Does Support for Abortion Rights in Red and Battleground States Signal Trouble for Republicans?
May 16, 2022
Last week I was at the American Association of Public Opinion Research (AAPOR) annual conference. I was hoping to get more than one post out, but it was busy – and I learned a lot. The research folks are doing on likely voter models, nonresponse problems, and some election polling innovations is fascinating. In fact, it has helped me think through the question posed for today’s post in ways I didn’t expect when I started drafting it last week.
It’s unclear if the abortion issue will be enough to redraw the political map. Perhaps it will fade, as it seems to have in Texas. But the stakes are not small for Republicans in this region: The predominantly white working-class voters who swung from Barack Obama to Donald J. Trump in the 2016 presidential election tended to back abortion rights.
In a postelection study, 58 percent of voters who flipped from Mr. Obama to Mr. Trump in 2016 said that they would support a law that would “always allow a woman to obtain an abortion as a matter of choice.”
Will the impending demise of Roe v Wade and the moves by Republicans to outlaw abortion and contraception imperil the chances for Republican Congressional candidates in this year’s midterm election? There has been widespread outrage over the draft Supreme Court opinion that was leaked two weeks ago, which suggests that this decision might significantly motivate voters to support Democrats this fall.
However, the short answer to this question is probably no. Why? For starters, have you seen the polling on gun control? The public broadly supports a number of gun control measures (although there is some evidence that support is decreasing), like waiting periods and bans on military-style weapons, but yet gun nuts win elections again and again and again. So, I am not optimistic that voters really care about the issues they say they do in polls since the results show so many responding to grievance politics and racist appeals from Republicans. If the outrage over two dozen first graders being murdered didn’t translate into votes for gun control candidates, it’s difficult to see how outrage over Roe’s demise will translate in to votes for pro-choice candidates. There is, however, a less cynical reason for why issue polling does not translate into electoral success, which we will discuss below.
Despite my cynical take above, choice is one issue that for a number of very good reasons resonates across the political spectrum. Where some people sincerely think guns are necessary for self-protection, most Americans think of abortion as such a personal decision that the government taking it away can feel like an attack not just on women or personal autonomy, but on our very idea of self-governance. So, this could be a very different case than gun control – even with unprecedented levels of trauma nationally from gun violence.
Just this month, Pew Research surveyed over 10,000 Americans and found that 61% said that abortion should be legal in most or all cases, while 37% disagreed. This breakdown has been largely consistent since 1995, when Pew found 60% said abortion should be legal in most or all cases, while 38% disagreed. When it comes to popular opinion on abortion, nothing much has changed in nearly 30 years (except the partisan gap). Not only have opinions not changed on this subject in three decades, but a large majority consistently supports legalization.
Source for this chart is here.
The problem, however, is that issue polling is not electoral polling. To provide some context to these considerations, let’s return to issue polling on gun control. It turns out we have a recent example of how issue polling does not translate into support at the ballot box.
In 2016, Nevada voters passed a ballot question to provide background checks for certain firearm purchases. The question passed by just over 50% - which means, obviously, that it came close to losing. Yet, public opinion issue polls in Nevada had previously shown support for such a measure at 86%. How could any ballot question nearly fail when it has the support of nearly nine in ten voters? Well, the fact is that the question probably never had that much support to begin with. Does this mean the issue polling was wrong – and had an error of over 30 points? It looks like that, but that’s not the problem. The issue poll was probably correct. The problem is, I believe, twofold.
First, campaigns matter. I am a bit of a broken record on this, but it demands repeating. A well-funded, strongly-messaged, aggressive campaign can make the difference in electoral outcomes. It won’t turn a hopeless campaign into a winner, but it can make up the difference even when starting with a double-digit deficit and particularly where there are a significant number of undecided voters.
Second, the Nevada ballot question was, in fact, asking a different question than what the pollsters asked in the issue polling. This insight is something I gained at last week’s AAPOR conference from the work of several researchers, but in particular the work of Jonathan Robinson, John Sides, and Christopher Warshaw. They looked at several issues that poll highly to see whether there is actually that much support for enacting laws to accommodate the policy changes through ballot questions. They posit that there are two ways to design (“frame”) issue questions that are relevant for use on a ballot: (1) by asking the respondent/voter whether or not they support the policy/issue (akin to a Likert scale survey response) or (2) by adding a “status quo frame” that some states do currently use on ballot questions. The first frame is the standard way of asking a question about an issue (“Do you support background checks for gun purchases? Yes or No?”). The second frame qualifies the question by asking the respondent/voter when they “support/oppose to changing the law” (“Do you support or oppose a change in the law to require background checks for gun purchases?”). The difference in language construction, as it turns out, is important.
What Robinson, Sides, and Warshaw reported1 is that support for popular issues, such as gun control (but not, oddly, cannabis legalization), drops when the status quo frame is added. Opposition does not appear to increase, but suddenly we see a lot of “undecideds.” Those undecideds are likely to vote no on the ballot question when the time comes, unless they have moved to “yes” in the meantime.
The reason for this appears to be similar to the undecided problem we see for Democratic presidential nominees. Democrats usually gets what they poll, but – when they win the popular vote – the margin is often closer than the reported polling margins. This is because those margins are usually created by a significant number of undecideds, who seem to break to the Republican at the last minute. We can’t say for sure why this happens, but considering that (at least rhetorically) the Democratic candidate is often the candidate of change while the Republican candidate embraces the status quo (or a ”return” to it) it is plausible to suggest that these voters, while either intrigued by Democrats or repulsed by Republicans, decide when they no longer can put it off that they prefer a “safe” choice to something new or bold.
In Nevada, it is likely that many of those 86% never gave a serious thought about what enacting a law might look like. The “this-law-is-irresponsible/impractical” political communication frame often used by opponents of ballot questions likely nested in the back of a lot of voters’ minds and when it came time to vote they did not vote against gun control as much as they just decided to keep things the way they are. Fortunately for the public, not enough did this to defeat the 2016 Nevada ballot question.
Let’s apply a made-up discount for issue polling and apply it to the abortion polling gap in the states to see if it means trouble for Republicans this November. If I recall correctly, in the AAPOR presentation the researchers found an eight point drop in support for a hypothetical popular issue when using the status quo frame. We simply do not know what an accurate estimate would be for a similar change in support on abortion rights across many states. To be conservative about it, let’s consider five and ten point drops in support from the “mostly legal” respondents in the abortion polling when it comes to the ballot box. Five seems reasonably prudent although less than an eight point change. Ten points gives us a broader picture of the possibilities, but it too may be too conservative for the electoral reality.
In Nevada the drop was 35 points, but remember that was for a single issue ballot question. For individual Congressional candidates, abortion rights may be the most or least important issue for voters but for many it will be only one of several considerations weighing on their votes. Voter preferences can change depending on what issues seem salient in a particular election – and that could be true for abortion this year. However, the partisan polarization we see today in American politics suggests that for most people abortion will not trump all other concerns – particularly if that means having to vote for a Democrat for a lot of Republicans. So, it would not surprise me at all to see that, at least in some red states, this discount is way too low.
The table below ranks the red and battleground states with polling advantages for abortion rights support. I included Missouri as well, which has equal numbers of support and opposition for keeping abortion “mostly legal.” For example, in Ohio there is a ten point margin of support in favor of keeping abortion “mostly legal.” The two right-hand columns subtract five and ten points respectively from that advantage.
As you can see even the five point discount wipes out the issue polling advantage abortion rights advocates have in Ohio, Iowa, North Carolina, Georgia - states that we might expect from the current issue polling could spell trouble for Republicans this fall. But there are other indications on this table that abortion rights will not cause the Republicans too much trouble at the ballot box. Florida has an 18 point issue polling advantage, but there is no electoral polling in the Sunshine State right now that suggests this is helping Democrats at all. Likewise, with Alaska and its 25 point issue polling advantage. Neither of those states is likely to become anymore competitive for Democrats than they already are.2
What does this mean for abortion and the midterms? Well, none of this is directly analogous. The AAPOR research concerned ballot questions, not candidates. The situation where a lot of voters are polling undecided and then voting for the Republican at the end of the campaign is mainly a presidential campaign phenomenon. Regardless, it does not seem unreasonable to assume that the furor over abortion rights today will not translate into pro-choice candidates defeating Republicans in the midterms. But this analysis, contingent as it is, is no reason for despair; consider it more of call to action. The furor over the release of the draft decision to overrule Roe has been more serious than conservatives expected, and the leaking of the decision will not prevent more furor when the actual decision is released.3
Let’s not forget: campaigns matter.
To be fair to the researchers, I am trying my best to capture what they presented but I do not currently have access to their slides or paper to check how accurately I am doing so. Any error is mine alone and not the researchers. I hope to write more about this once their paper becomes available.
This could be incorrect. Perhaps abortion rights will upend the politics in both states. Time will tell.
It is still possible that the justices will ultimately decide to tweak Roe to allow Mississippi’s 15-week ban rather than overturn the 1973 decision altogether. This could mute some of the anger we have so far seen.
If pro-choice advocates really want to codify the protections of Roe into law, then they should be using all of their energies into passing constitutional amendments on these Red/purple states and use them as a galvanizing issue for voters in the same way Republicans used anti-marriage amendments in 2004.