Yes: 7,001,084 votes; 52.3%
No: 6,401,482 votes; 47.7%
The official margin: Yes prevailed by 599,602 votes
340,611 people voted on November 4 but did NOT vote on Prop 8.
Total ballots cast was 13,743,177; ballots cast on Prop 8 was only 13,402,566. Thus, 97.5% of voters who cast a ballot in 2008 voted on Prop 8.
The official result on Prop 8 makes the vote looks closer than it was because No on 8 was the net beneficiary of wrong-way voting. Had all voters successfully cast a ballot to match their intention, Yes on 8 would have prevailed with 53.9% of the vote instead of 52.3%. See Appendix K for a full explanation of the calculations made to measure wrong-way voting. See Findings 1 and 7 for a discussion of the key lessons learned about wrong-way voting in this election.
Most political consultants believe, or assume, that voters who are undecided on the substance of a ballot measure tend to vote no on the measure. The idea is that if voters have any doubt, they vote no.
But that assumption was not true in Prop 8. On the contrary, the 10% of voters who expressed indecision on the issue of marriage for gay and lesbian couples—let’s call them inertial voters—started out leaning toward voting Yes on Prop 8. At the end of the campaign, they leaned even more strongly toward voting yes. This was true among all demographic groups.
It is not known why the undecided voters broke in the direction of voting Yes.
Perhaps a significant number of undecided voters were able to determine that yes was the vote that would retain the longstanding status quo where gay and lesbian couples were not allowed to marry.
Perhaps the Yes on 8 campaign made a stronger and more memorable impression on undecided voters than did the No campaign.
Perhaps voters—even self-described undecided voters, as measured in telephone polling—are not actually undecided on this topic in the same way that they are undecided on other topics. A voter who is undecided in a contest between two candidates, for instance, may be entirely indifferent or entirely uninformed. But perhaps almost no set of voters is entirely uninformed on the matter of marriage.
Whatever the reason, the result was that No on 8 did not benefit from inertial voting. In future elections, the pro-marriage side will not have to compensate for the loss of inertial voters. Unlike the wrong-way voters, the inertial voters did not boost No on 8’s vote total.
By the end of the campaign:
No on 8 was on the air forty-four days with thirteen ads. Combined, the thirteen ads aired 11,424 times;
Yes on 8 was on the air thirty-seven days with six ads. The six ads combined aired 11,300 times;
Yes on 8 was on the air with almost double the No on 8 buy in the first half of October;
No on 8 then caught up and outspent Yes on 8 by a margin of better than 2 to 1 in the closing week of the campaign.
See Appendix E for a complete description of the TV buy and for brief descriptions of all the campaign ads.
Voting on Prop 22 in 2000
Before Prop 8, California voted once before on the issue of marriage for same-sex couples. On March 7, 2000, Prop 22 (also known as the Knight Initiative after its best-known sponsor, State Sen. Pete Knight) passed by this vote:
Yes: 4,163,673 votes; 61%
No: 2,909,370 votes; 39%
The electorate in March 2000—a relatively low-turnout primary election; just over 7 million voted on Prop 22—was obviously different from the one in November 2008—a high-turnout presidential general election; just over 13.4 million voted on Prop 8. But the closer margin in the second election as well as polling data from a variety of sources support the conclusion that the LGBT community and our allies were closer to being able to win an election on marriage in 2008 than they were eight years earlier.
Votes required to win in 2012: 6.3 million. That number and all of the following calculations are based on estimated voter turnout of 75% of those registered. We chose 75% as the projected voter turnout by averaging the turnout in the last three presidential elections (2000, 2004, and 2008), all of which were highly competitive nationally, though not in California, in the presidential race.
Expected voter turnout: 12,682,180
Estimated ballots to be cast on same-sex marriage: 12,365,125–12,682,180 (the range reflects a potential 2.5% ballot drop-off as seen in 2008)
Estimated anti-gay no votes: 6,664,802–6,835,695 (based on a 53.9% win projection through correcting for wrong-way voting from 2008)
Estimated pro-gay yes votes: 5,700,323—5,846,485
Votes required to win in 2014: 4.1 million, based on 54.8% voter turnout. This projected turnout is an average of similar nonpresidential elections from 1998, 2002, and 2006.
Expected voter turnout: 9,266,446
Estimated ballots to be cast on same-sex marriage: 9,034,785–9,266,446 (the range reflects a potential 2.5% ballot drop-off as seen in 2008)
Estimated anti-gay no votes: 4,869,749–4,994,614 (based on a 53.9% win projection through correcting for wrong-way voting from 2008)
Estimated pro-gay yes votes: 4,165,036–4,271,832
The bottom line: we need approximately 73% more votes to win in a presidential election year than in a nonpresidential year.
The presidential year electorate may be more favorable to same-sex marriage, on a percentage basis, as polling shows more support than that of the electorate in a nonpresidential year; eg, younger voters are among the least likely to vote in an off year and are among the most supportive of same-sex marriage.
But it may be easier to win in a nonpresidential year because the absolute number of votes needed to win is so much lower. This would be particularly true if the LGBT community and its allies identify a significant percentage of the 4.6 million individual voters supportive of same-sex marriage needed to win, and if they develop the capacity to turn them out to vote in an off year.
Another way to think about it is this: in 2008, we were 1,000,000 votes behind Yes on 8; we lost by 1,000,000 votes. In a future presidential election year, we go into the election roughly 1,000,000 votes behind—perhaps a bit less if voter opinion has moved a bit in our favor, and if we have persuaded some voters to change their minds. In a future nonpresidential year, we go into the election roughly 730,000 votes behind—perhaps a bit more if the electorate includes fewer voters likely to participate who are favorably disposed toward us. Nevertheless, it might be easier to win an election where we start out somewhat more than 730,000 votes behind—even 800,000 or 850,000 votes behind—compared to one where we begin 1,000,000 votes behind.
I am flagging this here because I know it is of great general interest. But I am not recommending a specific year to go back, because (a) we have work to do before we’re ready to go back, and I cannot currently foresee when that work will be completed, and (b) the topic deserves further research that puts it beyond the scope of this report.