Finding 2: The Yes on 8 Ads That Worked

Yes on 8’s Ads Worked

The Newsom ad opened and closed with San Francisco Mayor Gavin Newsom saying, “It’s gonna happen—whether you like it or not.” As much as the words, the ad’s raucous and abrasive tone told voters they were not in control and that others were forcing them to accept same-sex marriage.

The “Princes ad opened with a little girl saying, “Mom! Guess what I learned in school today?” What she supposedly learned is that she can grow up to marry a female. The misleading ideas: impressionable kids are being indoctrinated; they will think it’s OK to be gay and may then experiment with sex or gay sex, be more likely to become gay, or choose to be gay. If any of these transpire, kids will be damaged by learning that gay people exist.

To read more about these ads, go to Appendix E.
The ads are displayed on the Yes on 8 Ads page.
The combination of Yes on 8’s “Princes” and “Newsom” ads moved voters in its direction. In particular, the fear-mongering message that children are in danger affected voters. 

The most effective decision made by either campaign—the one with the biggest impact on the outcome—was Yes on 8’s decision to air the “Newsom” and “Princes” ads back-to-back. “Newsom” caused a spike in voters refusing to say how they were going to vote; “Princes” then drove voters in the direction of yes. The combination of the two put Yes on 8 in the lead.

“Princes” had the greater impact. Its fear-mongering message about children drove Yes on 8 over the 50% threshold. It affected many types of voters, particularly those it most directly targeted: parents with kids under 18 living at home.

Key Lessons from Findings 1, 2, and 3:

·         When voters heard the lie that kids were in danger, some of them believed it.

·         As Finding 1 shows, parents in particular believed it.

·         More voters believed the lie the longer it went unanswered.

·         As Finding 3 shows, some voters no longer believed the lie once they heard it exposed as a lie.


How We Know What We Know: The Campaign’s Polling Data

The Data That Indicate Which Side is Ahead: the Horse Race Question

It’s possible to understand the story of how the “Newsom” and “Princes” ads thrust Yes on 8 into the lead because of the Lake Research polling data from its “standard horse race” question. Here is a quick overview of the horse race question; its strengths and weaknesses; and the ways in which it is complemented by a second question asked by Lake, the “Be Clear” question.

Definition: In its polls, Lake first asked the standard horse race question fairly early in the survey. Virtually identical to the horse race question taken in every poll by every pollster, the Lake standard horse race question asked voters how they would vote if the election were today. The horse race question aimed to gauge which side was ahead and by how much. Lake asked the horse race question by reading each voter the language on the ballot and then asking whether they planned to vote Yes or No on Prop 8.

Advantages: The strength of the horse race question is that it measures the combined impact of multiple phenomena that can affect the final election result, including:

Persuasion: voters changing their minds on same-sex marriage;

Self-correction of wrong-way voting: for voters who misunderstood the meaning of voting Yes or No, learning how to cast a ballot that expresses their opinion on same-sex marriage;

Decoupling: voters deciding that a vote on Prop 8 ought to reflect their view on an issue other than same-sex marriage, eg insulating children from learning about gay people.

The standard horse race therefore gives the clearest snapshot of how the electorate is divided at the time the survey is taken.

Disadvantages: The weakness of the horse race question is that changes in its result do not tell us the cause of the movement, since change could result from any of the above factors or any combination of them. As a result, the horse race is better at answering “where are we at this moment” than it is at explaining “why is this where we are at this moment.”

The Complement to the Horse Race Question: the “Be Clear” Question

Fortunately, Lake did not only ask the horse race question to track voter movement. Lake also asked in almost all of its polls a second question that was not asked by most of the other pollsters, a clarification question, in this report referred to as the “Be Clear” question. “Just to be clear”, the second question began, “is your vote to eliminate marriage for gay or lesbian couples in the state of California or NOT to eliminate marriage for gay or lesbian couples in the state of California.”

The purpose of the Be Clear question was to measure whether voters were voting in conformity with their view on same-sex marriage. Most voters understood how to vote to express their view on the issue; but many, it turned out, were confused about how to vote to express their opinion on same-sex marriage. Finding 7 and Appendix K discuss the phenomenon of “wrong-way voting” in detail.

Chronology: the Lake Horse Race Data on How Yes on 8 Got to 50% Support

1. In mid-September, before either side had TV ads on the air, Lake Research showed a virtual tie.

In mid-September 2008, all of the various polls, even those that disagreed on other matters, found No on 8 at the height of its support. On both of its principal questions tracking movement in voter opinion—the horse race question and the “Be Clear” question—Lake Research found No on 8 ahead by a tiny percentage smaller than the margin of error. The race was a dead heat. The following two charts show the situation. The first gives opinion from May through late October. The second is a close-up of the first and covers the period from mid-September through late October.

Chart 10: Support for Yes and No on 8 Among All Voters

Chart 11: Support for Yes and No on 8 Among All Voters

2. In late September, the horse race question showed that the “Newsom” ad moved a small number of voters directly to Yes and drove up the number who refused to say how they were going to vote.

On September 29, the “Newsom” ad went on the air in a major statewide buy. “Newsom” quickly moved a small number of voters to Yes. It also moved a chunk of voters to refuse to say how they planned to vote on Prop 8. Here are the details.

The two charts above show what happened after the “Newsom” message aired long enough to sink in. They depict the change in the standard horse race question at this point; it shows the first big, abrupt bump in the size of a group of voters. The group of voters that rapidly grew was the one that combined both undecided voters and refusals, those voters who refuse to tell the pollster how they plan to vote.

Chart 12: Gauging the Size of Each Side's Base

Chart 13: Breaking Down "Undecided" and "Refused" for Lake's Polling

Two additional charts, immediately above, are more precise because they separate the undecideds from the refusals and includes only those voters yet to vote. (Absentee voting began on October 6, so some voters had already voted; they are not included in the two charts immediately above). The additional charts reveal that the big bump was not a spike in undecided voters but instead a jump in refusals. Undecideds were steady as refusals increased 500%, from a mere 1% of voters to 6%. Prior to “Newsom,” refusals had been an almost constant 1% in the Lake polls from May through September. By October 13 and through the rest of the campaign, refusals returned to being a fairly constant 1% or 2%. But between September 29, when “Newsom” began to air, and the tracking poll of October 5-7 refusals jumped 5 points. Along with other attrition from the No side, the race was no longer a dead heat: No had lost ground and Yes had the lead.

The more details chart slices the yes and no votes and distinguishes among three kinds of supporters for each side:

·         Those who were strongly supporting each side

·         Those leaning toward each side

·         Those who were undecided but leaning toward each side. This last group was, perhaps counter intuitively separate from the undecided voters (mentioned above). The difference is that these voters expressed when polled at least a slight lean. A voter was relegated to the pure undecided category only when they admitted no preference whatsoever.

This chart shows that among those voters yet to vote, the undecided voters who were leaning toward voting no declined from 6% to 2%. This is likely the source of many of the new refusals.

The combined result of the movement among two groups—the refusals and the undecideds leaning toward No—was that No declined from 43% to 36%, while Yes declined from 47% to 45%.

These shifts support the inference that Yes on 8’s decision to air “Newsom” moved a slice of likely voters to have second thoughts about Prop 8.

3. No other campaign event or decision that occurred in late September or early October other than the “Newsom” ad explains the big shifts detected by the polling before “Princes” had a chance to penetrate.

The October 7 tracking poll shows a big movement away from No on 8 compared to the previous poll, taken less than a week earlier. No other big event favoring Yes on 8 occurred in that brief intervening time.

The October 7 tracking poll, like all Lake tracking, combined results for three days, the two days preceding and the date attached to the tracking poll result. For example, the October 7 tracking poll included results from October 5, 6, and 7. The only other big event favoring Yes on 8 during those specific days was its second ad, “Princes,” which began to air October 6 in Spanish. But “Princes” was largely unseen until October 8, when it went on the air in English; the size of the Spanish-language buy was relatively small. See Appendix E for the details. Similarly, “Princes” received no press reportage until October 8; and the ad did not receive extensive coverage until much later in October, when the declining fortunes of the No on 8 campaign attracted reporters to write about the Yes on 8 campaign ads.  See Appendix F for a more detailed discussion on earned media.

4. Then, “Princes” went on the air, first in a relatively small buy in Spanish and then beginning October 8 in a large buy in English. The full measure of the damage it instigated was not instantaneous. Its impact grew over the course of a week, an apparently steady accretion of support for Yes on 8.

“Princes” began to be widely seen after October 8. It joined “Newsom” in rotation and began to break voters away from refusal to state. Within a week, with “Princes” the dominant ad, voters moved steadily towards Yes. The electorate stabilized with Yes above the 50% level of support needed to win. Here are the details.

Over three days—October 8, 9, and 10, the first days that “Princes” began to be widely seen—the number of refusals plummeted. By October 10, refusals were back to 1%. Most or all had now declared when polled their voting intention on Prop 8. Some migrated back to their former position; but some switched sides and moved in the direction of Yes. Worse, Yes on 8 picked up a percentage point of voter support every other day. By the time “Newsom” and “Princes” were on the air together for a week on October 15, the horse race question showed Yes on 8 with a double-digit lead over No on 8, with 51% of voters planning to vote yes and only 39% planning to vote no.

The only good news here is that “Princes” did not have the ability to move an indefinite number of voters. It was not effective enough to move Yes on 8 significantly above the 50% threshold. But 50% (plus one voter) is enough to win.

The trend with “Princes” from beginning to the end of the week is one in which we can have a great deal of confidence because over seven days of tracking polling, the aggregate sample of voters surveyed was large, reducing the margin of error.

The trend from one day to the next day is one in which we can place less confidence because each daily sample of voters is smaller and the margin of error consequently larger. In other words, the margin of error is large enough that apparent daily progress for Yes on 8 may in fact have amounted to changes in support that actually bounced a bit up and down, trending upward over the course of the week but not necessarily increasing every single day.

The trends over the course of the week, however, are most consistent with the conclusion that the impact of “Princes” grew over time as voters were exposed to it again and again. A few responded to “Princes” right away. More responded to it after seeing it ten times in a week.

Alternatively, it is possible that the “Princes” message resonated with more voters over time as “Princes” was joined on the air by Yes on 8’s next ad, “Massachusetts,” which makes the same argument about kids. It is impossible to isolate the impact of “Massachusetts” with the data available, since its major air dates overlap completely with “Princes.”

In either case, the data support the conclusion that fear-mongering about kids, the sole focus of both ads, drove voters toward Yes and kept Yes at and above 50%. This conclusion is bolstered by the data that show that when No on 8 finally rebutted the kids argument directly in the “O’Connell” ad, voters moved the other way. For these voters, “Princes” seems to have decoupled Prop 8 from the issue of same-sex marriage: Prop 8 for them became a vote on whether kids should learn that LGBT people exist as a normal part of society.

Chart 14: Measuring Yes's Lead—The Margin of Yes Minus No

The chart immediately above, also discussed in Finding 3, shows that the No on 8’s hemorrhaging of support only stops—and even partially reverses—when the misleading information in “Princes” is directly rebutted by the “O’Connell” ad. It shows the margin by which Yes on 8 led throughout, from mid-September, when it had no lead, through late October. The rise from “Newsom” and “Princes” is remarkably linear. Its ascent is only broken after No on 8 airs the “O’Connell” ad on October 22 and it has had a chance to have penetrated, ie to have been seen by voters enough times that they might remember it. “O’Connell” helps No on 8 regain some of the ground lost, but not enough to change the result.5. Overall, the effect of “Princes” was greater than “Newsom.”

The data establish that “Princes” caused most of the erosion suffered by No on 8 during this time. It couldn’t have been “Newsom,” because Yes on 8 began phasing out “Newsom” as soon as “Princes” was on the air. Yes on 8 spent $1.9 million to air “Newsom” the week before “Princes” went on the air. Then Yes on 8 spent only $200,000 on “Newsom” in the debut week of “Princes,” as it was spending $2.1 million the same week to air “Princes” (in Spanish and English combined). By the following week, October 15–21, the buy for “Newsom” was a shrunken $6,000; it went off the air entirely on October 20. That same week, the “Princes” argument about kids and schools represented almost the entire Yes on 8 buy, both in the form of “Princes” and “Massachusetts.” See Appendix E for details of week-by-week ad buys in all media markets.

Chart 15: How Men and Women React to the Different Yes on 8 Ads

From October 11 through 28, Yes polled at or above 50%, the level of support it needed to win. The specific dates suggest that Yes on 8 strategically timed “Princes” to coincide with the beginning of absentee voting on October 6. For those eighteen consecutive days—each a miniature version of Election Day, with both sides banking votes—Yes on 8 daily increased its lead. Once voters cast absentee ballots, they couldn’t change their minds later in the campaign, even if they wanted to; they had voted. Yes on 8’s dominance as people voted absentee was particularly important since in this election, 42% of all California votes were cast before Election Day.

Chart 16: Men and Women, Corrected for Wrong-Way Voting

Yes on 8 spent $4 million airing “Princes.” It was the single most frequently aired TV spot of the entire campaign by either side. There is no other remotely comparable event or expense that occurred on or around October 8–10 that provides an alternative explanation for the abrupt shift among voters detected by the Lake tracking polls (the only tracking polls conducted at that time for the No on 8 campaign). On the other hand, to give “Newsom” its due, not only did it soften up voters in a way that seemed to have paved the way for “Princes”; it also had a particularly strong impact on male voters, compensating for the fact that “Princes” had a particularly strong impact on female voters.

6. No on 8 never recovered.

No on 8 never recovered from the combination of the “Newsom” and “Princes” ads. According to the Lake Research polling, No on 8 never again drew even or close to Yes on 8 on the horse race question. Although both campaigns created a variety of other ads and utilized a variety of other campaign tactics, the tracking polling suggests that none had the combined impact of “Newsom” and “Princes” among swing voters as a whole.

What The Data Show About “Newsom” and “Princes

Throughout the events described above, the Be Clear question did not move in lock step with the horse race question. Voters changed their minds more often and more quickly on the horse race; changes in the Be Clear lagged.

Chart 17: All Voters With and Without Correction for Wrong-Way Voting

The chart above compares the horse race and Be Clear questions and makes the difference between the two visible. One difference is particularly apparent at the time “Newsom” goes on the air. The number of undecideds and refusals on Prop 8 documented by the horse race question (illustrated with the darker line) and the number of undecideds and refusals on the issue of same-sex marriage documented by the Be Clear question (the lighter line) are markedly different. As “Newsom” became widely seen, the darker line shot up; the ad elevated the number of voters refusing to say how they were going to vote on Prop 8, as discussed earlier. But the lighter line stayed flat: the number of undecideds and refusals on the issue of marriage stayed the same, unaffected by “Newsom” or “Princes.” In other words, “Newsom” and “Princes” did not quickly persuade voters on the issue of marriage; voters’ views on marriage did not immediately change, even as their planned vote on Prop 8 changed.

The more gradual changes in the Be Clear question show that not every voter who changes their mind on how to cast their vote has also changed their mind on the issue of same-sex marriage.

The more gradual changes in the Be Clear also suggest that voters as a group are relatively slow to alter their opinion on same-sex marriage. This may reflect the increasingly common discussion of the issue over the past decade and many voters having developed a relatively firmly held opinion on the issue.

The Be Clear question helps reveal voters changing their minds on same-sex marriage. Considering the two questions together reveals self-correction of wrong-way voting and voters’ transforming their vote into one on kids’ safety rather than same-sex marriage, a “decoupling” of the two matters.

The advantage of asking the Be Clear question is that it reveals how much of the movement in the horse race question is due to voters changing their minds on the issue of same-sex marriage.

In addition, the disparity between the two questions quantifies voters’ self-correction of their wrong-way voting and suggests the success of the Yes on 8 message framing at decoupling Prop 8 from the issue of same-sex marriage. The fact that the same phenomenon recurs when No on 8 puts the “O’Connell” ad on the air rebutting “Princes” makes the de-coupling seem all the more likely. A slice of the electorate moved away from No on 8 without changing its view on same-sex marriage when “Princes” went up on the air; and a smaller segment of the electorate, perhaps including some of the same voters, moved back toward No on 8 without changing its view of same-sex marriage when “O’Connell” went up to answer it. Some voters, at least some of the time, did not see their vote on Prop 8 and their opinion on same-sex marriage as having to be congruent or as one and the same.

The non-congruence makes sense if “Princes” worked by stimulating or foregrounding anti-gay feeling and anti-gay prejudice to increase the number of voters ready to vote Yes on 8. My take on the ad is that, on the face of it, this was “Princes’” aim. Voters were encouraged to confuse their concern over the possibility of harm to children with the ballot measure allowing same-sex couples to marry. Marriage was a stand-in for public visibility of LGBT people and/or public approval of gay people as decent human beings equal to everyone else.

By the end of the campaign, Yes on 8 also stimulated significant movement among voters’ views on same-sex marriage

Yet decoupling is probably only one phase of voter persuasion for some voters, not the entirety of what the Yes on 8 campaign accomplished. As the chart on the Be Clear question shows, opposition to same-sex marriage grew from mid-September to late October; the initial 3 point lead grew to 13 points over the course of the tracking polls. Yes on 8 achieved a 10 point swing—it changed the minds of a minimum of 5% of voters—on the issue of marriage itself.

Even after “Princes” was off the air, it remained memorable

Post-election polling by David Binder Research provides evidence that “Princes” was a highly memorable ad even after the campaign was over. Binder Research’s survey showed that between November 6 and November 16, 23% of voters recalled “Princes” as a convincing Yes ad without prompting. No other Yes message or ad was mentioned by more than 3% of those polled. In addition, “Princes” was able to redefine Prop 8 to a majority of voters; 37% of those polled thought a yes vote on Prop 8 would stop the teaching of same-sex marriage in elementary schools, and another 17% were unsure whether Prop 8 would result in a change in how families were discussed in classes.

We are very likely to see future ads similar to “Princes”; we have to prepare

Based on the devastating short-term impact of “Princes” during the Prop 8 campaign, and the similar impact of comparable ads in Maine, it is all but certain that anti-gay forces will use the kids argument in future campaigns. Fears about gays and kids may not be uppermost in the minds of voters absent an anti-gay campaign, or voters may not readily volunteer latent concerns. But the Prop 8 experience suggests that:

·         When an anti-LGBT campaign alleging indoctrination of kids unfolds on TV; and

·         When that campaign is well-funded enough that the average voter sees ads exploiting anti-gay prejudice five or more times each week for four to five weeks; then

·         The ads generate, awaken, reawaken, or reinforce a response among some voters that moves them to vote against the LGBT community.

As a practical matter, anti-gay prejudice prepares some voters to believe the worst, and that’s precisely the spectre conjured up by the anti-gay campaign:

·         When voters hear a lie that kids are in danger, some of them believe it.

·         As Finding 1 shows, parents in particular believe it.

·         More voters believe the lie the longer it goes unanswered.

·         As Finding 3 shows, some voters no longer believe the lie once they hear it exposed as a lie.

See Recommendations 1-3 for work we can do now to reduce our vulnerability to the same kind of anti-gay campaign in the future.