No on 8’s “O’Connell” and “Thorons” ads were our most effective attempts to counter “Princes” and to confront the opposition’s exploitation of prejudice.
The two most effective ads made by the No on 8 campaign are the only two that directly take on the two main issues on voters’ minds during the campaign: same-sex marriage and kids being taught about gay people in school.
· “O’Connell” was the only ad that directly rebutted Yes on 8’s “Princes” ad and its assertions about children.
· “O’Connell” moved some voters to support No on 8. It seemingly did not, however, change their view on the issue of same-sex marriage.
· “O’Connell” aired October 22–28. As soon as it was off the air and replaced by No on 8’s “Feinstein” ad, voters reversed course and some moved back to the Yes side.
· No on 8 was beginning to gain a slight spending advantage at the time “O’Connell” aired, but that doesn’t explain the ad’s success. No on 8 ads that aired after “O’Connell” were much bigger beneficiaries of its growing financial edge; No was outspending Yes 2 to 1, 4 to 1, and even more in the final days. Yet none of the post-“O’Connell” ads had a measurable impact on voters. By itself, No on 8’s lopsided spending advantage at the end couldn’t make its ads effective.
· The only other No on 8 ad that may have had an impact on voters was the “Thorons” ad, the first to air and the only ad that directly talked about LGBT people and marriage.
The only ad that the polling data directly links to increased support for No on 8 is “O’Connell.”
None of the No on 8 ads had anything like the impact of “Newsom” or “Princes.” As Chart 18 demonstrates, once Yes on 8 opened up its lead after “Princes” had penetrated, it retained it for the rest of the campaign. This is true even though No on 8 aired thirteen ads over that time, including five for which it bought very substantial airtime. See Appendix E for the list of all ads and the subset of widely seen ads.
Chart 18: Measuring Yes's Lead—The Margin of Yes Minus No
The impact of “O’Connell” was slower and smaller than that of “Princes.” It is small enough that it is not obvious on the standard horse race, displayed earlier in Chart 11. But Chart 18, the margin of support between the Yes and No campaigns, shows an improved situation for No on 8 beginning October 23, shortly after “O’Connell” began to air, just as it was beginning to be seen by a large number of voters. No on 8 initiated no other ad or major voter communication in those intervening days—“O’Connell” was its principal ad on the air at that time—making it highly likely that the improvement was derived from “O’Connell.” At this point, No on 8 had not yet firmly established the financial edge that allowed it later to greatly outspend Yes on 8. So it was the “O’Connell” ad itself, not a spending advantage, that improved the situation for No on 8 among voters overall.
Chart 18 shows that Yes on 8’s lead over No on 8 peaked just before “O’Connell” went on the air, and then declined as “O’Connell” became widely viewed. Yes peaked both on the standard horse race question and the follow-up “be clear” question, and slipped on both questions after “O’Connell” went up and became widely seen. Yes slipped significantly more on the horse race question than on the “be clear” question, which is consistent with the message of “O’Connell.” In other words, voters who saw “O’Connell” were significantly more likely to switch their vote to No on 8 than to change their view on the issue of same-sex marriage. This is additional evidence that “O’Connell” was the campaign event having the impact on voters, since “O’Connell” was not about marriage. Like “Princes,” which it rebutted, it was about the issue of kids. So it didn’t persuade as much on the issue of same-sex marriage as it did on the separate issue of how to vote on Prop 8. To view “O’Connell”, go to the No on 8 Ads page.
Additional evidence of the value of “O’Connell” comes from an analysis of inertial voters. The inertial voters are those who were undecided on the “be clear” question; they didn’t know where they stood on same-sex marriage. Despite being undecided on same-sex marriage, most of these voters did have an opinion on how they planned to vote on Prop 8. That is, most of them were planning to vote Yes or No on Prop 8; they were not undecided on Prop 8, even though they were undecided on same-sex marriage. The 1.37 million inertial voters made up approximately 10% of the electorate. The inertial voters as a group were larger than the margin by which Yes on 8 won.
The term “inertial voters” is a new one coined for this report because there is not a term of art to refer to this very specific type of undecided voter. The inertial voters are not undecided in the usual sense of the word; most of them told the pollster which way they planned to vote on Prop 8. They were, however, undecided on the substantive issue of same-sex marriage.
In a ballot measure more typical than Prop 8, there are many, many inertial voters. For example, think of the large number of voters who don’t have a clear view on an obscure topic or a little-publicized bond issue. The conventional wisdom among political consultants is that inertial voters are likely to favor the status quo as their default position, which on most measures means they tend to vote no. In the case of Prop 8, however, a yes vote was more likely viewed as a vote to continue the status quo. Same-sex marriages had been legal in California for only a very brief time, and the ban on same-sex marriages that existed before June 2008 was longstanding. The term “inertial voter” captures the idea that this type of undecided voter tends to vote to retain the status quo.
The Lake Research tracking polling identified two types of inertial voters: those who voted absentee (starting October 6, when absentee voting began), and those who as of October 30 had not yet voted. Both types of inertial voters mostly broke the same way throughout October, when “Princes” was on the air but not yet directly rebutted by “O’Connell.” From October 8 through 16, both types of inertial voters broke in favor of Yes on 8 by a large margin. Additionally, inertial voters who had not yet voted—by far the larger group of inertial voters—continued to break toward Yes on 8 by a large margin through October 20.
“O’Connell” went on the air October 22. As “O’Connell” began to air, both types of inertial voters trended in the opposite direction for the first time, breaking by a lopsided margin in favor of No on 8. See Table Q below for the day-by-day numbers.
No on 8 continued to take the bigger share of inertial voters through October 28, when “O’Connell” went off the air and was replaced by the “Feinstein” ad. At this point, inertial voters reversed course, and once again broke in favor of Yes on 8 by a wide margin. The trends among inertial voters strongly suggest that “O’Connell” boosted No on 8 when it was on the air, and its replacement by other ads did the reverse.
The only evidence casting doubt on the impact of “O’Connell” comes from a portion of the tracking polling done by David Binder Research. No on 8 hired Binder to conduct tracking from October 21 through Election Day. From its first tracking poll on October 21 through October 25, Binder shows No on 8 losing ground essentially at the same time as “O’Connell” is becoming widely seen. On the other hand, Binder also shows No on 8’s fortunes improving October 26, 27, and 28, when “O’Connell” was still the only No on 8 ad on the air in a significant buy. There is no easy way to reconcile the conflicting tracking information, particularly because there is no earlier Binder tracking to provide a fuller comparison to the body of the Lake tracking and because the Binder tracking itself points in both directions. Ultimately, the smaller quantity of Binder tracking makes it a more limited tool for evaluation of the campaign’s performance.
“O’Connell” succeeded even though No on 8 had only an erratic and small financial edge when it aired. Later No on 8 ads, with even more money, did not duplicate its success.
No on 8’s fundraising improved greatly throughout October. As a result, its ability to air its TV ads increased throughout the month. The improvement was not steady, however. It came in stages by fits and starts.
During stage 1, from October 1 through October 15, Yes on 8 outspent No on 8 on TV every day. On six of those fifteen days, the two sides were fairly close, with No reaching at least 80% of Yes’s spending. On the other nine days, however, No’s spending averaged less than 50% of Yes’s spending. The Lake tracking polling shows that once “Princes” was widely seen, No on 8 was consistently 10 points behind Yes on 8 during this period.
During stage 2, from October 16 through October 27, No on 8 outspent Yes, but No couldn’t consistently hold onto a significant money edge. On six of the twelve days—almost every other day—spending between the two sides was almost even. On the other six days, however, No spent 50% to 100% more than Yes. The Lake tracking polling shows that Yes maintained a steady lead over No through this period but that No began to close the gap after “O’Connell” was widely seen. “O’Connell” was No’s only ad on the air from October 22 through October 26, the time when the tracking polling showed improvement. During this time, No picked up 5 points to move from 37% to 42%, while Yes declined 2 points from 52% to 50%.
Finally, during stage 3, beginning on October 28, No on 8 established financial dominance that continued through Election Day. Even during this period, however, No’s spending would rocket up one day—No outspending Yes 2.5 to 1 or 4 to 1, and on Election Day 33 to 1—but then return to a 20% spending advantage the next. Only in the final four days, starting November 1 was No on 8 overwhelmingly dominant every day. The Lake tracking polls pick up very little change after that point through their conclusion on October 30. The Binder tracking polling, which continued through Election Day, showed No on 8 gaining ground over a two-day period—October 31 and November 1—and then staying at the same level of support through Election Day. (Binder’s tracking showed No on 8 ahead of Yes on 8 from November 1 on, and predicted a win for No on 8. In general, if Lake’s tracking polling was somewhat pessimistic, Binder’s tracking was overly optimistic, showing No on 8 leading on October 21 and 22, October 27 and 28, and then November 1 through 4.)
In summary, from October 16 on,
· No on 8 ran its ads more frequently than Yes on 8, and ran more ads.
· The Lake tracking polling showed that the new ads other than “O’Connell” aired without improving polling numbers further until No gained an overwhelming spending advantage, better than 2 to 1 daily.
· The Binder tracking polling suggests the possibility that “Feinstein,” which went on the air October 28, might have had a small impact; it would have been most widely seen at the time that Binder shows a blip upward for No on 8 on October 31 and November 1. The next No on 8 ad, “Internment,” did not air till October 30 and would not have been widely seen until November 1, the point at which Binder shows no further improvement for No on 8. The even later No on 8 “Obama” ad similarly cannot be associated with improvement in any poll.
The weight of the evidence shows that No on 8 benefited from “O’Connell” and that “O’Connell” was No on 8’s most effective ad. That said, its impact was much smaller than “Princes.” The “O’Connell” ad’s relative weakness suggests that the pro-LGBT campaign either needs to (a) rebut the kids argument more quickly, or (b) develop a more effective rebuttal than “O’Connell,” or (c) both. Otherwise, ads like “Princes” will continue to dominate same-sex marriage campaigns.
For the day-by-day comparison of the Lake and Binder tracking polls, see Appendix D under “David Binder Research’s Standard Horse Race. For the day-by-day comparison of Yes on 8 and No on 8 spending, , and to see all ads, see Appendix E.
Outspending the opposition can boost our ability to make a successful rebuttal argument. But its effectiveness is limited. A spending advantage isn’t enough to stop the opposition’s message from getting through. We can’t drown them out if they have bought 500 gross rating points (GRPs) or more of airtime per week per media market. Conventional wisdom among media consultants is that this size buy is the minimum necessary for memorability and impact; it means that the average voter in the media market is exposed to the campaign’s ad or ads five times during a week.
Consultants recommend 1,000 GRPs per media market week to ensure having an impact. But for the anti-gay forces, the minimum can be enough to prevail when they run highly polarizing ads on the incendiary topic of danger to kids. This dynamic occurred in Maine in 2009. The pro–same-sex marriage No on 1 spent 140% of what Yes on 1 spent on TV advertising. But that did not impair the ability of Yes on 1 to get its message out. When No on 1 ran a phone bank testing our ability to persuade undecided voters to vote No, our team called a small sample of voters identified as undecided earlier in the campaign and scored as most likely to break toward No (having a “marriage score” of 90 or more out of a possible 100). Of this group that should have been very favorably disposed toward us, 29% ended the conversation planning to vote No; 44% planned to vote Yes; and the remaining 27% were concerned about the kids issue, and we were in danger of losing them. To see all the ads aired in Maine by both sides, and to see the persuasion script and memo to the No on 1 campaign manager, see Appendix R.
No on 8’s experience suggests that outspending the opposition by a similar margin can be enough to attract more voters to the pro-LGBT side, with an ad like “O’Connell” that directly rebuts the opposition’s argument. This was true even though the “O’Connell” rebuttal came late. This explanation most plausibly explains the body of evidence available.
Ultimately, No on 8 benefitted at the end to a limited degree from its late-in-the-game financial advantage when the advantage became overwhelming. (No on 8 also realized some important economies by purchasing some of its late-stage ad time prior to August at a discounted rate.) But given the success of Yes on 8 at raising money, and the improbability that pro-LGBT campaigns of the future can count on a 2 to 1 or 4 to 1 financial advantage over the totality of a campaign, money alone is not going to be enough to solve the problem with “Princes.”
There is limited evidence that “Thorons” had an impact on voters as well
The clearest indicator that a Prop 8 TV ad may have impacted voters was when its debut in a significant (greater than 500 points) buy coincided with a sharp change in the daily tracking polling. Charts 1 and 2 above provide this kind of evidence for the efficacy of the “Newsom” and “Princes” ads, for example.
But even when there is no sharp change reflected in the daily tracking, sometimes other polling data can suggest that an ad had impact. This is the situation with “Thorons,” the first ad aired by the No on 8 campaign, where a mom and dad (Ms. and Mr. Thoron) talk about their lesbian daughter.
Chart 19 shows that both Yes and No voters remembered the “Thorons” ad, even when it was off the air, and recalled it, unprompted, when asked about any TV ads they had seen on the issue. As late as October 21, a third of all voters polled recalled the ad. Chart 20 shows that a majority of voters recalled the ad when prompted by a description of it. These are consistent with the David Binder post-election poll that showed “Thorons” to be the most remembered No on 8 ad after the election was over, even though it was the first No on 8 ad broadcast.
Chart 19: The Percentage of Yes, No and Undecided Voters Recalling "Ad About Parents Wanting Their Gay Daughter Treated The Same" Unprompted
Chart 20: Percentage of Voters Who When Prompted Say They Have Seen the "Thorons" Ad
Chart 21: Effect of "Thorons" Ad—All Voters (more or less likely to oppose Prop 8)
For voters who remembered seeing the “Thorons” ad when prompted, Charts 21 through 24, show what they said about the impact of the ad on their thinking, whether it made them more or less likely to vote No on 8.
The most provocative of these are charts 22 and 24. Chart 22 shows that among those planning to vote Yes on 8, many more reported that the “Thorons” ad made them more likely to vote No; by comparison, few said that “Thorons” made them less likely to vote No. The persuasiveness of “Thorons’” reported by Yes voters grew every day between October 8 and October 12 (all the days for which we have this data). On October 8, an equal number of those planning to vote Yes reacted favorably and unfavorably to “Thorons” (19% for each point of view). But by October 12, 14% of the Yes voters said “Thorons” made them more likely to vote No, and only 8% said “Thorons” made them more likely to vote Yes.
Chart 24 shows a similar trend among undecided voters. On October 8, an equal number of undecideds said that the “Thorons” ad made them more likely and less likely to vote No (8% for each position). But over the next three days, the number who said they were more likely to vote No jumped to 15% and then settled at 12%, whereas those who said “Thorons” made them less likely to vote No had fallen to 5%.
On the other hand, here’s the evidence that the “Thorons” ad was not helpful to No on 8. From September 29 through October 2, Lake Research polled on the impact of “Thorons” by asking whether the ad made voters more likely to favor or oppose same-sex marriage. This is different from asking voters whether they were more or less likely to vote No or Yes on Prop 8, in the same way that the Lake “be clear” question was different from the horse race question. For voters who remembered seeing the “Thorons” ad who were already planning to vote Yes, 6% said the ad made them more likely to support same-sex marriage, and 32% said the ad made them less likely to support same-sex marriage. Among those who were already planning to vote No, 30% said the ad made them more likely to support same-sex marriage, and 6% said the ad made them less likely to. Among undecideds, it was a wash: 7% said the ad made them more likely to vote No, and 7% said the ad made them less likely to.
Putting all of this together, both Yes and No voters reported that the “Thorons” ad made them more likely to vote No on 8, but not more supportive of same-sex marriage. But “Thorons” is the only No on 8 ad that aired that explicitly made an argument on behalf of same-sex marriage. If that argument wasn’t having an impact on the merits of the issue, then its explicit content was not affecting voter opinion. It may have made voters more likely to vote No without having changed a single mind on the issue of marriage. Alternatively, it may be that all that registered from “Thorons” was that “Thorons” was the first ad urging a No vote. If that’s correct, that’s probably not enough to make “Thorons” an ad that could ever affect the electoral outcome.
On the other hand, for Yes voters to report that “Thorons” made them more likely to vote No at the time that “Princes” was already on the air may mean that something substantive from “Thorons” registered, even if it was not sufficient to persuade voters to alter their view on same-sex marriage. Also, since the effect of “Thorons” grew over time, it may be that later polls would have found “Thorons” affecting voters’ views on the issue of marriage if Lake Research had continued to ask that question.
On balance, there’s enough evidence that “Thorons” had an impact on voters favoring No on 8 to support more research on its message approach in the future. Campaigns in Utah, Texas, and Maine have also made ads featuring parents of grown LGBT children, and several of these have substantially more emotional punch than “Thorons.” See Appendix E for all of these ads. It is possible, though far from certain, that an improved version of the “Thorons” ad could have a directly measurable impact on voters in a future campaign.