What Can Ballot Initiatives Defending Abortion Rights Tell Us About Public Opinion on Abortion?
By Jonathan Robinson
Ballot initiative results have often surprised political observers, especially when public opinion polling seems strongly in their favor ahead of a vote. While a policy might poll extremely well in a jurisdiction, how voters ultimately cast their ballots is another matter.
In research with Chris Warshaw of George Washington University and John Sides of Vanderbilt University, we find that polls tend to overestimate support for ballot initiatives on issues where opinion is lopsided, regardless of whether opinion tilts in the conservative or liberal direction. In other words, the most popular liberal and conservative policies tend to underperform the most at the ballot box. In our analysis, we examine more than 200 ballot initiatives and paired polls on a variety of issues over time. However, in the aftermath of the Supreme Court’s ruling in Dobbs v. Jackson Women’s Health Organization, which overturned Roe v. Wade, I want to highlight specific findings related to abortion to help pro-choice advocates working on this critical issue.
Below is a graph of how much specific initiatives over- or under- performed their poll numbers. Most of the abortion-related initiatives were promoted by conservatives on policies where the liberal position in polls appears weak in the leftmost quadrant of the chart. The horizontal or x-axis of the graph shows how polling characterized support for the liberal position on the specific abortion policy being put on the ballot generated by a statistical methodology called MRP (multilevel regression with poststratification), which accurately projects data from national surveys to produce estimates of opinion in lower level geographies (like states). Meanwhile, the vertical or y-axis of the graph shows the difference between how the initiative performed compared to where the polls had the policies polling at in the state.
One example of such initiatives in this dataset is so-called “parental notification” or “parental consent” policies targeting women under the age of 18 who seek abortions. In 2011, Gallup found that 71 percent of voters favored such a policy, and 27 percent opposed it. Even more recently, the Pew Research Center found that even among the 61 percent of US adults who say abortion should be legal in most or all cases, 56 percent favored parental notification. Conservatives and anti-abortion advocates had hoped to use these ballot initiatives as a wedge to split liberal support, but instead, voters tended to reject these initiatives. Overall, conservatives lost 65 percent of the abortion-related initiatives in our database, despite the overwhelming majority of such initiatives stemming from conservative groups targeting relatively conservative voters in states where their policies poll well.
In fact, all abortion initiatives that conservatives proposed in our database underperformed relative to what polling would suggest, by an average of 16 percentage points.
We hypothesize that status quo bias may make voters less likely to follow through on stated positions in polls. This might be a result of a general aversion to change among voters with less strong views on an issue or who are cross-pressured with competing partisan or ideological signals on how to vote on an issue they are less familiar with or sure about. It may also represent an unwillingness to take individual-level responsibility for changing a law, even though a voter might support such a change if it were presented to them as coming from a state legislature, for instance. To study and understand the exact mechanisms at work here, we collaborated with Data for Progress on a survey experiment to test the effect of emphasizing that a certain policy the voter is being asked about — in this case, parental consent/notification for abortion — would be a change to the current set of laws and policies. We compared asking the question that way to asking it in a way that most pollsters and researchers normally do, which emphasizes opinion about an issue in a vacuum and not in the context of supporting a specific change in the law. We also included a question mirroring language from a 2006 initiative in California on parental notification/consent for abortion. Below is question-wording for the three questions:
Standard Issue Question Treatment: Do you support or oppose requiring that minors wait to have an abortion until 48 hours after a doctor notifies their parents or legal guardian (except in cases of medical emergency)?
Priming Change Treatment: There is a new proposal in your state to change current law about whether minors can have an abortion. This proposal would require that minors wait to have an abortion until 48 hours after a doctor notifies their parents or legal guardian (except in cases of medical emergency). Do you support or oppose this proposed change to the requirements for minors to obtain an abortion?
Ballot Language Treatment: Below is a proposition that could appear on a referendum in your state in the next year.
● Change your state's Constitution to require minors to wait to have an abortion until 48 hours after the doctor notifies the minor's parent or legal guardian. Allow the doctor not to notify the minor's parent if it is a medical emergency or the parent has waived their right to be notified.
● Permit courts to waive parental notification based on evidence of a minor's maturity or best interests. Require doctors to report any abortions performed on minors.
● Authorize damages against physicians for violating notification requirements.
● Require minor's consent to abortion, with exceptions.
Do you support or oppose this proposal?
The results show that asking the survey question in the language of the ballot initiative wording and the emphasis on what the policy would change reduces support for both parental consent/notification laws by 6 to 7 percentage points on average (the two lines on the graph are comparing the two “treatments” of the question to the “control”/standard issue question text). These changes are large and statistically significant, and could easily make the difference between an initiative failing or succeeding. Campaigners should keep these findings in mind when evaluating external polling, conducting their own polls, and estimating vote targets. Emphasizing that progressives want to protect the status quo on abortion is a powerful argument for abortion advocates and seems to be a big part of why we can beat back anti-choice initiatives, even in conservative jurisdictions.
Jonathan Robinson is the Director of Research at Catalist, a progressive data utility that maintains the longest running voter file outside the two major parties. It works with unions, progressive advocacy groups, and Democratic campaigns.