Not a Teammate and Not a Fan: Probing the Identities of Unaffiliated Registered Voters

Abstract

For decades, in most states with a party registration option, the percentage of voters registering as unaffiliated with a major political party has steadily increased. But who are these registered voters in these polarized partisan times, and why might they register without a major party? We address these questions by drawing on parallel large-N original surveys of registered voters in two southeastern states experiencing a notable rise in registered independents but with different electoral rules for unaffiliated registrants. The closed primary rule in Florida reflects a much greater share of major party registrants versus North Carolina, which has a semi-closed primary rule. Nevertheless, even with these different primary laws, in both states we find that the decision not to register with a major party strongly covaries with identity as a political independent. Hence, registration rules may alter registration patterns, but individuals claiming to be less attached to a major party are markedly more likely to manifest this position by registering unaffiliated.

Introduction

In the last few decades much has been made of the widening partisan divisions within the American electorate. In their assessment and review of affective polarization, Iyengar et al. (2019, 130) speak of “partisanship as a social identity.” This emphasis on partisanship in fueling contentious politics is an accurate portrayal of the historically polarized era in which Americans are living. Indeed, spanning the more than 70-year-old American National Election Study (ANES), the latest 2020 survey documented the highest share of strong partisans, at 44.2%. Nevertheless, the flip side of strengthening partisanship is a concomitant growth in political independents. For instance, matching its peak in 2014, 43% of independent identifiers in 2023 is a 35-year high for the Gallup Poll.¹ Hence, partisan polarization may also foster more unaffiliated registrants.

In this short article, we examine who registers as an unaffiliated voter in Florida and North Carolina. These southeastern states contain polarized electorates (Shor and McCarty 2011) experiencing notable population growth and change (Bullock III et al. 2019). Historically presidential battlegrounds (Huang and Shaw 2009), Florida and North Carolina’s substantial contingent of unaffiliated party registrants can determine the outcome of any statewide election.² Apart from the aforementioned commonalities, though, there is a key difference between these states: Florida holds closed primaries and North Carolina holds semi-closed primaries. Thus, in the Sunshine State, there is incentive to register with a major party so as not to be shut out of primary election contests, whereas not registering to vote under a party label imposes no such restriction in the Tar Heel State.

Dating back to the Progressive Era reforms in the late 1800s, with few exceptions, in order to vote Americans must take the step of registering.³ Are Floridians and North Carolinians who choose to register without a party making a conscious decision to disavow the two major parties? The polarization literature conceives of party affiliation as a social identity (Mason 2018), as does influential scholarship on party identification (PID) (Greene 1999, 2000, 2002, 2004; Hawkins and Nosek 2012; Theodoridis 2017). Likewise, with respect to explicit PID, we posit that political independence can be understood as an identity distinguishable from partisan affiliation (Greene 2004). Nevertheless, because there are gradations of partisan attachment, we hypothesize that weak party identification is strongly related to whether someone registers unaffiliated, but that this decision is mediated by a state’s primary rules.

In what follows, we estimate the probability of unaffiliated registration among Floridians and North Carolinians. In the summer of 2023, we administered original online surveys to representative samples of registered Floridians and North Carolinians to understand what factors are associated with unaffiliated registration – which is officially designated as “No Party Affiliation” (NPA) in Florida and “Unaffiliated” in North Carolina. First, we show a substantial disparity in the rate of NPA/Unaffiliated registration versus major party registration, depending on the state where a respondent resides. The closed primary rule in Florida is associated with a substantially higher share of major party registrants. Second, even taking account of state variation in registration rules, and controlling for a host of factors, strength of partisanship stands out as the principal variable conditioning the likelihood of registering unaffiliated. In this contemporary era of high partisan polarization, for many, the decision to register NPA/Unaffiliated reflects a social identity of political independence.

Not registering with a party in a polarized America

In a passage of The American Voter under the heading “Party Identification and Political Involvement,” Campbell et al. (1960, 143) offer the following commentary:

But if the usual image of the Independent voter is intended as more than a normative ideal, it fits poorly the characteristics of the Independents in our samples. Far from being more attentive, interested, and informed, Independents tend as a group to be somewhat less involved in politics. They have somewhat poorer knowledge of the issues, their image of the candidates is fainter, their interest in the campaign is less, their concern over the outcome is relatively slight, and their choice between competing candidates … seems much less to spring from discoverable evaluations of the elements of national politics.

Considering the political dispositions of registered voters six decades after Campbell et al. (1960), it strikes us that a polarized America has fundamentally changed partisans, but not their independent counterparts – located closer to the middle of the traditional seven-point PID scale. If anything, the intensification of partisan polarization should make independents even more inclined to exit the political arena. With PID frequently construed as a social identity (Green, Palmquist, and Schickler 2002; Greene 1999, 2000, 2002, 2004; Theodoridis 2017), among those who consider themselves to be independents, rising partisan acrimony should make this group the least vested in the trappings of politics and thus the least inclined to join a partisan team.

Of course, rules matter. For instance, studies by Burden and Greene (2000), Greene (2000), and Thornburg (2014) show that states with party registration foster greater partisan attachments.⁴ By comparison, not only do we consider variation in unaffiliated registration under different states’ rules, but we also ask to what extent does partisanship affect registering with or without a party? The latter dynamic is of particular interest because we contend that individuals who identify as independents act on this social identity by not registering with a party. Numerous studies confirm that voters conceptualize PID as a social identity (e.g., Green, Palmquist, and Schickler 2002; Greene 1999). Indeed, advances in this literature demonstrate that PID as a social identity can be explicit (simply asking one’s political affiliation along the ANES’s branched seven-point scale or employing a social-psychological battery of questions via Greene [2002]) or implicit, for example, based on implicit association tests (IATs) (Hawkins and Nosek 2012; Iyengar and Westwood 2015; Theodoridis 2017).

But whether PID is measured explicitly or implicitly, previous work has not assessed the influence of party affiliation on validated voter registration. Further, much of this scholarship on the social identity aspect of PID is fairly dated and hence was not undertaken at this current moment of historically high partisan polarization. Consider this statement by Greene (1999, 402), who writes: “An independent social identity may guide citizens’ self-perceptions, but it seems to have limited impact on political perceptions and behaviors.” Indeed, Greene (1999) found no significant relationship between an independent political identity and political participation. Nonetheless, five years later, Greene (2004, 139) writes that “the fact that in many states persons can register as ‘independent,’ just as they would as a Republican or Democrat, it may not be too implausible for some citizens to conceive of independents as just another political group to which they do or do not belong.” Not only do we agree with this pronouncement, but it also serves as the perfect jumping off point to assess empirically whether citizens closer to the independent “middle” of the seven-point PID scale are in fact significantly more prone to register unaffiliated.

Thus, controlling for a host of factors, we expect that the strength of partisanship is negatively related to registering NPA (Florida)/Unaffiliated (North Carolina). In other words, because we agree that political independence is a social identity (Greene 2004), it should manifest in a greater propensity to register unaffiliated. However, this relationship should be tempered by registration rules, leading to relatively less NPA registration in Florida versus North Carolina, as the former state holds closed major party primaries whereas the latter allows Unaffiliated registrants to participate.

Contemporary American politics revolve around the contentious interplay of the major parties. Harboring a weaker identity with either partisan tribe may undermine the motivation to engage in the political fray. Perhaps the most remarkable demonstration of independents’ distancing themselves from politics comes from the work of Layman and Carsey (2002). In their examination of conflict extension – that is, polarization across issue domains, among those who exhibit awareness of partisan differences on both social welfare and cultural issues – only pure independents do not polarize (Layman and Carsey 2002, 798–99). In other words, for even the most attuned independent voters, their refusal to polarize reflects an intentional decision. To somewhat twist and extend the metaphor of Schaffner, Streb, and Wright (2001), political independents do not play on partisan teams and therefore do not don uniforms; at best, they only half-heartedly root for Democrats/Republicans. Hence, it follows that individuals registering unaffiliated should be more detached from the major parties in terms of their partisan identification, even after controlling for numerous covariates.

Data and methods

We conducted two original web-based surveys of Florida and North Carolina, emailing registered voters in both states. For the Florida sample, we captured a random draw of 600,000 registrants from more than 15 million registered voters in the June 1, 2023, statewide Florida voter file, of which 597,054 emails were valid.⁵ The North Carolina sample of registered voters, randomly drawn from the more than 7 million registered voters in the state, was purchased from the vendor Aristotle, and included 610,271 registered voters of which we obtained 591,312 valid emails.⁶ The Florida survey was in the field July 18 through July 28, 2023, and the North Carolina survey was in the field from August 1 through August 12, 2023. Both surveys received IRB approval and responses were recorded using the Qualtrics platform.⁷

To test our propositions, we run logistic regressions with the dependent variable of interest a respondent’s official party registration status in the Florida or North Carolina voter registration database. Registration status is coded 1 for those respondents officially registered as No Party Affiliation (NPA) in Florida or Unaffiliated in North Carolina, and 0 for respondents registered as either Democrat or Republican in either state. We drop the handful of voters registered with minor parties, according to their official recorded status in the Florida and North Carolina voter files at the time the survey was conducted.

Our models include a host of control variables. Demographic indicators consist of age (in years); female; white (non-Hispanic); education; moved to Florida/North Carolina; and registration length (in years).⁸ Except for education and if a respondent moved to FL/NC, the other demographic variable data come directly from each state’s voter registration file. We also account for behavioral factors: talk politics (viewed as difficult); share views about politics (viewed as difficult); and surrounded by like-minded people. In addition, we asked about primary involvement, with variables that assess if one usually votes in primaries and if voting in primaries is important. Finally, we account for political variables: political engagement; political interest; ideology strength (moderate, slightly conservative/liberal, conservative/liberal, and extremely conservative/liberal); and PID strength (pure independent, independent leaner, weak partisan, strong partisan).⁹

Despite so many respondent characteristics taken into consideration, it is our expectation that strength of partisanship is a driver of unaffiliated registration. Conceiving of PID as a social identity, respondents nearer the independent end of the PID strength scale should exhibit a significantly greater probability of NPA/Unaffiliated registration in Florida and North Carolina, respectively. We run models separately for Florida and North Carolina and also show the results of a pooled model, which merely adds a binary variable coded 1 for Florida registrants. Because of Florida’s closed primary rule, compared with North Carolina respondents, we anticipate Florida registrants are significantly more likely to have registered with a major party.

Findings

Before turning to the multivariate findings for Florida and North Carolina, in Table 1, we begin by showing the distribution of partisanship among voters registered in Florida as “No Party Affiliation” (NPA) and voters in North Carolina registered as “Unaffiliated,” along with voters registered as Democrats and Republicans in each state. As anticipated, because of the restrictive primary participation rule in Florida, 76% of registrants in our sample are registered as Democrats or Republicans, according to Florida’s official voter file, with the remaining 24% registered NPAs. In contrast, in the semi-closed primary state of North Carolina, its more permissive participation rule is clearly reflected, as 47% of registrants in our sample are registered Unaffiliated, according to the official voter file, with the remaining 53% registered as Democrats or Republicans.

Table 1. Party identification by registration status, Florida and North Carolina

Note: Data computed by the authors from surveys (weighted) administered to registered voters in Florida and North Carolina.

Perhaps just as notable and interesting is the distribution of respondents’ partisanship among those registered NPA in Florida and Unaffiliated in North Carolina. Remarkably, of the three independent categories (lean Democrat, pure independent, and lean Republican), the cumulative total is 78% in Florida and 78% in North Carolina. However, once again reflecting the closed (Florida) versus semi-closed (North Carolina) primary rule, the modal category is pure independent in Florida at 34%; in contrast, pure independent is the smallest category (23%) of the three independent identifiers in North Carolina. These differences suggest the more permissive primary rule in North Carolina facilitates greater hidden partisanship among unaffiliated registrants, in keeping with the findings of Thornburg (2023). Stated another way, regarding the restrictive primary rule in Florida, the 34% plurality of pure independents registered NPAs indicates a substantial segment of the electorate appears to have no interest in participating in contests nominating partisan office-seekers. Conversely, the distribution of partisanship among partisan registrants (Democrats and Republicans) exhibits similar patterns in Florida and North Carolina, with all three independent categories (lean Democrat, pure independent, and lean Republican) in the single digits.

We now turn to Table 2, which displays the logistic regression models estimating the likelihood of registering unaffiliated (coded 1 = NPA in Florida/Unaffiliated in North Carolina, with major party registration coded 0), as well as the pooled model. Before turning to our variable of interest, PID strength, we will briefly mention the controls that are statistically significant. In Florida, women are less likely to register NPA; registrants on the voter rolls longer are less likely to be NPA; not surprisingly (because of closed party primaries), respondents who typically vote in primaries are less likely to be NPAs; and more ideological respondents are less likely to register NPA. In North Carolina, two demographic features correspond to a lower propensity to register Unaffiliated: age and education. Additionally, and as expected, in the pooled model the indicator for Florida respondents is negative and significant. Specifically, Florida registrants have a 0.24 NPA probability versus a 0.48 Unaffiliated registration likelihood for North Carolinians, a substantial 24-point difference.¹⁰

Table 2. Weighted logistic regression models for likelihood of registered NPAs in Florida and registered Unaffiliateds in North Carolina

*p < 0.05; **p < 0.01; ***p < 0.001.

In Figure 1, we display the probability of unaffiliated registration from the Florida and North Carolina models, based on our variable of interest, PID strength. Displaying the probabilities (bracketed by 95% confidence intervals) side-by-side for Florida (on left) and North Carolina (on right), we see the greater likelihood of unaffiliated registration across PID strength in the Tar Heel state with its more permissive semi-closed primary rule. The confidence intervals are also notably wider in North Carolina because of its much smaller sample. Among North Carolina registrants, a pure independent has a 0.91 probability of Unaffiliated registration, which plunges to a 0.23 probability among strong partisans. In contrast, a pure independent in Florida has a 0.63 probability of NPA registration, which plummets to a 0.07 probability in the case of strong partisans. In sum, the substantial and monotonic decline in unaffiliated registration going from pure independent to strong partisan, even after accounting for several other factors and state rules that greatly alter registration patterns, offers robust support for our contention that political independence is a social identity shaping voter registration decisions.

Figure 1. Predicted probabilities for registering NPA/Unaffiliated by strength of partisanship.

Note: Predicted probabilities generated from the estimates in the Florida and North Carolina models in Table 2.

Conclusion

In his study of intergroup behavior, Tajfel (1978, 63) defines a social identity as “that part of an individual’s self-concept which derives from his knowledge of his membership of a group (or groups) together with the value and emotional significance attached to the membership.” This generic concept should be applicable to a politically independent social identity, especially in our time of hyper-polarized major parties. In fact, regarding the realm of politics, Greene (1999, 395) constructed a measure of social identity for independents, and explained that “Given the strong civic virtue and social norms placed on political independence in the United States, it may be that either in addition to or in place of a social identification with a political party, some citizens may also identify with the category of political independents.”

By emphasizing explicit party identification along the strength of partisanship scale, we claim that most individuals do consider PID as a social identity. But, in considering gradations of partisan attachment from pure independent to strong partisan, we expected that individuals closer to the independent end of the scale would be significantly more likely to register as unaffiliated because those least attached to the major parties possess a politically independent social identity. Put differently, a weak association with a political party does not merely reflect the absence of a partisan identity but rather identity with political independence. As we have shown, a manifestation of this social identity of political independence is a notably higher propensity to register without a party, even after controlling for numerous respondent characteristics and assessing this relationship across states with primary rules that make it more or less likely to forego registering with a major party.

Scholars have repeatedly informed us that although a substantial chunk of the American electorate claims to be politically independent, a large segment of this group comprises closet/undercover partisans (e.g., Keith et al. 1992; Klar and Krupnikov 2016). Indeed, one experimental study shows that under the condition of implicit party identity, pure independents still profess a bias toward one of the major parties with respect to their policy preferences (Hawkins and Nosek 2012). We do not take issue with this scholarship but instead stress the other side of the equation: how might a greater identity with political independence influence a behavior? Thus, despite partisanship pervading the views of the contemporary American voter, at a time of historically high partisan polarization, a rise in unaffiliated registration suggests a strong linkage to a social identity of political independence.