In California, we have seven elections on the state level to examine: the 2003 Governor’s race and the six state legislative elections since 1994 (and, what the hell, we can throw in a seventh from 1914). We will also look at two other states that have the Yes/No question on Recalls, followed by a replacement vote.
In 2003, voters had real options to vote for a Democrat to replace Newsom. In this case, it was not a replacement-level Democrat, but a heavy hitter -- Lieutenant Governor (and former Assembly Speaker) Cruz Bustamante. Bustamante has been treated poorly by history due to the recall result, but you can see that the twice-elected state-wide candidate was serious. California has a split ticket Gov/Lt Gov arrangement, and Bustamante outperformed Gray Davis in his 2002 re-election race. There was every reason to believe that Bustamante could be the viable alternative. But what happened was very different.
55.4% of voters voted to remove Davis, which means that Davis received 44.5% of the vote. Bustamante performance? 31.5%. It was not divided among Democrats either. What happened? More than 62.5% of voters cast their ballots for Republican candidates to replace him. What ended up happening is that 8% of the voters skipped the replacement race (and, perhaps shockingly, 4.6% skipped the Yes/No question). You can say Schwarzenegger was special, but our look in the past shows this recurring again and again.
In 2018, State Senator Josh Newman (D) faced a recall vote. The vote was ostensibly over a gas tax, but it was also quite political, as the removal of Newman deprived the Democrats of a two-thirds majority in the Senate. Newman lost the recall badly, 58.1% against him. The replacement race saw a more than 6% fall off in turnout from the recall vote. Ling Ling Chang, who Newman beat in 2016 (and would lose to him in round three in 2020) won with 33.8% of the vote. However, there were three Republicans and three Democrats in the race. The combined vote total for Republicans was… 58.1%
In 2008, Senator Jeff Denham (R) faced a recall vote, in an attempt to give the Democrats a veto-proof two-thirds majority in the Senate. This is the only Democratic-led recall of the bunch. The Democrats ended up effectively abandoning this one, but it still went to a vote. Denham easily won with 75% in his favor. The only replacement candidate was a Democrat, so he got all the votes. But it was a big fall-off, with only 38% of voters casting ballots in the replacement race. Interestingly enough, 20,043 people voted to remove Denham, but 30,946 voted in his replacement race. So, over 10,000 voters chose to cast ballots in favor of keeping Denham and also selected a Democrat to replace him.
Now we get into 1995, when the Republicans gained a bare majority in the Assembly but then lost it when Democratic Speaker Willie Brown convinced one member to switch. Three Assemblymembers ended up facing recalls that term, all targeted by Republicans – Paul Horcer and Doris Allen, both of whom were elected as Republicans, but ended up supporting Brown, as well as Michael Machado, a Democrat from a marginal seat.
Paul Horcher lost overwhelmingly, 61.6% voted to remove him. In the replacement race, turnout dropped more than 16%. The Republican candidate won with 39.25%, but the combined Republicans received 76% of the vote, well outpacing the vote in favor of retaining Horcher.
The Democrat Michael Machado easily retained his seat, with nearly 63% voting in his favor. In the replacement race only 66% of the voters cast ballots. Despite Machado’s big performance, the replacement race totals were very different. A Republican would have been elected to replace Machado. Combined the Republicans won 68% of those votes.
Doris Allen, who flipped after Horcher was removed, was also kicked out. 65% of the vote was against her. 90.5% of voters who cast ballots in the recall voted on a replacement candidate. The Republicans won 68.44% of that vote, again outpacing the vote to retain Allen.
Finally, in 1994, State Senate President Pro Tempore David Roberti faced a recall over gun control. Roberti easily won, with more than 59% casting a no vote. The drop in vote for the replacement candidate was huge, with almost 40% leaving it blank. The sole Democrat would have won the replacement race, but that may be because the Republicans did not believe there was much a chance and did not coalesce. In fact, the four Republicans combined to get 63.5% of the vote.
In our seventh example, for which I don’t have the full numbers, State Senator E.E. Grant was removed in a recall vote in 1914. He was replaced by the Senator who he beat to win office, Eddie Wolfe, who was most assuredly on the other side.
I should point out that California had a law, overturned in 2003, that voters had to vote yes on a recall in order to vote for the replacement candidate. However, the pre-2003 recall numbers above (in which the replacement vote almost always topped the yes vote in the recall) suggests this law was not really followed. This provision has been around since the beginning, so I really can’t explain how it was used in these recalls, but it seems not to have had any effect.
Two other states are useful to look at here. Colorado has the exact same one-day/two-step process as California. Michigan had a two-day/two-step process, where the replacement vote takes place on a different day. (Michigan has since changed its recall law).
Colorado has only had two state-level recalls in its history, both on the same day. In 2013, Democratic Majority Leader John Morse and Senator Angela Giron lost their seats in recall votes. The Democrats did not run candidates in the replacement races, so the Republicans walked to victory. But there was still a vote on the replacement. The turnout dropped heavily, but it was all on the Democrats side. The vote to remove Moore was 9131-8812. The vote for his replacement was 8,932. The vote to remove Giron was 19451-15376. Her replacement got 19,391. Even with no real opposition, the Republican replacements kept almost the entire Yes vote.
Michigan has had four recall votes. In 1983, two Democratic State Senators were ousted in recall votes. Both were replaced by Republicans on a later date. The sole state-level example of this strategy working was in 2011, when a Republican House member Paul Scott was ousted by 197 votes (12,358-12,126). In February, 2012 he was replaced by another Republican (though Scott could have replaced himself. He declined to run) who won 10,290-8173. Turnout dropped more than 21%. Notably, this was months later so that really needs to mitigate the story.
Perhaps most revealing is the 2008 recall attempt against House Speaker Andy Dillon. The recall took place on Election Day with Dillon running for re-election. So Dillon appeared on the ballot twice, once for his election and once for his recall. He won both easily, but the drop in voting on the recall is noteworthy. Dillon won the regular election 27,864-14,311. He won the recall 23,987-14,257. While almost four thousand voters dropped off for Dillon (86%), the recall forces kept 99% of the vote and lost a grand total of 54 votes.
The idea that the party’s candidate will lose the recall, but someone else from the same party will swoop in and get all of their votes to beat back a divided opposition is not what seems to happen. In fact, the most likely result is that if you lose the recall, you’ll lose the replacement race. While the message was botched, Newsom and the Democrats were likely right to head-off a serious Democratic replacement candidate.