Unpacking Nuances Behind Party Trust Questions in the Polls
Over the weekend, a new national survey was released by ABC News/The Washington Post conducted by Ipsos using their probability-based KnowledgePanel. In general, the topline findings were consistent with most recent national surveys: Donald Trump’s approval rating is at a historic low in their polling (net -33; 33 percent approve – 66 percent disapprove) and Democrats lead by 8 points on the national generic ballot among both likely voters (52 percent Democrats – 44 percent Republicans) and all registered voters (50 percent Democrats – 42 percent Republicans).1 The share saying the country is headed in the right direction is just 26 percent, which is the lowest in polls commissioned by ABC News and the Washington Post in 15 years.2 Perhaps most troubling for Republicans and promising for Democratic electoral prospects is that by 24 points, respondents would rather see the next Congress controlled by Democrats “to act as a check on Trump” (59 percent) than controlled by Republicans “to support Trump’s agenda” (35 percent), up from a net 13 points from February 2025.
My eyes gravitated toward another set of questions in the survey that are frequently asked in political surveys and that I’ve been thinking about a lot lately: which party is more trusted to handle different issues facing the country? They asked party trust on 12 different issues, with Democrats holding a net advantage on 11 (crime was the one exception). The Washington Post’s characterization of the article read as:
The dissatisfaction with Trump inevitably has spilled onto a Republican Party that has been remade in his image. Registered voters said they trust Democrats over Republicans to handle the economy by 40 percent to 33 percent, the first Democratic advantage on the economy in Post-ABC polling since 2014. Voters split about evenly on this question in April and July of this year. Democrats also hold double-digit advantages on trust to handle inflation and the cost of living.
While the characterization is one story that is technically accurate, it obscures another critically important story about the seesaw nature of our national elections and politics over the last two decades: pluralities give neither party a trust advantage on the most important issues facing the country, including addressing the cost of living, curbing inflation, or reducing corruption in government. A plurality also said the same on the situation with Israel and Gaza, and fewer than 40 percent give either party a trust advantage on the situation with Iran, taxes, crime, transgender issues, or immigration). Notably, neither party has a majority advantage on any of the 12 issues tested in the survey.
Other recent surveys show similar results about the net advantage Democrats have on trust to handle issues, but methodological decisions by polling organizations are painting two different pictures of how robust these trust advantages are.
From my perspective—which wants Democrats to win big in the midterms in a couple weeks—it is critically important that they recognize the fragility of their trust advantages on key issues and remain vigilant and focused on them should they win back one or both chambers of Congress, and as they shift toward the presidential nominating process next year. Surveys conducted online reveal much lower trust in both parties to handle pressing issues—perhaps incidentally—than “gold standard” polls, and in this one specific way, they are more reflective of the current state of our fractious politics.
Most people generally know that different polling organizations take different approaches to sampling and recruitment of respondents. Some utilize live interviewers who conduct surveys on the phone, others recruit respondents via text message to take an online sample, others use proprietarily developed probability-based panels (like Ipsos or AP-NORC), and some use opt-in online panels where participants sign up to take surveys for an incentive. Most people are also probably aware that there is a long-standing frustration with polls that fail to push respondents toward revealing how intensely they feel about a given question, the most notable example being on vote preference questions in surveys that don’t capture leaners and thus have larger shares of undecided voters. In the case of party trust questions, both these dynamics are heavily in play: mode changes the survey response options available to participants, and also impacts the decisions of pollsters to push respondents toward one party or the other.
I took four other recent surveys from AP-NORC, Fox News, Quinnipiac and The New York Times/Siena University that also measured party trust on a number of the same issues in the last few weeks and compared them by these dynamics:
Mode: Ipsos’ survey for ABC and The Washington Post and AP-NORC’s survey are both online surveys that use probability-based panels (KnowledgePanel and AmeriSpeak, respectively). Quinnipiac and The New York Times/Siena University conduct all interviews by phone; Fox News conducted most of their interviews by phone, but also had text-to-web recruitment which accounted for about 25 percent of the interviews conducted.
Questionnaire design: As a consequence of Ipsos and AP-NORC being online surveys, they have designed their questionnaires to have explicit options for trusting both parties equally or trusting neither party. In the other surveys, respondents are offered only the options of the Democratic Party and the Republican Party, and have to volunteer both, neither, or that they are unsure.3
The chart below shows the combined percentage of those saying they trust either the Democratic Party or the Republican Party more on a set of issues. In online surveys that do not push respondents and/or offer explicit “both” or “neither” response options, roughly 30-40 percent of survey respondents are saying neither party has a party trust advantage. In live interviewer/text-to-web surveys where respondents are being pushed without a “both” or “neither” option explicitly offered, far fewer are volunteering that they don’t trust either party more than the other.
The next chart now shows the net Democratic advantage on these issues in the same five surveys, just calculated by those who trust Democrats more minus those who trust Republicans more without consideration of those who give neither party an advantage. If you were only to look at this chart, you would all find they generally tell the same story because the net advantages are quite similar between all of the surveys: Democrats have a massive advantage on health care, a double-digit lead on the cost of living and inflation, smaller advantages on corruption in government, and a roughly even split between the parties on immigration.
However, reporting the nets alone can be quite misleading if you only consider the universe of surveys that do not include explicit options to allow respondents to say they do not trust either party. When looking at the actual shares saying they trust the Democratic Party or the Republican Party on these issues across all of these polls, those with explicit response options that allow survey takers to say neither party has an advantage are 18 points lower on average for each party.
While I went through the exercise of looking across different polls with different approaches to respondent recruitment and sampling, you can also observe this by looking at the long trend of the ABC/The Washington Post’s own polling. In 2022, they migrated from conducting live-interviewer telephone surveys and started using probability-based panels where respondents completed interviews online. If you were only to look at their trend without that context, you would think both parties fell off a cliff starting in 2023; in reality, their methodological shift drove the seemingly precipitous declines for both parties on trust.
This chart is intentionally kind of ugly, and it should be noted that the volatility in 2022 is because they conducted surveys both using live interviewers and online, but the rise in no party trust advantage starting in 2023 and beyond is clearly reflective of their complete shift to online panel surveys.
The primary reason I think it matters to dwell on this—even if the overall net advantages are roughly the same across the polls right now—is that what is being revealed within these party trust questions in online surveys aligns with two things: (1) the chronic and deepening distrust in institutions and most Americans feeling inadequately represented by their elected officials, and (2) the rise of “double-haters”—those who are unfavorable toward both political parties. The Washington Post noted this in their write-up of their survey:
One-quarter of Americans dislike both major political parties, but as often happens, these “double-haters” swing against the party in power, favoring Democrats by 58 percent to 32 percent in support for Congress.
For a group that has been growing in size over time, and who consistently vote against the party in power, Democrats are likely to win the support of “double-haters” in these midterms as Trump won them by 20 points in 2024, Republicans won them by 17 points in 2022, Biden won them by 21 points in 2020, and Democrats won them by 17 points in 2018. Yet, the common characteristic of their recent voting behavior is to reject the incumbent party in power, and having a large share of voters giving neither party a trust advantage will continue to make them fluid to whoever is in power for the foreseeable future.
The party trust advantages Democrats may be consistent across surveys, but failing to measure the true share of those not trusting either party can result in overstating how strong and enduring party trust of Democrats actually is.
Developing strategies to cultivate trust on issues among those who trust neither party seems like an obvious strategic imperative for Democrats, but it also requires consistently measuring how delicate their current trust advantages actually are, and how easily those advantages could dissipate should they win control of Congress.
1 One place where this survey slightly deviates is that there is no difference between registered voters and likely voters; most surveys show an expansion of the generic ballot lead in Democrats’ favor among likely voters.
2 In November 2011, only 22 percent said the country was headed in the right direction. This survey was conducted by Langer Research Associates, not Ipsos. This was also long enough ago that it was somewhat credible for the survey to ask a presidential horserace quesetion between Barack Obama and Herman Cain (Obama +5 among likely voters, for the record!).
3 Fox News actually asks a follow-up if respondents initially do not align with one party or the other to try to force them to choose; however, they unfortunately do not report the results of the initial ask in their publicly available data.
