Basking in Their Glory? Expressive Partisanship among People of Color Before and After the 2020 US Election

Abstract As the number of people of color (PoC) grows in the United States, a key question is how partisanship will develop among this important electoral group. Yet many open questions remain about PoC partisanship, due to limited availability of panel data, a lack of sensitive instrumentation, and small samples of PoC in most public opinion surveys. This brief report leverages a unique panel of African American (N = 650) and Latino (N = 650) eligible voters, before and after the 2020 Presidential Election between Democrat Joe Biden and Republican Donald Trump. Using measures that tap expressive partisan, racial, and national identity attachments, we find that Biden’s electoral victory significantly intensified partisan identity among his Democratic PoC supporters, relative to PoC who were not Democrats and supported Trump. We do not find significant changes in racial or national identities. Our results advance research on PoC’s partisanship.


Introduction
People of color (PoC) are a fast-growing demographic in the United States (US), but we lack a clear understanding of their partisan identification.PoC generally vote for Democrats (Fraga 2018), but there are reasons to question the depth of their Democratic identity (Hopkins et al. 2020).Substantial shares of PoC identify as unaffiliated with political parties (Hajnal and Lee 2011), especially if they have immigrant origins.Moreover, many Democratically-identified PoC believe their votes are under-valued because their policy preferences are weakly represented in Democratic governing agendas (Frymer 1999).
This article has earned badges for transparent research practices: Open Data and Open Materials.For details see the Data Availability Statement.
The intensity of PoC's Democratic partisanship is unclear for two reasons.First, most political surveys contain relatively small PoC samples.Additionally, partisan identity among PoC is generally measured with a 7-point scale that yields coarse gradations in partisanship strength (Huddy, Mason, and Aarøe 2015), which may not capture the nuances of PoC's partisan identification.
We address these limitations with a large panel survey of PoC before/after the 2020 Presidential Election between Democrat Joe Biden and Republican Donald Trump.We use unique instrumentation to capture expressive partisanship: a form of political identity that draws its intensity from a convergence of political (e.g., Democrats) and social (e.g., race/ethnicity) categories (Huddy, Mason, and Aarøe 2015).A strength of expressive partisanship is that it reflects a more granular partisan intensity than a traditional 7-point scale of partisan identity.Using a quasiexperimental design, we test whether PoC's expressive partisanship was affected by 2020's electoral result.
We find that PoC Democrats who supported Biden have intense levels of expressive partisanship, which became even stronger after the election.This identity shift was uniquely partisan, as we find no changes in PoC's racial or national identities.

Theoretical expectations
We aim to better understand PoC's Democratic partisanship.Elections are useful for studying this because their results provide information about group status.Social identity theory suggests that when an in-group gains power, members' identification intensifies (Tajfel 1981).This is especially likely when few out-groups exist and inter-group relations are institutionally enshrined (Brewer 1999), as in America's two-party system.
The 2020 election was especially high stakes for PoC Democrats.Donald Trump ran on the slogan "Make America Great Again," which promoted a vision that marginalized PoC.Joe Biden ran with a Black-Asian vice presidential candidate -Kamala Harrisand promised a country that was more inclusive of PoC.Given these stakes, we expect Democratic partisanship to intensify after Biden's victory.

Research design
We use panel data on 650 Black and 650 Latino eligible voters, surveyed 2-3 weeks before and after the 2020 election.This allows a difference-in-differences analysis comparing changes in expressive partisanship of Democratic Biden supporters to those of non-Democratic Trump supporters.We measure expressive partisan attachment with two items ("I identify as [Democrat/Republican/Independent/Other]," and "I am glad to be [Democrat/Republican/Independent/Other]") that we scale together.Full details on survey items are in Appendix 1.
We assume the expressive partisanship of Biden and Trump supporters would have followed common trends in absence of the election result, which lets us isolate the effect of having one's preferred candidate win the presidency.We cannot directly test this assumption with our data, but reassuringly detect no pre-election differences in partisanship levels.Appendix 1 fully details our research design, measures, estimation strategy, and assumptions.

Results
Column 1 in Table 1 compares expressive partisanship levels between Biden and Trump supporters before and after the election.Column 2 adds individual fixed effects to account for time-invariant characteristics.While pre-election partisanship levels are similar across Biden and Trump supporters, they decrease among Trump supporters after the election (this change is statistically insignificant).However, expressive partisanship significantly increased among Biden supporters by 6.5% points relative to Trump supporters.This shift is similar across Black and Latino respondents (see Appendix Table 5.2) 1 and robust to excluding Independent Trump supporters (see Appendix Table 5.3).
Pre-election levels of PoC's expressive partisanship among PoC Biden supporters were fairly high (µ 0:81), suggesting this attachment was already intense.But the magnitude of change in intensity indicates expressive partisanship among PoC is adaptive to political developments, such as Biden's victory.Columns 3-6 reveal no evidence of a shift in PoC Biden supporter's racial or national identity, suggesting the observed effect is uniquely partisan.

Implications
Prior research finds low levels of partisan identification among some PoC, but our focus on expressive partisanship reveals intense attachment to the Democratic party The dependent variable is a scaled [0,1] index of identity attachment based on items "I identify as : : : " and "I am glad to be : : : ".Biden supporter indicates preference for Biden over Trump and post-election after election survey.Individualclustered standard errors are in parentheses.Average Trump supporter ID is Trump supporters' pre-election mean partisan attachment.p0:001; p0:01; p0:05; p0:1. 1 We are underpowered to conduct sub-sample analyses by nativity.
Basking in Their Glory?

Table 1 .
Effects of a Preferred Candidate Victory on PoC Identity Attachments