Measuring Trust Among Southwestern Pennsylvania’s Realigning Voters
Over the course of the past month, the RealClearPolitics Institute of News & Information (RCPINI) has engaged in a lengthy, in-depth study of southwestern Pennsylvania and the determinants of political trust in the region. As a reminder, the ten counties that we study are Allegheny, Armstrong, Beaver, Butler, Fayette, Green, Indiana, Lawrence, Washington, and Westmoreland. We’ve studied them via a partnership with Emerson Polling, one of the most accurate polling companies in the country.
After laying out the scope of the project, we examined trust as a general matter. Next, we looked at the rural/urban divide and the education divide. Today, as befits RealClearPolitics, we examine the politics of the region, both in general and how those politics relate to trust.
As a brief matter of annotation/nomenclature, reporting the error margins on finding becomes tedious quickly. To avoid this, I’ll forgo reporting +/- 3.5% after every reported value, and adopt the more academic approach of simply calculating the result for you. So if you see something that is 13.2% (9.6%, 16.7%) that simply means that it is 13.2%, +/- 3.5%. With rounding, if you add and subtract 3.5% from 13.2%, you would get 9.6% and 16.7%. Note too that these are Bayesian calculations, so even if the error margins overlap, we may have a high degree of confidence that, say, Trump led among independents.
This is a worthy endeavor because, as mentioned in the opening piece, southwestern Pennsylvania has been the site of one of the more stunning political transformations. In 1988, when Michael Dukakis came 2 points away from carrying the state, he won only two counties in the eastern half of the state: Philadelphia and Lackawanna (Scranton). This wasn’t a matter of Dukakis holding down margins in eastern Pennsylvania either; George H.W. Bush won each of the four suburban Philadelphia counties with around 60% of the vote. In western Pennsylvania, however, it was Dukakis who won half of the counties in the region with around 60% of the vote, including the most populous, Allegheny. In Greene, Fayette, and Washington counties, tucked into the heart of Appalachia in the southwestern corner, he won 64% of the vote.
Today, the script is flipped. Kamala Harris won Allegheny County by 20 points, but lost the other counties in the area by at least 20 points. Her competitiveness in the state hinged on (barely) carrying Lackawanna County, and then running up large margins in heavily populated Philadelphia, Montgomery, Chester, and Delaware counties.
To see how Trump did it, we can examine various registration groups and see how they voted. One thing that immediately pops out: Trump managed to split off more Democrats than Harris did Republicans. This is not entirely surprising, as ancestral Democrats in the region voting increasingly for Republicans is a well-explored phenomenon. Nevertheless, the scope is still jarring to witness. Trump peeled off 13% of registered Democrats in the region (10%, 17%). Harris, on the other hand, managed to convince only 5% of registered Republicans to vote for her (3%, 7%). Trump probably won independents as well, with Harris winning 32% (25%, 40%) to Trump’s 40% (33%, 48%). Put differently. Harris’ coalition was 84% registered Democrats (80%, 87%) while Trump’s was 75% Republican (71%, 79%).
When we dive into issues of trust, some surprising results appear. Trump’s voters appear more trusting overall than Harris voters. Voters with a high degree of trust in the U.S. voted strongly for Trump, giving him 53% of their votes (43%, 63%) to Harris’ 32% (23%, 42%). This is surprising because multiple surveys have shown Trump voters to have lower degrees of social trust. We could be witnessing two things. First, Trump voters with low degrees of social trust may simply be unwilling to respond to the survey, skewing the sample. The other possibility is that this reflects Trump voters’ enthusiasm over the Trump administration. In other words, the causal mechanism may be reversed: Being a Trump voter may lead to a high degree of trust in the U.S., rather than the other way around.
Some support for this can be found in other areas of trust that we examine. For example, when we inquire about trust in Pennsylvania, the breakdown is roughly the same: 13% (10%, 17%) of Harris voters report a great deal of trust, compared to 14% (10%, 17%) of Trump voters. For a “fair” amount of trust, it is also balanced: 55% (50%, 60%) of Harris voters fall into this category while 52% (47%, 57%) of Trump voters agree. We do find some evidence for the theory of “missing Trump voters” as well; among the (admittedly small) subsection of voters with “not much” trust, Trump wins 56% (38%, 72%) to 25% (12%, 41%); there is a 98% chance that he came out ahead. If we look at “community trust,” the same pattern repeats.
Going through other institutions reveals unsurprising preferences. Voters who expressed a great deal of trust in the military, police, religious leaders, and ICE broke strongly for Trump. Individuals with a great deal of trust in religious figures gave Trump 59% (51%, 66%) of the vote to Harris’ 25% (18%, 31%), while individuals with a “fair amount” of trust gave Trump 54% of the vote (49%, 59%) to Harris’ 32% (28%, 37%). Individuals with no trust in religious leaders, on the other hand, gave Harris 57% of the vote (48%, 65%) to Trump’s 25% (18%, 33%). ICE was particularly divisive: People with a great deal of trust in ICE gave Trump 83% (78%, 87%) to Harris’ 7% (4%, 11%); the numbers for “none at all” were reversed. Of course, this doesn’t mean that support for ICE caused Trump support – the intensity of support for deporting people in the country illegally likely drove both support for Trump and support for ICE.
On the other hand, people with high degrees of trust – scientists, professors, and journalists – all supported Harris. Support for scientists was strongly divisive, with Harris leading Trump 63% (58%, 68%) to 24% (20%, 29%) among voters with a great deal of trust in scientists. Professors showed a similar split. Perhaps notably, among voters reporting no trust in professors, Trump led 82% (75%, 88%) to 7% (3%, 12%).
Trust in business leaders, teachers, nonprofits, and federal, state, and local officials behaved predictably with respect to vote, but they displayed less dominance for one party or the other.
Interestingly, if we compare Trump Democrats to Trump Republicans or independents, some meaningful differences do appear. Trump Democrats were less likely to say that they had a “great deal” of trust in ICE – 30% (18%, 44%) – than Trump Republicans or independents – 52% (46%, 58%). They were also more likely to say that they had a “great deal” of trust in scientists – 26% (15%, 39%) – than their registered independent/Republican Trump-supporting counterparts –17% (13%, 21%). They also appeared more trusting of teachers, professors, and journalists, although the differences weren’t as marked. Overall this suggests that Trump Democrats weren’t just “DINOs” – there really were ways that their preferences more closely resembled Democrats than those of Republicans. A comparison of Harris Republicans to Harris Democrats/independents suggests complementary findings, but the error margins are simply too large to draw reliable conclusions.
Regardless, there are important partisan differences in levels of social trust. While they might not be surprising on their face, the degree of partisan differentiation is surprising.
