For decades, party registration has been a strong marker of political identity in American life, even though not all states have party registration.
Just over half of voters live in one of the 30 states that allow voters to register with a party. In other words, around 124 Million Americans can choose to register with the Republican or Democratic Party.
Today, roughly 85 million Americans — about 37% of U.S. voters — have marked their formal party registration as a Republican or a Democrat. Party labels have long served as a durable signal of political identity. Once voters chose a party, that designation typically remained a reliable indicator of their civic identity.
That assumption is becoming less true.
New evidence suggests a growing gap between the party voters are formally registered with and how they actually identify politically. This shift reflects changing party brands, evolving coalitions, and a more fluid relationship between political identity and institutional affiliation. The result is a form of partisan misalignment that has implications not just for elections, but for how brands, institutions, and strategists understand the American public.
Party coalitions in the United States were relatively stable for roughly three decades prior to the mid-2010s. But recent political developments have reshaped both major parties.
The rise of Donald Trump significantly altered the Republican coalition, while the Democratic Party’s traditional base has been challenged by new and more progressive currents. At the same time, both parties have tapped into broader populist sentiment, accelerating changes in their public image and ideological positioning.
As party brands evolve, voters do not always update their formal affiliations. Many remain registered with a party they joined years earlier, even if their political identity has shifted.
This creates a growing gap between institutional party labels and how voters see themselves.
Party registration remains a useful indicator of political behavior, but it is less predictive of partisan identification than commonly assumed.
Our recent survey research finds that only about 56% of voters registered within either major party identify with that party when asked directly. Registered Democrats are slightly more likely to identify with their party than registered Republicans, but the differences are modest.
Misalignment exists in both directions. A meaningful share of registered Republicans identify as Democrats when surveyed, and a smaller but still notable share of registered Democrats identify as Republicans.
Even participation in partisan primaries, while still a strong signal of party loyalty, does not fully resolve the gap. Looking at primary voting history improves alignment, and is a large part of what Tunnl uses to inform our party framework, but it still leaves a significant share of voters whose formal behavior does not match their stated identity.
In short, traditional indicators like registration and primary participation capture partisan behavior, but they do not fully capture partisan identity.
Several forces help explain the widening gap between party registration and party identification.
Party brands are changing faster than voter identities. Many voters formed their partisan attachments under earlier versions of the parties and have not formally updated their affiliation as party platforms and coalitions evolve.
Political identity is becoming more situational. Voters respond to specific candidates, issues, or cultural signals rather than maintaining consistent loyalty to a party label, especially among younger voters.
Institutional behavior is sticky. Changing registration requires effort, and many voters simply never revisit a decision made years or decades earlier.
The result is a political environment where formal party affiliation may reflect a voter’s past alignment more than their present worldview.
This growing misalignment suggests that American politics may be less stable and more fluid than traditional partisan categories imply.
First, party coalitions may be more volatile than they appear. If many voters do not strongly identify with their registered party, shifts in messaging, leadership, or policy emphasis could produce faster changes in political support.
Second, the electorate may be less rigidly polarized in identity than commonly assumed. Strong partisans certainly exist, but a significant portion of voters occupy a more ambiguous space between formal affiliation and personal identification.
Third, future presidential elections will likely play a major role in redefining party brands. Campaigns in 2028 and beyond will not simply compete for votes: they will compete to redefine what each party represents and who belongs in its coalition.
These changes have implications far beyond campaign strategy.
Brands, media organizations, and institutions frequently rely on stable partisan categories to understand public opinion, consumer behavior, and cultural alignment. But if party labels are becoming weaker indicators of identity, those assumptions may need to be reconsidered.
Political identity influences consumer behavior, institutional trust, and social attitudes. When identity becomes more fluid, public opinion becomes harder to segment and predict. Organizations that rely on static political categories may misread their audiences or misunderstand emerging social trends.
More broadly, the weakening link between party registration and political identity signals a shift in how Americans relate to institutions themselves. Party affiliation, once a defining social marker, may be becoming more symbolic and less determinative of behavior.
Traditional measures of partisanship — registration records, voting history, and demographic patterns — still provide valuable information. But they increasingly need to be supplemented by dynamic measures that capture how voters currently identify and how their attitudes evolve over time.
Understanding American politics today requires recognizing that party identity is no longer fixed. It is evolving alongside shifting party brands, changing coalitions, and a political environment defined by rapid cultural and institutional change.
The labels remain. But what they mean and who they represent is becoming less certain.
Tunnl has a deep history of approaching partisanship with rigor and nuance. Our perspective, what we call our party framework, has always combined extensive survey research, careful study of voter file data, a rich training dataset, and a sophisticated machine learning methodology to fill in unknown or incomplete partisan signals. Rather than relying on a single static indicator, we identify partisanship using a blend of observed data and modeled survey insights. This creates a dynamic method of assigning partisan alignment and behavior that adapts as coalitions and identities shift.
The framework itself creates the partisanship spectrum that rolls up into 3 easy to use categories of Democrat, Republican, and Independent.
Over the last several months, in light of what we’ve seen shifting, we’ve taken an even harder look at how partisanship is assigned across the voter file.
The reality is that party identity is not captured cleanly in any one signal. Registration rules vary state-to-state, and registration itself often lags behind how people actually behave at the ballot box.
Because of that, we are updating the party framework we work from to prioritize observed voting behavior first, before leaning on modeled party scores. When the data shows us what someone does, such as primary participation patterns or cross-cycle voting behavior, we treat that as more meaningful than their registration status if they have one. Modeled scores remain critical for filling gaps and ensuring national completeness, but for Tunnl they will always complement, not override, clear behavioral evidence.
This shift reflects a broader philosophy at Tunnl: in a fluid political environment, measurement frameworks must evolve alongside voter behavior.
Party identity is becoming more dynamic, and the tools used to understand it must be equally adaptive.
As party labels grow less predictive on their own, our approach ensures that partisanship in the Tunnl Platform reflects real-world behavior first, and modeling where it adds value second, allowing our clients to operate with a clearer, more current understanding of the American electorate.