter to House control, but it would certainly put an exclamation point on Democrats’ dominance in the Golden State in 2018.