The State of American Division

What the book has argued

Twenty-four chapters of evidence. Three meta-claims.

First, the country has polarized along multiple axes simultaneously, but not at the same rate or on the same items. Some divides are older and have stabilized (the gender gap, the Black-coalition-Democratic baseline). Some are middle-aged and growing (the diploma divide, the cultural-attitude partisan gaps). Some are very recent and dramatic (the post-Dobbs Supreme Court split, the post-COVID Republican retreat from science, the Trump-era White-voter realignment). "Polarization" as a single phenomenon is a useful summary but is too coarse to describe what the GSS shows. The country has polarized in specific ways during specific periods on specific items, and the contemporary partisan landscape is the additive result of fifty years of those distinct movements.

Second, the polarization mechanism is mostly asymmetric movement rather than mutual entrenchment. On every item this book has documented where partisan gaps have widened, the two coalitions have moved at different speeds. Sometimes (homosex, racial attitudes) both moved Left, with Democrats moving faster. Sometimes (financial satisfaction, police force) Democrats moved while Republicans held. Sometimes (military, environment) Republicans moved while Democrats held. The standard "two sides dug in their heels" framing of polarization is rarely what the data shows. The standard "one side moved and the other stayed put" framing is closer but oversimplifies. The accurate framing is "both sides moved, in opposite directions, at different speeds, on different items, through different decades — and the cumulative result is a much wider partisan gap on virtually every item than existed in 1974."

Third, the contemporary partisan coalitions are ideologically sorted in a way the 1974 coalitions were not, and this sorting is the structural feature that makes contemporary American politics different. The 1974 GSS showed 25% of Democrats identifying as Conservative and 19% of Republicans identifying as Liberal — substantial cross-pressured voters in both parties. The 2024 GSS shows 11% and 6% — small marginal populations. The within-coalition cross-pressuring that produced the compromise legislation of the post-war era no longer exists in the contemporary electorate. The reason American legislative politics has become so much harder than it was in the 1960s and 1970s is not that the average voter has become more extreme; it is that the cross-pressured voters who used to mediate within each coalition have been sorted out of it.

What the book has not been able to show

A short list of important findings that the book has gestured at but cannot fully evidence from the GSS catalog.

The geographic distribution of the realignments. The GSS's regional coding is too coarse to evidence the Southern realignment of the 1970s-1990s or the Rust Belt realignment of the 2010s-2020s directly. The aggregate White-voter swing of §10 is the GSS's signal; the geographic distribution comes from other sources.

The within-coalition sub-divides. The book has argued in §14 that the contemporary Democratic and Republican coalitions contain meaningful sub-coalitions with cross-cutting positions on authority, religiosity, and identity. The GSS does not directly measure these sub-coalitions, and the inferences the book has drawn from specific-item disagreements are suggestive rather than direct.

The mechanism of attitude change at the individual level. Survey data shows what people report; it does not show whether their reports reflect underlying attitude change, partisan-motivated perception, social-desirability shifts, or some combination. The book has flagged the financial-satisfaction presidential-control effect (§7.1) as evidence that partisan-motivated perception is a non-trivial component of self-reported attitudes; the same caution applies to most well-being and life-satisfaction questions in the GSS.

The role of partisan media in producing the post-2010 acceleration. The 2010-2024 leg of nearly every chart in this book has been faster than the 1990-2010 leg. The political-science literature has substantial work on the role of Fox News, MSNBC, conservative talk radio, and the post-2010 partisan social-media ecosystems in producing this acceleration. The GSS does not directly measure media-consumption patterns, and the book has had to defer to that literature rather than evidence the claim directly.

Three predictions

The book closes with three modest predictions about the next decade of American political-attitude change. None is a confident forecast; each is a "what the contemporary trajectory suggests, if nothing major intervenes" projection.

Prediction 1: the diploma divide will continue to widen. The §8 chapter on educational realignment showed that the contemporary college-vs-non-college Democratic-margin gap has been widening through 2024 with no sign of reversal. The non-college share of the electorate has been falling slowly over fifty years (more Americans have college degrees in 2024 than in 1974), but the partisan implications of that shift have not yet stabilized. The contemporary Republican Party is winning a larger share of a slowly-shrinking demographic group. Whether that arithmetic works out for Republicans in the long run is one of the more contested questions in contemporary American politics; the GSS data does not, on present trends, suggest a reversal.

Prediction 2: the partisan gap on race-related questions will continue to be the largest in the GSS. The §4.4 chapter on racial attitudes documented a 48-point partisan gap on the discrimination-as-explanation question — the largest gap on any item in this book. The Democratic-coalition shift toward structural explanations of racial inequality has accelerated since 2018. The Republican-coalition rejection of those explanations has intensified. There is no contemporary signal that the gap will narrow; the demographic composition of the parties suggests it will widen further as the Republican coalition becomes more White-dominated and the Democratic coalition more diverse.

Prediction 3: the contemporary "post-liberal" intellectual movement on the Right and the "online progressive" intellectual movement on the Left are the within-coalition sub-coalitions that will shape the next decade of partisan politics. The §14 chapter argued that within-coalition disagreements are where the next political changes will happen. The post-liberal movement is wrestling for the soul of the Republican coalition; the online progressive movement is doing the same on the Democratic side. Which sub-coalition wins each fight will shape the platforms of the two parties more than any single between-party shift will. The GSS data cannot directly evidence which side will win, but the demographic trajectories of religious vs. secular conservatives and college vs. non-college progressives suggest that both sub-coalitions face uphill arithmetic against the secular-Republican and the older-college-Democrat tendencies they are opposed to within their own parties.

A closing note on data and humility

This book has been built on one of the longest, most-cited, most-rigorous survey archives in American social science. Every chapter has been written with the rule that no claim is allowed in the prose that the data files do not directly support. Where the data showed something different from what the proposal that started this project expected, the data won. The audit pass documented in the book's commit history caught roughly forty numerical issues where the prose had drifted from the underlying numbers, and each was fixed.

That said, even the GSS has limits. It measures self-reported attitudes, not behavior. It collects survey responses, not deliberation. It captures aggregate distributions, not individual mechanisms. The chapters of this book have tried to be honest about those limits — flagging where the catalog could not evidence a claim (geography in §10 and §13), where the mode shift might be a confound (every chapter's caveat box), and where partisan-motivated perception might be confounding the response data (§7.1 most directly).

What the book has shown is the contemporary political-attitude landscape of the United States, mapped against fifty years of comparable survey data, with all the major partisan and demographic findings documented chapter-by-chapter. Read together, the picture that emerges is more layered than the contemporary discourse usually allows for. The country has polarized; the partisan coalitions have sorted; the gaps on most issues are wide and growing. But the mechanism of the polarization is asymmetric, the timing differs across items, and the demographic and ideological dimensions of the contemporary divide are not reducible to a single axis.

Acknowledgments and method

The data this book is built on comes from the General Social Survey (GSS), conducted by NORC at the University of Chicago since 1972. The aggregated graph-ready JSON files were pre-computed using design-based survey methods (Taylor-series linearization on stratified, clustered samples with the GSS's official weights) and stored in a Tigris-backed object store. The chart components are part of the ontopic-viz-components registry. The book's writing process — the audit-then-correct discipline that produced the final prose — is documented in the commit history at github.com/ctzn-pub/ctzn-pub.

The book's central debt is to the GSS team and the half-century of survey methodologists, NORC interviewers, and survey respondents whose work produced the data. The contemporary American political-attitude archive exists because they continued the project, wave after wave, through every administration, with the methodological discipline that makes the data trustworthy. The book is a small attempt to use what they produced to describe what has actually happened.

Citation: Smith, Tom W., Michael Davern, Jeremy Freese, and Stephen Morgan. General Social Surveys, 1972–2024 [machine-readable data file]. Principal Investigator, Michael Davern; Co-Principal Investigators, Tom W. Smith, Jeremy Freese, and Stephen Morgan. NORC ed. Chicago: NORC, 2024.