Content uploaded by Dongshu Liu
Author content
All content in this area was uploaded by Dongshu Liu on Sep 28, 2021
Content may be subject to copyright.
Minority versus Minority: Partisanship and Inter-Group Competition
among Asian Americans
Dongshu Liu
City University of Hong Kong
Nathan Carrington
Syracuse University
Abstract:
Increasingly salient in democratic politics are the divides among political parties regarding how
they mobilize support between ethnic majorities and minorities. Why, then, do some members
of a minority group support political parties seemingly antithetical to the interests of minority
groups? We draw on group conflict theory to suggest that a partial explanation rests on
perceived competition within minority groups. We test this theory by focusing on Republican
Party support among Asian Americans in the United States. Based on two representative
surveys and an original survey experiment of Asian Americans, we demonstrate that perceived
competition among racial minority groups has a significant effect on the partisanship of Asian
Americans, pushing them toward the Republican Party. Our findings provide critical
implications on how race affects politics in democracies with increasingly diversified ethnic
minority groups.
Accepted at American Politics Research
Acknowledgement
We woud like to thank the Editor and two anynomous reviewers for their helpful comments.
Thanks to Shana Gadarian, Dimitar Gueorguiev, Emily Thorson, Steven White, Seth Jolly,
Chris Faricy, Li Shao, Colin French, Neil Malhotra, Beng Quan Zhang, Participants in 2018
MPSA Panel and participants in Political Science Research Workshop at Syracuse University
for their suggestions.
2
Introduction
Ms. Wen, an Asian American in Virginia, campaigned tirelessly for President Trump
during both the 2016 and 2020 elections. Noticeably, Ms. Wen, and many others like her,
represent a non-ignorable portion of Asian Americans firmly supportive of Donald Trump and
the Republican Party. The 2020 election saw an increase in Republican partisanship and a
decrease in Democratic partisanship among Asian Americans, despite Democrats’ consistent
endorsement of seemingly pro-minority policies and their nominating an Asian American
woman to the vice presidency. The increase in Republican support is particularly striking given
the GOP’s “China Virus” rhetoric coinciding with an increase in hate crimes against Asian
Americans and GOP immigration policies that make it challenging for Asian Americans
(particularly from India and China) to find a pathway to citizenship. Why, then, do a sizable
portion of Asian Americans support a party that seems antithetical to their interests?
To those following American politics, the Democratic Party is often regarded as the most
attractive for minorities, while the base of the Republican Party is usually believed to be
primarily white evangelical voters. This Majority-Minority paradigm dominates the discussion
and analysis of U.S. politics, with an underlying assumption that minority groups benefit from
the Democrats while the GOP benefits white Americans.
1
However, this majority-minority
paradigm is only part of the story. Much mobilizing of Asian Americans by the GOP in the
2020 election cycle targeted other minority groups instead of speaking in terms of the majority-
minority paradigm. For example, California Proposition 16, which sought to remove the ban
on affirmative action for education and employment, was cited by many Asian Americans in
explaining support for Trump, as they believe the proposition privileged Latino and African
Americans over Asian Americans.
2
We argue that this minority-minority paradigm is an
undertheorized and overlooked factor in determining the partisanship of minority groups.
Studies focusing on political relations among minority groups in the United States are
3
overwhelmingly dominated by the Black-white paradigm (McClain et al., 2006; McClain et al.,
2007). However, it is reasonable to expect that minority groups interacting with each other
would shape political partisanship. Past studies demonstrate that perceived competition or
threats from other minority groups can affect the social and economic positions of various
minority groups (Bobo & Hutchings, 1996; McClain et al., 2006; McClain et al., 2007). A
natural extension of this research is to ask how competition affects partisanship.
Focusing on Asian Americans, this paper argues that when members of a minority group
perceive low commonality and high competing interests with other minority groups, they are
more likely to identify with the Republican Party. Specifically, Asian Americans who believe
that they are in competition with African Americans or Latinos for privileges, rights, or benefits
are more likely to view policies and positions of the Democratic Party as less favorable, even
though the Democratic Party is generally viewed as being more favorable for minority groups
overall. Consequently, these individuals become less Democratic and more Republican. To test
the theory, we use two national Asian American surveys fielded in 2008 and 2016
(Ramakrishnan, Junn, Lee, & Wong, 2008; Ramakrishnan, Lee, Lee, & Wong, 2016), and
combine that with an original survey experiment on Amazon’s Mechanical Turk (MTurk).
Both observational and experimental data provide support for our theory. These findings
have implications for not only race scholars but also scholars of American politics more
broadly, including in political parties and voting behavior. Finally, this research highlights how
the majority-minority narrative in interpreting the partisanship of minority groups and support
for the Democratic Party is incomplete. A minority group may still identify with the GOP even
if it is socially excluded from the majority since other minorities are more threatening.
Asian Americans and the Formation of Political Partisanship
Political partisanship remains one of the strongest predictors of political attitudes and
beliefs (Campbell, Converse, Miller, & E., 1960; Kinder & Kalmoe, 2017), influencing how
4
people evaluate the political world (Campbell et al. 1960), how they process new information
(Lenz, 2012), and even their economic and political behavior (Gerber & Huber, 2009).
Consequently, it is unsurprising that scholars continue to discuss how these partisan loyalties
develop. Some, for example, argue that political partisanship is a product of socialization from
an early age (Campbell et al., 1960), with others arguing that partisanship forms as a result of
partisan stereotypes interrelated with self-identity (Green, Palmquist, & Schickler, 2002).
Under this latter view, people develop stereotypes of the groups that comprise each political
party’s coalition and self-selects into the party that most closely matches their self-identity.
Problems begin to arise, however, when multiple identities are stereotyped into multiple
parties. For example, although minority groups are often viewed as being included in the
Democratic Party’s coalition, many Asian Americans have interests that place them in
competition with other minority groups, such as their religious attainment (Wong, 2018),
higher median income, and educational attainment, often stereotyped with Republican partisan
identity. Asian Americans are also less likely to experience the same socialization as other
groups, particularly given their status as heavily first-generation. This realization has sparked
an interest in researching the partisanship of minority groups, especially Asian Americans.
According to Pew Research Center, the Asian American population has increased almost 81%
since 2000. In each presidential election cycle, the Asian American community is responsible
for half a million votes, enough to be determinative in some states.
3
Most of the literature on Asian American political behavior addresses two questions: 1)
why do Asian Americans participate in politics; and 2) why do Asian Americans develop
partisanship, whith more research focusing on the first question. Acculturation (Lien, 1994),
socialization (Cho, 1999), immigration status and length of time in the US (Xu, 2005), and
cognitive costs of voting (Bedolla & Michelson, 2009) have all been shown to influence the
degree to which Asian Americans participate in politics.
5
As for the second question, scholars offer several explanations for the development of
partisanship. Phan and Garcia (2009) argue that socioeconomic status, political attitudes, and
immigration experiences increase the likelihood of Asian Americans developing partisanship.
Focusing on Asian Americans and Latinos, Wong (2000) argues that the length in the United
States, naturalization status, English skills, and media usage are positive indicators of having
partisan identity. When investigating which partisan identity Asian Americans develop,
Raychaudhuri (2020) argues that a liberal local political context makes Asian Americans
identify Democratic. Media consumption and political efficacy are also positively correlated
with Democratic identity (Masuoka, Han, Leung, & Zheng, 2018). Approaching the question
from an intergroup perspective, Kuo, Malhotra, and Mo (2017) find that social exclusion (by
the majority white Americans) makes Asian Americans identify with the Democratic Party.
Zheng (2019) argues that partisanship of Asian Americans is determined by whom they feel
closer to between majority (white) and (other) minorities. Hajnal and Lee (2011) find out-group
commonality and experience of discrimination diminishes Republican identity among Latino
and Asian Americans. Other scholars attempt to explain Democratic partisanship among Asian
Americans with attention to traditional factors such as socioeconomic status and perspectives
on immigration (Wong, Ramakrishnan, Lee, & Junn, 2011).
To our knowledge, these are the only studies in the literature on why Asian Americans
would identify with a specific political party, and they focus exclusively on individual
characteristics of Asian Americans or their relation to mainstream society under the majority-
minority paradigm. This paper, however, expands our understanding of Asian American
partisanship by proposing a minority-minority perspective. We theorize that not all minority
groups share similar goals and this actual or perceived between-group competition, which has
previously been shown to influence prejudice of one group towards another (Esses, Jackson,
& Armstrong, 1998; Sherif, Harvey, While, Hood, & Sherif, 1961; Zárate, Garcia, Garza, &
6
Hitlan, 2004), is likely to influence political partisanship.
4
As such, we present a theory of
minority political partisanship that incorporates the role of inter-minority group competition.
5
Racial Competition, Similarity, and Political Opinions
Several theories exist underscoring the influence of inter-group relations on racial group
opinion formation. Such theories expect one racial group to be more hostile towards another if
they perceive the other group to be a threat. Overwhelming, however, these theories focus
primarily on white-Black relations or, more broadly, the majority-minority paradigm (Blumer,
1955; Giles & Evans, 1986; Oliver & Mendelberg, 2000).
Some work by McClain, however, focuses on Latino and Black Americans, highlighting
the importance of considering a minority-minority perspective (McClain et al., 2006; McClain
& Karnig, 1990). Specifically, this work argues that Latino American immigrants “hold
negative stereotypical views of Blacks and feel they have more in common with whites than
with Blacks” (McClain et al., 2006). Similarly, Kim (2000) documents the competition
between Korean and African Americans in New York and demonstrates how such competition
shapes politics. These findings indicate that minority groups face competition with both white
Americans (the majority) and other minority groups, though it is not immediately clear the
extent to which, if any, this latter competition affects political partisanship and voting behavior.
Although Asian Americans may feel competition with other minority groups for a variety
of reasons, including electoral interests (Lai, Cho, Kim, & Takeda, 2001) or rank in the
American racial order (Kim, 2000), the research investigating the political consequences of
Asian Americans’ perceptions of other racial groups is scarce. Ramakrishnan, Wong, Lee, and
Junn (2009) find that commonality with Black Americans motivates support for Obama among
Asian Americans in the 2008 primaries against Hillary Clinton. Similarly, Samson (2015) finds
that commonality with Black Americans results in increased support for a path to citizenship
for undocumented immigrants.
7
Although much of the literature on intra-group affect highlights the importance of external
factors (such as a population increase among a “rival” group or economic competition), these
feelings can also originate psychologically. As Bobo and Hutchings (1996) argue, “orthodox
prejudice, such as negative feelings and stereotyping” matters for threat perceptions.
Psychological motivations are not just negative. Indeed, group perceptions have previously
been identified as an important source for partisanship. In their foundational study, for example,
Hajnal and Lee (2011) show that the racial identity of minorities groups, measured by out-
group commonality and discrimination experience, affects the partisanship of Asian Americans
and Latinos. Our theory builds on this line of argument by arguing that a significant difference
exists between Majority-Minority relations and (our focus of) minority-minority relations.
In Hajnal and Lee (2011), for example, out-group commonality is measured using an
additive index of commonality with both White Americans and other minority Americans and
their discrimination questions do not specify whether the discrimination is from a majority or
minority group member. This suggests that Hajnal and Lee (2011) conceive of perceptions and
relations with out-groups as unidimensional and that commonality and competition with both
minority and majority groups will contribute to partisanship in one direction.
Our argument, however, is that minority-minority relations affect partisanship differently
than majority-minority relations. Below, we show that relations with White Americans work
in the opposite direction than relations with other minorities in shaping Asian American
partisanship. We develop our theory of minority-minority competition in the next section. It
should be clear, though, that our theory expands on the group-identity perspectives initially
made by Hajnal and Lee (2011), Kuo et al. (2017), and Zheng (2019), among others, by
pointing out the importance of the competing interests from a minority-minority perspective.
A Theory of Minority Group Competition and Partisanship
We draw from group conflict theories in psychology (Esses et al., 1998; Sherif et al., 1961;
8
Zárate et al., 2004) to argue that feelings of competition with other minority groups shape the
partisanship of racial minority groups by influencing the perception of benefits they can gain
from the two major political parties. Like others, minority groups likely identify with the party
that can provide the most benefits, whether material or psychological. In general, minority
groups will tend to identify with the party with pro-minority policies.
We argue, however, that perceptions of the policies proposed are shaped by perceived
shared interests with other minority groups. If one minority group feels a strong commonality
in political rights and interests with other minority groups, they form the same political alliance
and are unlikely to react negatively if the other groups in the coalition gain from pro-minority
policies as that gain benefits everyone in the coalition. On the other hand, if a minority group
feels in competition with other minority groups, they are likely to view the pro-minority
policies as meaningless because, even if such policies can bring about benefits, few (if any) of
these benefits flow to them. Because no political common ground for any at-large alliance
exists, this minority group is likely to feel negatively toward the pro-minority policies because
they benefit their competitors (Esses et al., 1998; Zárate et al., 2004). This is especially
pronounced for the weaker minority groups in terms of size and political power since they are
more likely to lose and to feel the pro-minority policies benefit others rather than them.
A Competing Interest Theory of Asian American Partisanship
We expect Asian Americans to identify with the Democratic Party if they perceive similar
interests with other minority groups. Policies aimed at helping minorities are heavily shaped
by the concerns raised by African Americans and Latino Americans, both much larger than
Asian Americans in terms of population and political influence.
6
Only when Asian Americans
feel they are not in competition with other minority groups but, instead, share common political
interests would they believe they also benefit from these policies.
The commonality and perceived competition of interests might lead to a zero-sum view
9
of politics. According to Zárate et al. (2004), competition among groups can create a feeling
of prejudice against other minority groups. As a result, some Asian Americans might feel less
in common with other minority groups regarding political representation and government
policies. Consequently, we predict such Asian Americans to move away from the Democratic
Party and towards the Republican Party (either becoming independent or identifying with the
GOP) due to a feeling that policies promoted by Democrats only help Latino and African
Americans. Republicans, although not widely known as being pro-minority, promote policies
likely to benefit Asian Americans, such as tax breaks (Grossmann & Hopkins, 2016). For some
Asian Americans, the Republican Party does not seek to benefit competitors. Because policies
of the Democratic Party can be perceived as benefiting competitors whose interests are not in
common with Asian Americans’, some may feel the Republican Party deserves support.
As noted, this minority-minority view is one of the techniques used to mobilize
Republican Asian Americans. For example, some Asian Americans believe that Democrats
support African and Mexican Americans by raising taxes on Asian Americans.
7
A rumor that
California plans to offer reparations of two million dollars to African Americans for slavery by
taxation was widely spread among the Chinese American community as evidence of how
California treats Asian Americans unfairly.
8
All of this indicates a clear pattern of how
perceived competing interest with other minorities can push some Asian Americans towards
the Republican Party.
Beyond Hajnal and Lee (2011), Zheng (2019) also finds that commonality with other
minorities enhances the Democratic identity while commonality with white Americans leads
to the Republican identity. However, Zheng (2019) still adopts a majority-minority paradigm,
as he argues that racial consciousness is used as a cognitive shortcut to decide whether the
interests of Asian Americans lie with the majority White or other minority groups. In this
paradigm, Asian Americans are assumed to pick a side between the majority (white) and
10
minority (Latino and African) groups. Our theory, however, derives from an intra-minority
perspective. The strong competition and low commonality with other minority groups are
pushing some Asian Americans toward the GOP, but it does not necessarily mean that they
feel closer to white Americans. Asian Americans form their partisanship by evaluating how
they benefit from pro-minority policies based on how competitive politically they are with
other minorities, instead of merely which side they are closest to between the majority and
minority groups. In other words, we argue that the majority-minority relation and the minority-
minority relation are two separate processes determining partisanship. Both are influential, and
our analysis indicates that these two processes co-exist (although opposite in effects), rather
than being a “one-out-of-two” choice in deciding partisanship.
Personal Level Hypothesis
We further theorize that the this zero-sum perception of minority-minority relations at the
group-level may need to reach the personal level in order to shift one’s partisanship identity.
Junn and Masuoka (2008) find that the “proportion [of Asian Americans] overall who say race
is important to them in their racial consciousness is much smaller overall than for Blacks.” If
being Asian American is not important to one’s racial consciousness, it is likely that one may
perceive the competition between Asian Americans and other minorities as merely an abstract
concept without tangible consequences. Therefore, we theorize that Asian Americans only
adjust their partisanship when they feel the competition has personal relevance allowing them
to transform an abstract idea into a tangible consideration with concrete emotional reactions.
We test our theory using both observational and experimental data on Asian Americans.
Specifically, two rounds of the National Asian American Survey are combined with an original
survey experiment, all of which support our theory.
Study 1: Observational Data
The theory outlined above offers the following hypotheses:
11
Hypothesis 1A: Asian Americans who perceive more in common politically with African
Americans will identify less with the Republican Party.
Hypothesis 1B: Asian Americans who perceive more in common politically with Latino
Americans will identify less with the Republican Party.
Data and Methods
To test these hypotheses, we analyze both the 2008 and 2016 National Asian American
Surveys (NAAS). In 2008, the NAAS sampled 5,159 Asian Americans. Table A1 in the
appendix describes the ethnic breakdown of the sample. For the primary dependent variable of
interest (political partisanship), we rescale the 7-item scale to range from zero to one, with 0
being a strong Democrat and 1 being a strong Republican.
9
One major concern with this variable, however, is the “don’t know” responses that plague
studies of Asian American partisanship. To alleviate the concern with such a large proportion
of “don’t know” responses, we follow previous studies (Kuo, Malhotra, and Mo 2016,
Ramakrishnan et al. 2009) and treat this in two different ways. First, these responses are treated
as missing data and excluded from the analysis entirely, leaving 2,993 valid respondents.
Second, we treat them as the midpoint of partisanship (0.5).
10
Political commonality is generated by asking respondents how much they perceive to have
in common between Asian Americans and African and Latin Americans “thinking about
government services, political power and representation” (See appendix for full question
wording). Because our theory emphasizes a zero-sum, competitive relations with other
minority groups regarding benefits gained from government policies and political decisions,
political commonality is a appropriate measurement. If an Asian American perceives high
political commonality with other minorities, he or she is less likely to perceive their relation as
zero-sum. The question is re-scaled from a four-scale index to an index with a range from zero
(nothing in common) to one (a lot in common). Missing data were treated like the partisanship
12
variable above. We include commonality with white Americans in all analyses to control for
the effect of the majority-minority perspective and compare whether the minority-minority
perspective we propose can have an independent and differential effects on partisanship from
the minority-minority one.
For the other control variables, we included gender, income, age, education, time spent in
the United States (as a percentage of total life years.), religiosity (whether they have religious
beliefs), citizenship, and ideology (liberal-conservative). In addition, we include a binary
variable for having experienced racial discrimination, as indicated in Kuo et al. (2017). We
rescale all the non-binary control variables to range from zero to one. Finally, missing data
among the control variables are coded as a zero, with a binary variable generated as missing
data indicators. These binary variables are included in all analyses.
11
We adopt the same coding strategies and variables as much as possible for the 2016 NAAS
data. In total, 4,438 respondents are included under the “midpoint” approach. The “missing
data” approach leaves 2,706 valid respondents. Political interest is not included because the
2016 NAAS omits the question. Ideology questions are also omitted, but the 2016 survey offers
two questions gauging support for abortion rights and a transgender bathroom policy which we
model as proxies for ideology. See Table A1 in the appendix for details of 2008 and 2016 data.
Results
We employed OLS regression with Models 1 and 2 using the 2008 survey data and Model
3 and Model 4 using the 2016 data. Missing data are excluded from Models 1 and 3, with
Models 2 and 4 using the midpoint approach. Moreover, following the practice in Kuo et al.
(2017), all analyses are one-tailed where appropriate because our theory is directional. See the
appendix for summary statistics.
The results presented in Table 1 largely support the hypotheses. An increase in the
perception of commonality with African Americans decreases the Republican Partisanship by
13
5 to 7 percent, after controlling for commonality to white Americans and other variables. The
effect is similar on Latino Americans. This relationship holds in both the 2008 and 2016 data,
irrespective of how the missing data are treated. While most of the results are statistically
significant, commonality to African Americans in Model 2 is very marginally insignificant
(p=0.051) but in the correct direction. In addition, while it appears that the relationship is
stronger for the commonality with Latino Americans, the difference is insignificant. It also
seems that the effect of common interests is larger in 2016 compared to 2008. Multiple
robustness checks were conducted, including the logit, ordinal logit, and multi-logit regressions,
showing that our findings are robust regardless of model specifications and assumptions.
Similarly, a non-exact, near-neighbor matching was conducted to show that our findings are
robust without a linear and additive parametric mode. The results can be found in the appendix.
An alternate explanation for our finding is that people use outgroup commonality to
decide which party is closer to themselves (Zheng (2019). If this is true, we would expect
commonality with white Americans to be negatively associated with commonality to other
minority groups. Only in this way can commonality serve as a shortcut for Asian Americans to
evaluate which side is closer. Thus, we regress commonality to white Americans on the other
two minority-group commonalities and other control variables, with results shown in Table 2.
As demonstrated, the three commonality perceptions are significantly and positively correlated,
confirming our argument that low commonality with other minorities does not push Asian
Americans towards white Americans. Asian Americans who identify with the GOP do so not
because they feel closer to white Americans, but partly because they feel competition for
political interests with other minorities so they cannot gain from supporting Democrats.
Furthermore, our analysis in both Tables 1 and 2 indicate that the minority-minority
perspective is independent of the majority-minority paradigm in shaping partisanship.
Although commonality with white Americans positively correlates with the other two minority
14
commonalities, the three commonality measures push partisanship in different directions.
Commonality with whites enhances GOP identity among Asian Americans, while
commonality with other minorities enhances Democratic identity. Therefore, Asian Americans’
partisanship is affected by both the majority-minority and the minority-minority perspectives
simultaneously, and scholars should pay attention to both when studying partisanship.
Lastly, we sought to test our personal level hypothesis in the observational study, though
neither wave of the survey included ideal questions. Because of this constraint, we utilize the
level of linked fate with other Asian Americans as an exploratory test of our personal level
hypothesis. In the appendix, we explain our rationale for taking this approach and its potential
limitations. The results, shown in the appendix (Table A17-1 and 17-2, Figures A3 and A4)
support the personal-level hypothesis. We fully acknowledge that, because the linked fate
question is not a perfect measurement, this evidence is only exploratory. We urge readers to
consider the evidence in the observational data together with the experimental evidence below.
Study 2: MTurk Survey Experiment
While the observational data offer strong support for our theory, they have shortcomings.
First, the causal direction remains ambiguous. It might be the case that Asian Americans
perceive a threat from other minority groups because they are Republican (Lenz, 2012).
Republican elites have continuously made known their views towards Latino and African
Americans. As a result, Asian Americans identifying with the Republican Party (because of
early familial socialization, for example) could receive these signals and update their attitudes
accordingly. Second, the questions used to gauge commonality are less precise than we would
hope in terms of measurement. Third, there might still be some unobserved confounders. Lastly,
although the moderating effect of the linked fate seems supportive, we still need more evidence
to test our personal level argument and see whether competition with other minorities shifts
the partisanship only when it reaches personal level.
15
To address these concerns, we fielded a survey experiment on MTurk reaching a total of
161 Asian Americans.
12
MTurk has become widely accepted for theory testing and
experiments (Berinsky, Huber, & Lenz, 2012). Clear, explicit language was used in the
experiment recruitment indicating that only Asian Americans are eligible, and participation
was limited to those who had previously completed 5,000 Human Intelligence Tasks (HITs)
with a 97% or greater approval rating.
13
Only those MTurk users with USA-based IP addresses
are included.
14
Requiring such a high level of approved HITs, combined with questions asking
respondents about ancestral origins at the very beginning of the survey, increases our
confidence that respondents are not fabricating data. Indeed, analyzing the data for
inconsistencies or abnormalities (using things like IP addresses or coordinates) yields no
suspect findings. Finally, there is no reason to expect that respondent, especially those with our
inclusion criteria, are more inclined to lie about their ethnicity than any other questions.
The treatment we used in our study concerned the claimed unfair treatment of Asian
Americans in university admissions caused by affirmative action policies. This treatment was
selected for several reasons. First, it is a real-world situation where Asian Americans are told
that they compete with other minorities, allowing our treatments to replicate reality and
enhancing external validity. The issue is also framed this way by politicians to mobilize Asian
Americans.
15
Second, it is perceived by many Asian Americans as generating competition
with other groups. As some view it, in helping African or Latino Americans, elite universities
are hurting Asian Americans. Taken together, the salience and weaponization of the issue by
political elites seeking to attract Asian Americans makes it an ideal treatment.
16
Once in the experiment, respondents are randomly assigned to one of three groups. The
first group exposes respondents to a short excerpt summarizing a complaint filed by several
Asian American organizations to the Department of Justice arguing that, by helping other
minority groups, affirmative action unfairly discriminates against Asian Americans. The
16
second group combines the manipulation in group 1 with a prompt asking respondents to
describe “when they, a close friend, or their children” were the victims of discrimination in the
university admissions process. This “bottom-up” treatment method is widely used in the field
of political psychology as a way to manipulate individual affect (Albertson & Gadarian, 2016).
The purpose of this prompt is to personalize the competition, consistent with our theory that
competition must be localized. Finally, the third group is a control group that reads no material.
The detailed treatment conditions and wording are in the appendix. After exposure to treatment
(if applicable), participants are immediately asked about their partisanship using the method
utilized by both the American National Election Studies and the NAAS. Afterward,
respondents are asked how each political party represents the interest of people like them, as
well as their overall attitude towards each party in a feeling thermometer.
We also include a question to determine treatment acceptance asking respondents whether
they agree that SAT scores should have a greater weight in university admissions. Because
Asian Americans have higher SAT scores on average than African and Latino Americans,
weighting SAT scores greater in admission is supported by many Asian Americans to gain
advantages in the competition of college admission with other minority groups. Therefore,
those perceived stronger competition with other minorities (i.e., people in the treatment group)
should support giving more weight to SAT to gain more advantages over other minorities. For
details of the experiment design, see the appendix.
Experimental Results
As shown in the descriptive statistics and the balance check (in Appendix), our
randomization is largely successful. Additionally, experiment groups are significantly different
in our question measuring treatment acceptance (shown in the appendix), as only the
respondents who are exposed to the competing interests and the personal prompt question
demonstrate significantly supportive towards increasing the weights of SAT in admission (F-
17
test=4.248; p-value=0.016). This indicates that our treatments work as expected.
Overall, the experimental data confirms the causal direction and lends further support to
our theory. As is apparent in Figure 1, those randomly assigned (N=50) to treatment group 2
with competing interests information combined with the personal prompt are about 15%
stronger in GOP identity. However, those that were exposed to only the competing interest
information without a personal prompt (Group 1; N=53) were not stronger in Republican
identity. This finding suggests that a strong feeling of losses in the competition with other
minorities triggers Republican Partisanship, but the grievance must be personal and not simply
a group-wide perception. This is consistent with our theory expressed above.
17
One may question whether Asian Americans support the GOP because they feel the GOP
supports the lawsuit. We have two reasons to believe this is not true. First, our treatment did
not connect the lawsuits to any specific political party. Second, respondents in the “competition
only” group also read the same lawsuit information but did not change their partisanship, which
implies that the respondents changed their attitudes because they perceived the competition
with other minorities at personal level rather than the GOP doing something they like.
18
Some may further worry that our findings are driven by the personal prompt question
rather than the competing-interests message of university admission. While this is a valid
concern, we have evidence to suggest that it is unwarranted. By analyzing the answers to the
personal question, it becomes clear that most of the unfairness recalled is about being Asian
American and having disadvantages compared to African and Latino Americans. Thus, our
experiment is effective in reminding respondents about inra-minority competition at a personal
level. The only way our theory cannot hold, therefore, is if the effects of the personal prompt
question is associated with just having a question to answer but not with what they recall.
We examined whether this is true by separating the respondents in the combined group
(Group 2) further into two subgroups: one subgroup with all respondents who report no
18
unfairness and the other subgroup who reported unfairness in the admission compared with
other minorities. We then compared the partisan identity of these two subgroups together with
the control groups and competing-interests group. The result, shown in Figure 2, clearly shows
that only respondents who recalled unfair experiences after reading the personal question
become more Republican (about 20% stronger). Reading and answering the personal prompt
question but not recalling any unfair experiences did not make our respondents different from
the control group. Thus, we have some confidence in saying that it is the unfair memories
triggered by the personal questions that make respondents significantly more Republican.
One might argue that the reason that we do not find significant results on the competition-
only treatment is because the treatment is weak. While this is a valid concern, there are a few
reasons we do not believe this to be the case. First, as noted above, the treatments were designed
to mimic signals that the respondents are likely to receive in the real-world, rather than
overemphasizing the scale of group conflict. Secondly, and more importantly, the condition
which combines the conflict treatment with a personal prompt also serves as a stronger conflict
treatment by forcing respondents to cognitively recall how favorable policies to other
minorities affects them or their loved one. We can compare the prompt treatment to a “bottom-
up” style manipulation commonly used in research on emotions in politics (Albertson &
Gadarian, 2016). The fact that we do not see an increase in Republican partisanship among
those who received the personal prompt but wrote no experience of unfairness indicates that
even a stronger treatment cannot produce effects if it is not personally relevant.
Although the experimental results are largely robust, a few results regarding attitudes to
the Democrats do not pass the conventional significance threshold despite the correct direction.
We have three possible reasons. First, the experiment was fielded shortly after widespread
media coverage of the lawsuit, and MTurk samples are more educated than the general public
(Berinsky et al., 2012). Therefore, our respondents should have a high level of prior exposure
19
of treatment, which reduces the treatment level (Gaines & Kuklinski, 2011). Second, we have
a small sample due to the difficulty of recruiting Asian American participants given their small
portion in population, which inevitably hurts statistical power. Lastly, the participants recruited
from MTurk are more likely to be liberal and the Democratic, so any treatment that pushes
them towards a Republican identity would be more difficult. Given these reasons, we think our
experiment is a conservative test. Consequently, the overall effects would be stronger with less
media coverage of the Harvard lawsuit and a larger and more representative sample of the
general population. The fact that our results are largely robust in a conservative test makes us
confident that our theory extends beyond the experimental context. Even so, future
experimental research should seek to achieve a greater sample size.
Extended Analysis
To further verify our results and test alternate explanations, we conducted several
extended analyses with the details and results of each in the Appendix. First, we address the
ongoing debate in the literature about the presence of a “common identity” among Asian
Americans to see how the internal diversities of Asian Americans may affect our finding. We
also acknowledge the high diversity of social-economic status (SES) of different Asian
American groups and investigate whether the effect varies across SES. We find that our theory
holds across all subgroups, although wealthier groups might be more sensitive to our theory
(See section 4 in Appendix).
19
Next, a mediation analysis shows that only 9.7% of our
treatment effects are mediated insignificantly through ideology, which indicates that ideology
is not driving our experimental results.
20
Finally, we test our theory on Latino Americans
finding similar results (Table A16 and Section 4 in Appendix). This latter point suggests that
our theory is generalizable beyond Asian Americans, but further research is needed.
Discussion and Conclusion
In this study, we argue and demonstrate with both observational and experimental data
20
that the relationship between one minority group and another is crucial in shaping the political
partisanship of an important and growing minority group. The majority-minority paradigm is
not the only way of studying the partisanship of racial minorities. Instead, intra-group relations
among minority groups also affect partisanship. The partisanship of a minority group is not
only shaped by its relations with the majority group, but also by its relations with other minority
groups. If one minority group perceives having less competition with other minority groups,
they are more likely to support the Democratic Party. If, on the other hand, a minority group
perceives they compete with other minority groups, they are more likely to identify with the
Republican Party because they do not feel they benefit from the pro-minority policies of the
Democratic Party. The relationship with other minority groups influences the perceptions of
the two parties, as well as their policies, which subsequently affects party affiliations. We
further show that the competition perception is effective only when it reaches the personal level.
The partisan identity of racial minority groups has long been of interest to scholars given
its relation to political behaviors and attitudes, as well as practitioners who seek to maximize
the effectiveness of campaign strategies. As a result, the implications of this research are far-
reaching. This study reveals the importance of intra-group competition between minority
groups on partisanship and shows that it works differently from competitions with the Majority.
Furthermore, minority groups should not be viewed as unified when assessing the effects of
pro-minority policies simply because of the majority-minority paradigm—the independence
and autonomy of the various minority groups and their different interest calculations must be
recognized. This study provides a new and important influence in understanding the
partisanship of Asian Americans, and arguably other minority groups, as well.
Our finding are consistent with some studies arguing for self-interests-centered incentives
in partisanship (Downs, 1957), but different from studies arguing the socialization-based
partisanship (Campbell et al., 1960; Green et al., 2002). We suspect that the reason is that Asian
21
Americans (and perhaps also Latinos) have a high percentage of first-generation of immigrants,
so they are more self-interest oriented and less socialized in U.S. context. We encourage further
studies on this question.
These results also have implications for political practitioners who seek to mobilize and
influence political partisanship during campaigns. Our study suggests that an effective method
for political parties to influence the political partisanship of minority groups is to highlight the
cleavages between the minority groups. For example, a mailer sent to Asian Americans
emphasizing that affirmative action policies (favored by the Democratic Party) benefit African
and Latino Americans at the expense of Asian Americans is likely to have a significant effect
on the proportion of Asian Americans who vote Republican. We have no reason to suspect that
a similar result would not occur among other minorities and urge further research in this area.
Our study demonstrates the importance of a detailed analysis of racial minorities.
Contemporary American politics has long viewed minority groups as a monolith, regardless of
whether it includes African, Latino, or other ethnic groups. For example, when analyzing the
shocking victory of Donald Trump in 2016, the media and pundits frequently used terms like
“white identity” and “minority voters,” indicating a clear “white-versus-minorities” narrative.
21
However, the findings of this study demonstrates that each racial minority group should be
viewed separately and independently. Therefore, it is crucial to understand the political attitude
of each racial minority group independently and examine their relationship with other minority
groups, particularly when ethnic minority groups become increasingly diversified. This will
offer a better understanding of partisanship for racial minority groups to scholars, practitioners,
and everyone with interests in how race affects the U.S. electoral politics.
22
Endnote
1
For example, see: https://www.washingtonpost.com/politics/2020/02/07/republican-party-is-white-southern-
how-did-that-happen/
2
https://www.latimes.com/california/story/2020-11-01/affirmative-action-divides-asian-americans-ucs-largest-
overrepresented-student-group
3
See: https://goo.gl/kvCMnr；
4
Within the traditional group conflict theory, actual or perceived group competition results in an increased
prejudice against the out-group. Thus, it is often prejudice that is the mechanism by which perceived competition
manifests in certain outcomes.
5
Group competition is theoretically distinct from social exclusion and has different remedies and expectations.
As Kuo et al. (2017) demonstrate, social exclusion reinforces Democratic partisanship with the goal of increasing
overall societal inclusion whereas group competition results in an increase in Republican partisanship given the
Republican Party’s discounting of group identity vis-à-vis the Democratic Party (Grossmann & Hopkins, 2016).
6
Although proponents of such policies argue that they benefit all minority groups (including Asian Americans),
for some Asian Americans, this is not true. For example, see: Kim and Lee (2001). Here, we are focused on
subjective (or perceived) competition and, as a result, we do not address whether these policies actually
(objectively) benefit all minority groups.
7
https://cn.nytimes.com/usa/20200916/cold-war-heats-up-in-chinese-families/
8
See an example: https://www.tiaotiaowen.com/articles/9zcm1z. Some Chinese Americans use this rumor to
claim that the Democrats favor African American by taxing Asian American.
9
When this index increases, people are moving away from the Democrats and towards the Republican. They can
become independent from being Democrats or being Republican from being Democrats. In either way, they
become more inclined to the GOP.
10
Future studies may also consider using imputation to address missing data problem.
11
We follow Kuo et al. (2017) in processing and rescaling data.
12
MTurk has been used to study racial subpopulations (Jefferson, 2019). Our sample is not large as recruiting
subpopulations is hard, but it is comparable to some other experiments of Asian Americans, e.g., Kuo et al. (2017).
13
In this regard, we are like those used in lab experiments recruiting specific demographics.
14
In appendix, we further discuss how it may affect our findings if someone outside USA use technical method
to bypass the IP address requirement. In short, we believe it does not hurt our finding.
15
See an example of how Harvard lawsuit is framed as a competiton between Asian Americans and other
minorities: https://politi.co/35M0bHL (Access in Feburary 2, 2021)
16
It is noted that our treatment reflects more about the zero-sum competition, rather than commonality used in
the observational study. However, both competition and commonality share the same core of our theory: the zero-
sum relation in the distribution of interests from a minority-minority perspective. Therefore, our two studies are
still intellecturally connected. We actually use two elements to test our theory: the perception of commonality and
the perception of competition. We believe these two share the same core of our theory and provide a
comprehensive test of our theory. Interested readers who are considering using these two concepts in their studies
may also refer to the appendix, in which we provide a more detailed discussions on these two concepts and how
they maybe constructed and used in our study and other existing studies.
17
Figure A1 in appendix shows that the treatments have similar positive effects on the GOP and the negative
effects on the Democratic party regarding favorability towards the Party and the perceived representativeness of
the Party. These results provide some evidence to support the mechanism, as it shows that our treatments do
increase respondents evaluation on the GOP and make them believe that the GOP is representing their interests.
18
To address the potential heterogeneous effect of control and demographic variables, we ran an OLS regression
controlling all demographic variables. The result is higher similar (Figure A2 in Appendix).
19
According to latest data, Indian American, Fillipino American, Chinese American, Japanese American and
Korea American are among the highest income subgroups in Asian American populations.
20
Indirect effect through ideology=0.0159, total effect=0.1641, p=0.268
21
See: https://goo.gl/H5ia4C (NY Times) and https://goo.gl/9zJJp3 (CNN)
23
Reference
Albertson, B., & Gadarian, S. K. (2016). Did that scare you? Tips on creating emotion in
experimental subjects. Political analysis, 24(4), 485-491.
Bedolla, L. G., & Michelson, M. R. (2009). What do voters need to know? Testing the role of
cognitive information in Asian American voter mobilization. American Politics
Research, 37(2), 254-274.
Berinsky, A., Huber, G., & Lenz, G. (2012). Evaluating online labor markets for experimental
research: Amazon. com's Mechanical Turk. Political analysis, 20(3), 351-368.
Blumer, H. (1955). Reflections on theory of race relations. Race relations in world perspective,
3-21.
Bobo, L., & Hutchings, V. L. (1996). Perceptions of racial group competition: Extending
Blumer's theory of group position to a multiracial social context. American Sociological
Review, 951-972.
Campbell, A., Converse, P. E., Miller, W. E., & E., S. D. (1960). The American voter. Chicago:
University of Chicago Press.
Cho, W. (1999). Naturalization, socialization, participation: Immigrants and (non-) voting. The
Journal of Politics, 61(04), 1140-1155.
Downs, A. (1957). An economic theory of democracy. Journal of public economics, 65, (2),
135-150.
Esses, V. M., Jackson, L. M., & Armstrong, T. L. (1998). Intergroup competition and attitudes
toward immigrants and immigration: An instrumental model of group conflict. Journal
of social issues, 54(4), 699-724.
Gaines, B. J., & Kuklinski, J. H. (2011). Treatment Effects. In J. N. Druckman, D. P. Green, J.
H. Kuklinski, & A. Lupia (Eds.), Cambridge handbook of experimental political
science (pp. 445-458).
Gerber, A. S., & Huber, G. A. (2009). Partisanship and economic behavior: Do partisan
differences in economic forecasts predict real economic behavior? American Political
Science Review, 103(3), 407-426.
Giles, M. W., & Evans, A. (1986). The power approach to intergroup hostility. Journal of
Conflict Resolution, 30(3), 469-486.
Green, D. P., Palmquist, B., & Schickler, E. (2002). Partisan hearts and minds: political
parties and the social identities of voters. New Haven, Conn: Yale University Press.
Grossmann, M., & Hopkins, D. A. (2016). Asymmetric politics: Ideological Republicans and
group interest Democrats: Oxford University Press.
Hajnal, Z. L., & Lee, T. (2011). Why Americans don't join the party: Race, immigration, and
the failure (of political parties) to engage the electorate: Princeton University Press.
Jefferson, H. (2019). Policing Norms: Punishment and the politics of respectability among
Black Americans.
Junn, J., & Masuoka, N. (2008). Asian American identity: Shared racial status and political
context. Perspectives on Politics, 6(4), 729-740.
Kim, C. J. (2000). Bitter fruit: The politics of black-Korean conflict in New York City: Yale
University Press.
Kim, C. J., & Lee, T. (2001). Interracial politics: Asian Americans and other communities of
color. Political Science & Politics, 34(03), 631-637.
Kinder, D. R., & Kalmoe, N. P. (2017). Neither liberal nor conservative: Ideological innocence
in the American public: University of Chicago Press.
Kuo, A., Malhotra, N., & Mo, C. H. (2017). Social Exclusion and Political Identity: The Case
of Asian American Partisanship. The Journal of Politics, 79(1), 17-32.
Lai, J. S., Cho, W. K. T., Kim, T. P., & Takeda, O. (2001). Asian Pacific-American campaigns,
elections, and elected officials. Political Science & Politics, 34(03), 611-617.
24
Lenz, G. S. (2012). Follow the leader?: how voters respond to politicians' policies and
performance. London;Chicago;: University of Chicago Press.
Lien, P.-t. (1994). Ethnicity and political participation: A comparison between Asian and
Mexican Americans. Political Behavior, 16, 237-264.
Masuoka, N., Han, H., Leung, V., & Zheng, B. Q. (2018). Understanding the Asian American
vote in the 2016 election. Journal of Race, Ethnicity and Politics, 3(1), 189-215.
McClain, P. D., Carter, N. M., DeFrancesco Soto, V. M., Lyle, M. L., Grynaviski, J. D.,
Nunnally, S. C., . . . Cotton, K. D. (2006). Racial distancing in a southern city: Latino
immigrants’ views of black Americans. Journal of Politics, 68(3), 571-584.
McClain, P. D., & Karnig, A. K. (1990). Black and Hispanic socioeconomic and political
competition. American Political Science Review, 84(02), 535-545.
McClain, P. D., Lyle, M. L., Carter, N. M., Soto, V. M. D., Lackey, G. F., Cotton, K. D., . . .
Kendrick, J. A. (2007). Black Americans and Latino immigrants in a southern city. Du
Bois Review: Social Science and Research on Race, 4(01), 97-117.
Oliver, J. E., & Mendelberg, T. (2000). Reconsidering the environmental determinants of white
racial attitudes. American Journal of Political Science, 574-589.
Phan, N., & Garcia, J. A. (2009). Asian‐Pacific‐American Partisanship: Dynamics of
Partisan and Nonpartisan Identities. Social Science Quarterly, 90(4), 886-910.
Ramakrishnan, K., Junn, J., Lee, T., & Wong, J. (2008). National Asian American Survey, 2008.
Ramakrishnan, K., Lee, J., Lee, T., & Wong, J. (2016). “National Asian American Survey
(NAAS) 2016: Post-Election Survey.
Ramakrishnan, K., Wong, J., Lee, T., & Junn, J. (2009). Race-Based Considerations and the
Obama Vote. Du Bois Review: Social Science Research on Race, 6(01), 219-238.
Raychaudhuri, T. (2020). Socializing Democrats: Examining Asian American vote choice with
evidence from a national survey. Electoral Studies, 63, 102114.
Samson, F. (2015). Asian American attitudes towards a US citizenship path for illegal
immigrants: Immigration reform as racialised politics. Journal of Ethnic and Migration
Studies, 41(1), 117-137.
Sherif, M., Harvey, O., While, B., Hood, W., & Sherif, C. (1961). Intergroup cooperation and
competition: The Robbers Cave experiment. Norman, OK: University Book Exchange.
Wong, J. (2000). The effects of age and political exposure on the development of party
identification among Asian American and Latino immigrants in the United States.
Political Behavior, 22(4), 341-371.
Wong, J. (2018). Immigrants, evangelicals, and politics in an era of demographic change:
Russell Sage Foundation.
Wong, J., Ramakrishnan, K., Lee, T., & Junn, J. (2011). Asian American political participation:
Emerging constituents and their political identities: Russell Sage Foundation.
Xu, J. (2005). Why do minorities participate less? The effects of immigration, education, and
electoral process on Asian American voter registration and turnout. Social science
research, 34(4), 682-702.
Zárate, M. A., Garcia, B., Garza, A. A., & Hitlan, R. T. (2004). Cultural threat and perceived
realistic group conflict as dual predictors of prejudice. Journal of experimental social
psychology, 40(1), 99-105.
Zheng, B. Q. (2019). The Patterns of Asian Americans’ Partisan Choice: Policy Preferences
and Racial Consciousness. Social Science Quarterly, 100(5), 1593-1608.
25
Appendix
Table 1: OLS Results for the Commonality on Partisanship
2008 Survey
2016 Survey
Model 1
Model 2
Model 3
Model 4
Commonality to African American
-0.05*
-0.03+
-0.07**
-0.05**
(0.03)
(0.02)
(0.03)
(0.02)
Commonality to Latino American
-0.05*
-0.04*
-0.07**
-0.05**
(0.03)
(0.02)
(0.03)
(0.02)
Commonality to White American
0.06**
0.04**
0.06**
0.05**
(0.02)
(0.02)
(0.03)
(0.02)
Control Variables
Yes
Yes
Yes
Yes
Missing Variable Indicators
Yes
Yes
Yes
Yes
N
2993
5159
2706
4438
The dependent variable is the partisanship index, with higher value means more Republican.
In model 1 and 3, we treat the missing data as missing. In model 2 and 4, we treat them as
midpoint. The independent variable in all models is treated the same way. * p<0.05 **
p<0.01 *** p<0.001, + (p=0.051) (one-tailed)
Table 2: OLS Results for the Commonality between Minorities and White American
2008 Survey
2016 Survey
DV: Commonality to White American
Model 1
Model 2
Model 3
Model 4
Commonality to African American
0.32***
0.31***
0.31***
0.31***
(0.02)
(0.02)
(0.02)
(0.02)
Commonality to Latino American
0.24***
0.24***
0.24***
0.24***
(0.02)
(0.02)
(0.02)
(0.02)
Control Variables
Yes
Yes
Yes
Yes
Missing Variable Indicators
Yes
Yes
Yes
Yes
N
5159
5159
4438
4438
The dependent variable is the commonality with White American. In model 1 and 3, we treat
the missing data as missing. In model 2 and 4, we treat them as midpoint. Mode 1-2 use 2008
NAAS data. Model 3-4 use 2016 NAAS data. * p<0.05 ** p<0.01 *** p<0.001, (one-tailed)
26
Figure 1: CIplot for Experiment Groups on partisanship
Figure 2: Extended Analysis of Experiment Scrutinizing Personal Prompt Subgroup
.3 .4 .5 .6 .7 .8
Control Competing Competing+indi
w/o unfairness recall
Competing+Indi
w/ unfairness recall
Note: CI=95% (One−way); dashed line is the mean of control group
Level of Republican Identity
