The funny thing is that going into the election I figured Measure G, the parcel tax, would pass narrowly—in the 67 to 68 percent range. It did. How it got there was stunning, however.
I have been doing this for a long time—and in fact, even before I founded the Vanguard. I remember pouring over numbers in Florida in the 2000 Presidential Election and Ohio in 2004. Based on those numbers, I quickly realized that they called Florida too soon for Bush and that Kerry would not win in Ohio.
But those are tricky because you have such different population characteristics and you have to look at what areas have not come in.
In Davis, there are a few things that tend to hold. Early returns generally tend to stand up. You look at two numbers coming out of election day—what was the percentage of votes on election fay (as opposed to early balloting) and how many ballots are there still to count.
The first returns for Measure G had it at 63.9 percent. But on election day it polled 66.9, which pushed the overall number to 65.1.
We then calculated based on 10,000 ballots that it needed about 68.8 to win.
In actuality there were far more than 10,000 ballots to be counted. The number was 15,540 ballots counted after the election—more than were counted prior to the election, but as it turns out that did not factor hugely into the shock.
We know that usually the make up of the election day voters mirrors that of the late ballots cast. Based on the discrepancy between 66.9 percent on election day and the 68.8 percent needed to get 
to two-thirds and pass, we deemed it not impossible but highly unlikely that Measure R would win.
Privately to a person, everyone on the campaign and people on the school board and others conceded this fact. We were all wrong.
It turns out that the second, third and fourth counts after the election far exceeded 70 percent yes. In fact it was 70.5 percent overall.
I don’t know why this was the case, but the later we got into the balloting, the better. The numbers went from 63.9 percent pre-election day to 66.9 percent on election day, to 69.4 in the first count, to 71.5 percent in the second count, to 74.5 percent in the last two counts.
We have never seen that kind of trend.
I think the question that we need to answer is: what’s going on? We think we might have an idea. We have requested data from the county elections office to get a better idea.
This may simply be a one-off thing. In fact, several factors appear to have come together.
The early voters were definitely a lot more conservative. That makes sense if you think about it. The Republicans did not really have a dog in the primary fight. So those Republicans that were going to vote voted early. If they had an absentee ballot they sent theirs in or they voted on election day.
Democrats, on the other hand, had a crowded primary field that all of a sudden was narrowing rapidly.
That explanation makes some sense, but not complete sense. The problem is that one constant in the race was the left—the progressive left supported Bernie Sanders and Bernie Sanders handily won California and Davis.
The establishment Democrats went late for Joe Biden, coalescing around him and giving him the primary win. But Joe Biden didn’t do that well here.
An alternative explanation is that we have a lot same-day registration and a lot of those voters—students—voted for Bernie Sanders. They also—and you can see it in the precincts around the university—overwhelmingly supported Measure G.
But we had a lot of reports that even in non-student precincts, people made up their minds late and brought their absentees in to the polling place where they had to be counted late.
The trend across the board was similar.
Look at the fourth supervisorial district. Jim Provenza went from about 50 to 51 percent on election day to 48 percent. David Abramson, the third candidate in the race, polled an astonishing 14 percent.
In the fifth supervisorial, Duane Chamberlain was up by a scant 90 votes on election day but ended up losing by over 450 votes and nearly 5 percent.
The late vote moved everything somewhat sharply to the left.
Is this a one-time phenomena or a trend? That we will see. One possibility is that, with same-day voting, the late vote will skew heavily left. Another is that this was simply a one-time phenomena.
At any rate, in the future we will be much more reticent to call even what appears to be a fairly safe conclusion.
—David M. Greenwald reporting
I’m not enough of an insider to comprehend your statement:
Can you explain or define a “second count”, a “third count” and a “fourth count”? And why do you refer to them as “after the election” when they are all obviously part of the election? There is a difference between election day and the election. The election, as I understand it, is the tally of all valid ballots regardless of how long it takes.
Sorry – they didn’t release all the results at once, instead, they released it in batches, that’s what the counts refer to.
It would seem that as more and more people vote absentee, even delivering their ballot to the polling place on election day, that there will be increasing uncertainty in election returns if the desire is to call the election before it’s over. In this election I do think the Democratic primary with all of the “debate” appearances and new candidates at the last minute had many people wondering whether to switch for one reason or another and most of those voters are yes on Measure G. In our household there were three different votes for presidential candidates but all three votes for G. It’s the first time we waited until election day to turn in our absentee ballots. So if that is any indication, then it was probably a one time thing.