Independent polling sees promising opening for Yang & Whitman’s Forward Party

Independent polling of voter sentiment by e.pluribus.US found promising opportunities for the new third party recently launched by former presidential candidate Andrew Yang and former New Jersey Governor Christine Todd Whitman.

85% of respondents polled across the US advocate seeking compromise with political opponents, vs. working harder to overcome them.

A second interesting finding: not a single survey respondent expressed support for the predominant current approach toward engaging with political opponents.

With surprising consistency, the pro-compromise result held true in red, blue and moderate communities — including in the most extreme left and right-voting districts — in urban, suburban and rural communities, and across the north, south, east and western US. Even in the district with the nation’s most extreme right-voting congressional representative, over 70% of respondents reported prefering compromise.

Figure 1 shows the percent of respondents supporting compromising with, vs overcoming political opponents, across ten regions polled. Support for compromise was consistent and overwhelming.

Ironically, the polling organization, e.pluribus.US, was actually attempting to discover a polling outcome that revealed greater support for overcoming political adversaries than for seeking compromise with them. Based upon the partisan tenor of current public political discourse, our hypothesis was that we would find a public strongly in favor of defeating political opponents.

We found the opposite. Initial returns showed over 85% of respondents advocated compromise. We then extensively iterated the targeting, phrasing, participant selection process and other variables in an attempt to elicit a more confrontational response.

Through these iterations we found that it made little difference if we targeted left vs. right-leaning districts, more extreme vs. more moderate voting representatives, whether the compromise was in abstract or on specific controversial issues, de-personalizing the individual that would compromise to be a third party, identifying vs concealing the sponsor and purpose of the survey and even going so far as demonizing the other party that must be compromised with as one that “obstructs things.” Over 70% of respondents advocated attempting a compromise outcome even with political opponents that were themselves obstructing.

Ultimately we were able to extract an anti-compromise response only when conspicuously specifying that compromise would threaten the respondents’ key valued principles. Even then, we’ve only yet identified that anti-compromise response in the US’ most extreme left and right districts (we haven’t yet tested that last phraseology in more moderate districts).

Our polling methodology was created to understand rough directionality of sentiment, for a separate project e.pluribus.US is pursuing. Method details are provided at bottom, but notably, this approach was not designed to rigorously refine results down to small margins of error. There therefore do remain uncertainties regarding precise sentiment and follow-up polling is encouraged to clarify these.

But the responses we received so heavily leaned in one direction that, even with a high margin of error they remain highly instructive and in many ways fascinating.

Interesting findings:

Overwhelming support exists for compromise with political opponents.

Across the first ten districts we sampled, in aggregate 85% of respondents broadly advocated seeking compromise with opponents, rather than working harder to overcome them. We asked this question over a dozen different ways. Figure 2 displays a sample subset of questions we tested.

Both the left and right are majority pro-compromise.

Preference for compromise exists in both red and blue communities. Indeed, even in the districts whose representatives have — literally – the most extreme right and left voting records in the current Congress, over 70% of respondents advocated for compromise. (According to VoteView these districts were Texas’ 29th District on the Left and New Mexico’s 2nd on the Right.) Figure 1 above shows results across the first ten communities sampled. As respondents were able to express a preference across an array of five choices, we were able to calculate a “Compromise Score” for each region, presented in Table 1. The score indicates the weighted average response in that region on a 1 to 5 scale where 1 indicates 100% advocated doing “much better” at overcoming opponents and 5 indicates 100% advocated for doing “much better” at compromise. Regions in the table are sorted from most left-leaning to most right. Note none of the scores are below the 3.0 level that would indicate majority preference for overcoming opponents.

Interestingly, in the last iteration of the poll, wherein we finally elicited a net anti-compromise response, said preference for overcoming opponents was identical on both the left and right (as represented by, according to VoteView, the highly right-leaning TX District 13 and highly left-leaning Illinois District 9). Figure 3 illustrates the identical responses from those districts. When key values were threatened, in TX-13 75% preferred overcoming opponents and in IL-9, 74% did so. This offers guidance to policy officials structuring compromise-seeking efforts on the importance of avoiding explicitly threatening key values.

Pro-compromise sentiment extends even to controversial specific issues.

Expanding upon the result mentioned immediately above, one may wonder whether, in practical application of policy, the obstacle to compromise is that all specific issues are considered “key values.” It turns out that majority support for finding compromise does appear to exist on specific issues, even the most controversial. In one region, we tested whether support for compromise varied by whether we phrased the object of compromise as “politics in general” or alternatively one of three specific issues: abortion, gun laws and economic policy. In that test we found majority support for compromise on all four topics, as shown in Figure 4. That specific test was in only two districts in Colorado, but the strength of results suggests they may hold up across the broader population.

No one is happy with status quo.

Perhaps the most interesting finding? No one — zero respondents — support the status quo approach our leaders currently take toward dealing with political opponents. Ten of our twelve polls allowed the respondent to select across a usually five-point range of “confront vs compromise” options, with the middle point always being the status quo, some version of “leaders are doing fine already.” Across all our polling, not a single respondent chose that option. Not one. Skeptics of the Forward Party, take note.

Interpretation of results.

These results may not surprise political science researchers. Prior studies using more rigorous analytical methods have evidenced that voters tend to be more moderate than the Congress. Other studies suggest many voters who advocate for compromise as an abstract concept lose their temperance when “push comes to shove” on specific, controversial issues. That said, we remain struck by the glaring disconnect between the appearance of intransigence in the national political dialogue and these poll results that — even allowing for their limitations — suggest a shocking degree of hunger for finding compromise resolution with political opponents.

We believe two explanations are most likely.

First, representatives are “out-of-sync” with their constituencies, but the latter can’t do anything about it. This is not to suggest the representatives are dumb or do not understand their constituencies. Simply, while hundreds of thousands of voters in a district represent a range of views, usually expressed as a bell curve, as an individual a representative can only occupy one point on that range at a time. While it may seem advantageous that they vote in a way consistent with the center of that bell curve, from a campaign politics standpoint there are strong incentives for their records to diverge away from that center point, and for them to publicize this quite loudly. For one, they need to differentiate themselves from challenger candidates and that is difficult to do if all candidates are bunched in the center. Second, the voters at the extreme tend to be the most motivated, advocating with the greatest force, donating the most money, helping most passionately on the campaign trail. So if a candidate wants to optimize support, they are wise to energize their extreme supporters and that requires legislative behavior supportive of positions more distant from the center.

The result of all this is, while an underlying population may be willing to engage in compromise to achieve progress, their representative is almost certainly less likely to do so. Indeed that representative is incentivized to do quite the opposite: to whip up divisive behavior in order to energize the electorate. “Fear and anger get people to the polls.” And that results in a constituency with a broad range of mostly pro-compromise views having to choose from among a tiny selection of nearly always non-compromising candidates.

Second, further to the point above about the disproportionate volume of the extremist voices, if one is listening to the national discourse, one is more likely to hear those extremist viewpoints. Additionally, media today, being polarized and eyeball-driven, has its own incentive to sensationalize the perception of discord and conflict.

This is all to say that it may indeed be that the majority of the electorate does strongly support compromise as a tool to move policy goals forward, but that we simply don’t see that sentiment represented by our elected officials, nor in the prevailing national dialogue.

If true, it suggests a promising opening for the Forward Party.

What’s next?

e.pluribus.US works to overcome the divisive partisanship preventing the nation from getting things done. This polling is an element of our Project INFLUENCE, which seeks to find a way to influence attitudes toward cooperating with opponents, initially through social media messaging. We needed a method to measure whether our messaging campaigns succeed at that. We created these polls as part of the process of developing that measurement tool. Next steps for Project INFLUENCE are to identify an appropriate subject community, use the measurement tool to baseline that community’s current attitudes toward cooperation, then begin testing our messaging campaigns on them.

Survey methodology.

It was not our intent to execute a polling campaign that absolutely minimized margin of error. We are not trying to answer these questions with the highest possible degree of precision. We are instead trying to efficiently and rapidly develop a tool that will tell us with actionable accuracy if a target community’s attitudes toward compromise have meaningfully changed due to an influence campaign. That said, the results returned from this type of polling can reveal confident insights as long as we view them in the context of the limitations of the system. For example, if a poll done with this methodology returns a result that leans 90% in one direction, it is exceedingly unlikely the underlying reality is 90% in the opposite direction.

The polling was of social media users on Facebook, Instagram and Linked-In from May through July 2022. We ran campaigns at specific geographic zones, for instance, perhaps a congressional district. We then targeted ads (multiple iterations of creative, two are illustrated in Figures 6 and 7 at bottom) at all users within that zone, inviting them to anonymously express their opinions about (various wordings of) “compromise vs confrontation in politics,” via an off-site poll. Clicking on the ad would take them to a poll executed through SurveyMonkey.com (though Colorado polls were executed through Instagram’s in-feed polling feature). Once on the poll page they were asked some variation of “As a nation do we need to do a better job of (a) overcoming or (b) compromising with our political opponents?” They were then offered generally a five point scale of radio buttons to respond, ranging from “Much better overcome” to “Much better compromise.” Figure 5 shows a representative range of response options.

We continued a poll until its outcome became obvious. In some cases this was rapid; for instance, if 8+ of the first ten responses were all pro-compromise, it is exceedingly unlikely we would learn anything new from the next ten, so we would stop that poll evolution and move on to the next. In all, thus far we have reached over 100,000 users and captured 323 responses.

Known meaningful limitations

• ● Skews associated with online-only audience. There are demo- and psycho-graphic skews associated with the survey pool being online-only. We do not believe this significantly invalidates the directional conclusions of the results.
• ● Selection bias: Survey respondents self-selected to participate. In an ideal survey they would be randomly selected so the results would not be skewed by any bias inherent in people who are more likely to engage with the subject topic and/or polls in general. For the purposes of what e.pluribus.US is trying to do, this is not a disqualifying factor. If we are able to measure a change in attitudes among those more likely to respond to surveys, it is reasonable to conclude we have done so for the broader population we messaged. However, for purposes of most accurately evaluating the percentage of voters who support compromise, the selection bias in the methodology we used is noteworthy and one would want to eliminate it in any attempt at more robust polling on this topic.
• ● Miscellaneous: We sampled more right than left-leaning districts because initially we assumed that we’d find more anti-compromise sentiment on the right. That turned out to be incorrect. We also did not sample any districts in the Northeast. There was no particular reason for this, but we do not believe there is a meaningful chance that adding a northeast district at this point would alter conclusions. *For the South Carolina poll we observed minor corruption in the response pattern and partially corrected it, but suspect that an inaccurate 5% or so skew away from from compromise remains in that specific poll.

.