Fault Lines Among Asian Americans: Convergence and Divergence in Policy Opinion
Sociologists have long argued that a racial and ethnic category can be made under certain circumstances and remade under anothers. The pan ethnic category of Asian American was in fact devised to mobilize different immigrant-origin groups in the 1960s. Today’s Asian Americans have diverse opinions on numerous issues. Given this divergence, it is possible that the category will be subject to remaking or unmaking. Using survey questions from the 2016 pre-election National Asian American Survey, this article analyzes where respondents’ policy opinions converge and diverge. Using latent class models, it shows that though many Asian Americans support government interventions in health care, education, climate change, and racial justice, some diverge sharply in regard to Muslim immigration. Logistic regression models show that different experiences of immigration and differences in national origins undergird such divergence. I discuss the implications of these fault lines for the future of the Asian American category.
Asian American, NAAS, policy opinions, immigration, latent class models
Broadly speaking, Asian Americans are people who can trace their roots to countries throughout East Asia, South Asia, and South East Asia. Obscured by this broad definition of “Asian” and “Asian American” is a staggering diversity of peoples that represent twenty-four distinct groups. . . . it is fair to ask whether there is even one “Asian American,” or one “Asian American History.” Asian Americans with long roots in this country may wonder what they have in common with today’s recent arrivals. Similarly, new Asian immigrants and their descendants may not think that the histories of earlier Asian Americans are relevant to their own experiences. But they should.
In her comprehensive history of Asian America, Erika Lee (2015) opens her discussion by acknowledging the complex meaning of the term [End Page 46] Asian American. Undercutting the purported unity of geographic origin—the continent of Asia covers more than 30 percent of the earth’s total land area—is “a staggering diversity of peoples” whose experiences diverge sharply from one another. This heterogeneity notwithstanding, Lee argues, these peoples should believe in the relevance of a collective past to their everyday lives. In Lee’s account, each successive generation of immigrants from Asia faced similar sets of challenges in the United States, and in immigrant communities, often relied on the direct and indirect support of the their predecessors, however disparate, while forging pathways in their newfound homeland. In a way, diverse groups of immigrants from different parts of the vast continent of Asia have been connected and continue to connect through the shared history of immigration and adaptation.
Whether they could or actually do believe and invest in a common identity as a people is, however, an empirical question. Philosophers and historians of science have studied various occasions in which “making up people” occurred (Hacking 2002). Sociologists have tirelessly pointed out the historical contingency of labels such as race, ethnicity, and nation. That is, although these labels have exercised much power over the making of our times, they are by no means natural entities that spontaneously emerge from shared traits among individuals. Instead, they are products of particular social and historical circumstances, and as such can be made, unmade, and remade through state policies, social movements, and cultural campaigns, to mention just a few factors (Wimmer 2012; Brubaker, Loveman, and Stamatov 2004; Anderson 1983).
The mutable character of ethnicity is especially more prominent in the case of panethnic categories such as Asian American, in which both primordial and instrumental dimensions of ethnicity give way to political and cultural processes (Espiritu 1992; Okamoto and Mora 2014). In the first half of the twentieth century, immigrants moving to the United States from various parts of the continent labeled Asia did not identify as Asians. When asked by inspectors of immigration stations, such as in Angel Island (Lee and Yung 2010), where they were from, they uttered village name, regional origin, point of departure, or family pedigree. Usually, however, they stuck to their national origins, in most cases China or Japan. As David Hollinger (1995) explains through his concept of “ethno-racial pentagon,” it was not until the aftermath of the civil rights and antiwar movement of the 1960s that the category of Asian American emerged as a collective identity encompassing these diverse groups of immigrants and their children, presumably to range them alongside whites, blacks, Hispanics, and Native Americans—who were, in fact, also equally heterogeneous in their own regard.
In short, Asian American was a category made from the particular circumstances of the 1960s out of diverse experiences of immigration. Hence its content and boundaries can be renegotiated in different circumstances. Moreover, in an unlikely but not entirely impossible scenario, we can even imagine an unmaking of the category, in which various subgroups making up the panethnic label embark on their identity formation or regroup under different banners. Although many Asian Americans have made notable strides in education and business in the latter half of the twentieth century (Alba and Nee 2003), many of the prejudices and stereotypes targeting Asian Americans (C. Kim 1999)—such as “foreign,” unable to speak English, lacking in leadership and creative qualities, overachieving, and so on—have not disappeared entirely. In fact, in many cases Asian Americans are slotted in marginal, technical positions in organizations and barred from the powerful, decision-making posts—the phenomenon commonly referred to as “bamboo ceiling” (Chin 2020). In addition, with a turn toward the knowledge economy and the rise of Silicon Valley geek culture, the stereotype of smart but unimaginative Indian and Chinese tech workers emerged as a target of anti-immigrant advocates, who imagined such undeserving and “un-American” workers taking jobs that should be given to “American” workers.
Yet in recent years, Asian Americans have also witnessed important changes in how they were understood in American society. The economic development of East Asian countries— “Asian Ascendancy” (Hoang 2015)—and increasing [End Page 47] significance of pan-Pacific economic connections are complicating the positions of Asian Americans in American society. As the widely popular movie Crazy Rich Asians (2018) showcased, Asian Americans are now being portrayed not just as perpetual foreigners but also as gatekeepers to glamorous foreign wealth. Andrew Yang’s unexpected performance in the presidential primary suggested that the wealth and cultural competency depicted in the movie have a potential to turn into real-world political gains. However, the most recent turn of events seems to negate all these incremental positive changes: the incendiary rhetoric of “kung flu” amid the unprecedented global pandemic ushered a new wave of physical and verbal abuse targeting all Asian-looking persons across the United States, regardless of their citizenship status or immigration history, once again reminding Asian Americans that their social standing in American society has not moved very far from that of the Yellow Peril era (Lee and Yadav 2020). Although Asian Americans have certainly grown in numbers and gained some visibility, in the popular imagination they remain at the margins of American social and cultural landscape. Kung flu is a cruel reminder that their social standing is still precarious after all these years in the United States, despite their achievements in education, culture, and economy.
It is unclear how these trends will affect Asian American as a category of identity. They may turn the clock backward and remake or unmake the category, possibly reenforcing peoplehood based on national origins and class over panethnic identity. It is equally plausible, however, to expect Asian American to increase its significance and gravity for the generation born well after the emergence of the label. As researchers of the National Asian American Survey (NAAS) project have recently demonstrated (Ramakrishnan et al. 2018), Asian Americans are becoming a political force to be reckoned with, especially as the most dedicated supporters of the Democratic party and its liberal policies (see also Wong and Shah 2021, this issue). Thus it is reasonable to assume that politically Asian American will be more prominent as a category, even as the people linked to the label become culturally and socioeconomically more differentiated (see Drouhot and Garip 2021, this issue).
This article attempts to identify some of the potential fault lines among Asian Americans, based on which making, remaking, and even unmaking of the category can occur in the future. In the 1960s, a common cause united immigrant families from China, Japan, and other parts of the world as Asian Americans. Using survey data on policy opinions of Asian Americans from the 2016 pre-election NAAS, I explore whether a common cause still holds together various constituents of the category and how different subsections of the Asian American population converge and diverge on the important issues of our times. Latent class models on seven policy opinion questions ranging from the Affordable Care Act and free college tuition to the admission of Syrian refugees and the Muslim travel ban reveal that opinions on the latter two policies sharply diverge among different subgroups of Asian Americans who otherwise share the support for active government intervention on other issues. Certain demographic variables correlate to the policy positions. Analysis reveals how these findings map onto ethnic divisions, forming fault lines within particular Asian American subgroups. Combined with the results from additional regression analyses of the questions on Syrian refugees and the Muslim travel ban, the analysis shows a major divide within the Asian American category: the older, less educated, and foreign born versus the younger, educated, and native born. The ethnic division between Southeast Asians (Cambodian and Vietnamese) and other Asians (Chinese, Japanese, Korean, and Indians) somewhat corresponds to this divide but not entirely.
PANETHNIC IDENTITY AND POLITICAL MOBILIZATION: A HISTORICAL OVERVIEW
Before the twentieth century, the people living in the continent designated Asia by Europeans did not see themselves through a unified category, or as Asians. The long, often violent, history of Sinocentrism in China shows that within Asia, ethnic differences mattered more than the panethnic identity, and the complicated histories of various states in South and [End Page 48] Central Asia invalidate such simple categorization. Even when European race theorists embarked on their long journey to catalog and categorize peoples of the world, Asia was seldom seen as a place for one people. For instance, in the late eighteenth century, Johann Fredrich Blumenbach, known for coining the “color races” scheme (white, black, yellow, and so on), distinguished between yellow and brown people, or between Mongolians and Malayans (Keevak 2011). The rise of nationalism and imperialism in the nineteenth century complicated this scheme further, adding ever more fine-grained divisions among peoples of Asia and other parts of the world to account for their supposed differences (Gossett 1963).
The demise of scientific racism and challenges against European imperialism in the first half of the twentieth century changed the situation somewhat. As Donna Jones (2010) demonstrates in her reading of early twentieth-century Caribbean writers, pan-African consciousness emerged as an aesthetic and political counterpoint to pan-European conceptions of whiteness, those which undergirded the cultural landscape of imperialism. In an entirely different context, the discussions around immigrant assimilation in the United States also forged a way for a less monolithic concept of whiteness that encompassed differences among various European immigrants (Roediger 2005; Jacobson 1999).
The situation was, however, quite different for various immigrants of Asian origin. From the 1880s and onward, ethnic rivalry and conflict among Asian immigrant groups had always been boiling under the blanket imposition of white supremacy and nativism. Nativists in California used the term Oriental to simultaneously deride Chinese, Japanese, and Korean immigrants, but members of these groups strongly rejected the term, not only because it was de-meaning but also because they did not appreciate being labeled into a single category with other groups. As the historian Eiichiro Azuma (2005) discusses at length, asserting distinction from Chinese workers was in fact a focal point of the early Japanese immigrant identity. Chinese immigrants to the United States in the nineteenth century were mostly poor farmers from rural villages of Guangdong province. As such, they seldom had any education before they immigrated and were often victims of extortion and trafficking by both Chinese and American labor brokers. Many were temporary workers in the most demanding and dangerous jobs, such as railroad construction and mining, and intended to return to their hometowns as soon as they saved enough money. To the progressive reformers of the era, these Chinese workers cannot be any more at odds with the American ideals of citizenship upholding the value of free, dignified labor (Jung 2006).
After passage of the Chinese Exclusion Act of 1882, Japanese immigrants began to arrive to fill the gap in the labor market left by the Chinese workers who had returned to China. Unlike their Chinese predecessors, Japanese immigrants had enjoyed relative high standing in Japanese society, as sons of independent farmers and Samurais, but then experienced relative downward mobility after the Meiji restoration. Given this background, many had secondary education, often from Western-influenced schools, and hoped to eventually continue their education in the United States. They saw themselves as settlers, not sojourners, and envisioned a future in which they would command the respect of Americans—native-born whites, to be specific—through honest work and upward social mobility. Although Chinese and Japanese workers were often working alongside each other as farmworkers in the Central Valley and domestic servants in San Francisco households, in their minds they could not have been any more different from each other. That Americans confused or did not care enough to differentiate these two groups frustrated ambitious Japanese immigrants. Hence they distanced themselves from Chinese immigrants whenever the opportunity was presented, and made such distancing (“desinofication”) a major strategy to advance their position in American society.
Korean immigrants maintained a similar attitude to assert their difference from the Japanese. Although the peninsula was colonized by Japan in the early twentieth century and Koreans were officially the subjects of the Japanese empire, in many instances they strongly denied such affiliation, instead opting to argue for Korean identity and supporting the independence [End Page 49] movement. Filipino immigrants were situated in a very different position from these groups because they were once U.S. subjects through colonization and culturally identified more with Spanish-speaking, Catholic communities, such as Mexicans. South Asians were known as Hindoos, but no one thought to group them together with other Asian immigrants. In short, the idea that these immigrants from Asia had something in common—other than being immigrants and working in manual jobs, which was also true for many European and Mexican immigrants—was in this historical context virtually unthinkable (for a comprehensive history of these immigrant communities, see E. Lee 2015). The 1920s ushered in a series of immigration policies focused on restriction, and all of these groups were barred from entering the United States under the Asiatic barred zone. Under the influence of the Naturalization Act of 1790, which only allowed whites to become citizens, no one with origins in Asia was permitted to naturalize, though many mounted legal challenges that tested the cultural boundaries of whiteness (Haney-Lopéz 2006). In the meantime, the children of these immigrants were being born on U.S. soil and, by virtue of the Fourteenth Amendment, were automatically granted U.S. citizenship, even though they did not enjoy the full civic and social rights reserved for whites.
In the 1960s, however, the circumstances around Asian immigrants and their children changed somewhat. As the Jim Crow regime and restrictive immigration policies were struck down through the Civil Rights Act and the Immigration Act of 1965, various actors began to reimagine racial and ethnic identity. The Black Power movement, spearheaded by the Black Panther Party, envisioned a collective front of “black people” united against white supremacy, a movement that included not only African Americans but also all members of the African diaspora in various colonies of the world (Bloom and Martin 2012). Panthers later extended their conception of blackness, arguing that, just like the color itself, blackness can and should encompass all people of color, most notably the peoples of Vietnam and North Korea, who were engaged in their own struggles against U.S. imperialism. Many others followed this frame and envisioned political movements of panethnic “people.” The “red power” movement and American Indian ethnic identity renewal (Nagel 1997) was partially inspired by the black militancy of the 1960s and 1970s. Although the Hispanic category emerged more through the cultural work of experts rather than popular social movements (Mora 2014), the renewed sensibility around the political usefulness of panethnic identity certainly helped its initial take-off (Espiritu 1992).
Asian Americans, however, were directly influenced by the Black Power movement, at least in terms of how they came to envision their collective identity (Ishizuka 2016). Drawing on interviews of the activists, William Wei (1993) pinpoints the specific time and place in which the term Asian American emerged. In 1968, Berkeley and San Francisco were the hotbeds of everything radical, including new forms of political mobilization based on racial and ethnic identity. During the Third World Liberation Front strike at San Francisco State College, student and community activists of Chinese and Japanese origins needed a new way of overcoming differences among themselves and mobilizing for political action. Just like many other Americans of their generation, these activists grew up surrounded by the consumer-oriented, conservative culture of the 1950s, but came to witness the social ills woven into the ostensibly peaceful fabric of Cold War era material abundance. In their eyes, domestic racism, of which they were direct victims, and the war in Vietnam were the two sides of the same coin representing the contradictions of white supremacy. They became active participants of campus-based social movements, including the Students for a Democratic Society and the Black Panthers. Neither organization, however, wholly embraced the presence of Chinese and Japanese participants, and those children of immigrants started to question their places in a society marked by the black-white divide. In the end, they realized that they needed their own social movement, one that addressed their shared concerns around war, racism, and their immigrant communities. The category of Asian American was consciously devised to address these circumstances.
After the turbulent decade in which Asian [End Page 50] American was used more as a slogan in street demonstrations than as an everyday term, the term finally became a part of the official lexicon. In 1977, Asian American was enshrined into civil rights legislation as a rightful component of the “ethno-racial pentagon” through the Office of Management and Budget (OMB) Statistical Directive No. 15 (Hollinger 1995), and suddenly it seemed that Chinese, Japanese, Koreans, Filipinos, and other immigrant groups had something in common with each other, as Asian Americans.
Although the term has survived to our times, the actual makeup of the people under the label shifted dramatically. In the 1960s, Asian Americans denoted mostly those of Chinese, Japanese, and Filipino origins but included a small number of immigrants and their children from South Asia. Because of the immigration restrictions placed in the 1920s, these people have been in the United States for a generation or two, and the young members of these groups grew up effectively cut off from their parents’ or grandparents’ homelands. But they came to see each other as members of a unified racial group, largely through shared opinions on issues (such as the Vietnam War and racism) and consequent political mobilization. As noted in the introduction, the Vietnam War ended a long time ago and Asia has since become an economic powerhouse; many Asian Americans have moved out of ethnic enclaves and some of them exercise a considerable amount of economic and cultural power within American society (Jiménez 2017). Many more waves of newcomers, some of them refugees, from different parts of Asia arrived since the 1960s, embodying a dizzying array of diversity in their social and cultural backgrounds. Young members of these groups maintain vibrant connections to their origins and other countries of Asia, navigating a very different cultural landscape than the one presented to Chinese and Japanese youth of the 1960s. Political mobilization under the label of Asian American still remains, but in a different format and with less passion. The discontinuity between the Asian Americans of the 1960s and those of the 2020s is pronounced, yet as the kung flu reference to the coronavirus has painfully demonstrated, a clear connection exists between the marginalized positions on which the old and new groups stand. In other words, just like then, today’s Asian Americans cannot be separated from their direct and indirect experiences of immigration, and thus are vulnerable to the nativist ideology that plagues the American political culture every so often.
Where, then, does this put Asian American as a category, given rifts and shifts among groups? It is one thing to trace how the American public—mostly native-born whites, to be specific—perceived Asian Americans, but how Asian Americans saw themselves is an entirely different matter. In the 1960s, shared opinions on the Vietnam War and domestic racism brought together different immigrant youth into the panethnic label of Asian American, supposedly to empower their collective mobilization. How do today’s Asian Americans feel about issues? Do they have a more or less common and coherent stance on various policies, or are there points of divergence within people categorized as Asian Americans?
POLICY OPINIONS: CONVERGENCE OR DIVERGENCE?
Based on this historical overview, we arrive at a succinct articulation of the relationship between policy opinions and group-making in the case of Asian Americans. First, panethnic identity, such as Asian American, is made when policy opinions converge among ethnic groups; second, and conversely, panethnic identity can be remade or even unmade when policy opinions diverge, not only along ethnic lines but also in terms of demographic characteristics. The first point is well documented in the emergence of the Asian American label in the late 1960s, and the second in the demise of Asian American movement in 1970s. The aim of this article is to draw inspirations from these experiences and examine whether any of the trajectories are useful in thinking about contemporary Asian Americans.
In doing so, I focus on five policy dimensions: welfare, immigration, culture, environment, and race. Each is an important pillar of contemporary political divide, progressives and conservatives diverging sharply on how government should approach them. In terms of welfare, I use the questions on the most visible government-led programs, which are often topics [End Page 51] of debate: the Affordable Care Act and the proposal for government-sponsored free college tuition. The same applies to the questions related to immigration: Syrian refugees and the Muslim ban had received much media attention in the period leading up to the time of the survey. Culture, environment, and racial justice are each measured by a single question asking for opinion on specific government policy—legalization of marijuana, limiting coal-based power plants, and government stepping up its role in assisting blacks. Although these five dimensions leave out some of the important, volatile policy issues—religion, gender, international relations, and economic policy come to mind—they should provide enough grounds on which policy opinions can converge or diverge.
To map out convergence and divergence along these dimensions, I focus on identifying clusters of opinions. That is, using an exploratory data analysis technique—latent class analysis (LCA), to be specific—I highlight how opinions on different issues appear together in groups of respondents. One cluster of respondents may support all of the seven policies being questioned; in another respondents may support five out of seven; in still another respondents may object to a specific policy but support all the others. Although permutations of these configurations are quite large, LCA statistically reduces the potential possibilities into a legible number, demonstrating existence of statistically plausible clusters. The dividing lines that mark the boundaries between clusters can be read as potential fault lines on which remaking or unmaking of the category can occur in the future.
In term of expectations, I envision the results to fall between the two somewhat extreme possibilities. First, we can imagine a near-perfect convergence of policy opinions among Asian Americans: a large chunk of respondents will belong to a single cluster, leaving not many for other clusters, and because the single cluster is so large, ethnic differences in policy opinion clusters will be minimal. This result is close to what occurred among the activists in the 1960s, in which many Chinese and Japanese children of immigrants rallied behind the label of Asian American based on their shared opposition to the Vietnam War and domestic racism. Second is the opposite scenario, in which the divergence of policy opinions is marked: many clusters will have unique configurations, and their distribution will show a sharp divergence along ethnic lines. This result is similar to the times before or after the 1960s, in which differences in opinions along ethnic lines undermined the potential panethnic mobilization under the label of Asian American. In the first case of near-perfect convergence, the category of Asian American is stable and the possibility for its dissolution minimal. In the second case, we can conclude that the possibility that Asian American category will be remade or even un-made is strong.
Both possibilities are extreme scenarios, and the results are not likely to point conclusively one way or the other. The result will be far from a near-perfect convergence but not close to a complete divergence along ethnic lines, either. More productive than testing the two unrealistic hypotheses, though, is to statistically pinpoint where the status of Asian American category lies between the two scenarios. Moreover, by closely combing through the results and considering not only ethnic differences but also demographics, I aim to identify the potential fault lines along which the Asian American category can be remade or unmade in the future. Scholars interested in the political future of Asian Americans would benefit much from paying attention to these fault lines.
I explore these possibilities by applying latent class models to the 2016 pre-election NAAS (for theoretical and methodological principles on using LCA to study racial and ethnic differences, see S. Kim 2019; Drouhot and Garip 2021, this issue). I also use logistics regression models for a closer examination of key variables that emerge from latent class analysis.
DATA, VARIABLES, AND METHOD
Data are drawn from the 2016 pre-election NAAS (N = 4,787), a nationally representative, multilingual survey of Asian Americans. The data set includes nine national origin and ethnic groups (Chinese, Filipino, Indian, Vietnamese, Korean, Japanese, Hmong, Cambodian, and Native Hawaiian–Pacific Islander) as well as whites, blacks, and Latinos as comparison [End Page 52] groups. Except in figure 1, in which I compare Asian American policy opinions with those of whites, blacks, and Latinos, my analyses focus exclusively on the nine groups (N = 3,170). I use the person weight included in the NAAS data set on all models, including the comparison presented in figure 1.1
The main variables used to construct latent classes are questions on policy opinions. The question wording is as follows:
“Next, we will ask your opinions on certain policies. In each case please tell me if you support or oppose.”
Q1. Do you support or oppose the health care law passed by Barack Obama and Congress in 2010?
Q2. Do you support or oppose major new spending by the federal government that would help undergraduates pay tuition at public colleges without needing loans?
Q3. Do you support or oppose accepting Syrian refugees into the United States?
Q4. Do you support or oppose legalizing the possession of small amounts of marijuana for personal use?
Q5. Do you support or oppose banning people who are Muslim from entering the United States?
Q6. Do you support or oppose setting stricter emission limits on power plants in order to address climate change?
Q7. Do you support or oppose the government doing more to give blacks equal rights with whites?
The responses were coded as support, oppose, or “don’t know or refused to answer,” providing 3^7 = 2,187 possible variations in combinations of the answers. In addition, I included an array of individual-level demographic characteristics (gender, education level, foreign born status, and party identification) as covariates to further substantiate the profiles of classes. For racial and ethnic categories, I followed the initial sampling frame and used the nine national origin and ethnic groups.
Table 1 displays the overall response patterns for the seven indicator variables, and figure 1 summarizes how the policy opinions of Asian Americans as a group compare with whites, blacks, and Latinos on those variables. As widely noted, Asian Americans have considerably liberal policy opinions when it comes to government actions: an overwhelming majority support the Affordable Care Act, free college tuition, limits on power plants, and interventions on behalf of blacks. In the same light, 65 percent of the Asian American respondents oppose the Muslim travel ban. However, Asian Americans are less supportive of Syrian refugees (51 percent support and 31 percent oppose) while many oppose the legalization of marijuana (42 percent support and 49 percent oppose). As shown in figure 1, these numbers demonstrate that Asian Americans are far more progressive than native whites, almost on par with Latinos, and slightly behind blacks.
For each question, a considerable number of respondents (from 10 to 20 percent, except for free college tuition) opted to answer don’t know or refused to answer. In his famed essay criticizing public opinion research, “Public Opinion Does Not Exist,” Pierre Bourdieu (1979) argues that public opinion has two layers: the first layer is the ability to have an opinion and express it in a survey interview; and the second layer is the opinion itself. To become legible in public opinion surveys, respondents need to overcome the first hurdle of “having” and “expressing” an opinion on issues, even before deciding what those opinions are. For instance, to properly answer the question on the Syrian refugees, a person needs to understand who these refugees are and why they became an issue in contemporary politics (the first layer), before deciding whether to support or oppose [End Page 53]
their admission to the country (the second layer).
Many scholars of public opinion, according to Bourdieu, only pay attention to those who are capable of expressing opinion, and ignore many members of the public who don’t know or are hesitant to reveal their opinion. However, this does not mean that they are not a part of the democratic public. As Taeku Lee (2002) has shown through his study on the changes in latent public opinion leading up to the civil rights movement, people express what they think through various means, and these expressions may or may not be captured through standard public opinion surveys (see also Lee and Pérez 2014). This criticism is especially meaningful in light of the fact that fewer than half of the Asian American respondents in the 2016 pre-election NAAS data set expressed firm opinions for all of the seven questions, and that more than half had at least one don’t know answer. To affirm these answers as an expression of [End Page 54] opinion, I consider don’t know as a valid response alongside support or oppose and use the response as an important dimension along which subgroups of respondents diverge.
Method: Latent Class Analysis
LCA considers observed variables, such as survey responses, as a function of underlying, unobserved categorical variables termed latent classes. Using a maximum-likelihood approach, LCA finds the optimal number of categories for the latent class variable to explain the variation in the observed variables (Goodman 1974). More intuitively, this approach is akin to how a doctor would diagnose a disease from observed characteristics of a patient. That is, a doctor who saw fever, runny nose, and coughing (observed variables) would likely point to cold (latent class), an underlying variable that predicts all these features. In a nutshell, LCA is a model-based data reduction tool that reduces complexity in real-world data by providing a few templates to think through variations. Some scholars have argued that LCA is close to the “ideal types” usually employed in the analysis of qualitative data (Hagenaars and Halman 1989). The formal equation of the model I use here is as follows (see Bonikowski and DiMaggio 2016):
The left side of the formula notes the probability of a given combination of observations on the variables (y) included in the model. X stands for latent variable, which comprises a number of latent classes (t) (see also Goodman 1974). Latent class models attempt to identify the minimum number of latent classes, or t, to account for the variations in data.
I also explore three extensions of LCA (Bonikowski and DiMaggio 2016; see also Knight and Brinton 2018). First, in constructing latent class variables with the seven survey questions on policy opinions, I consider local dependence between the survey responses (Vermunt and Magidson 2002). Standard latent class models assume that all of the indicator variables for latent class variables are independent from each other, but this assumption in many cases is not satisfied, which results in underfitting of the model. Local dependence is usually confirmed by observing bivariate residuals between indicators after the model’s initial fitting. Specifying these relationships in the model improves the fit in many cases. Second, I include a number of individual-level demographic characteristics as covariates, which do not intervene in construction of latent classes but do contribute to the assignment of individual cases into the classes. Scholars use covariates in LCA much like independent variables in regression models, to assess how individual-level background characteristics map onto latent classes. Third, in observing distribution of latent classes across racial and ethnic groups, I use the improved three-step approach developed by Jeroen Vermunt (2010). A simple observation of distributions often ignores classification errors embedded in the individual assignment of classes, and therefore produces biased results. The three-step method takes a probabilistic approach to latent class allocation and thereby produces a more accurate distribution across categories. I use Latent Gold 5.1 for all analyses (for a detailed mathematical exposition of latent class models, see Drouhot and Garip 2021, this issue).
No single, absolute criterion has been established to determine the number of classes for a given latent class model. Many scholars, though, adopt Bayesian Information Criteria (BIC), along with other parameters and substantive meaning of the model, as their guideline. Unlike other parameters, BIC considers both model fit and number of variables to examine whether an additional class actually provides more information or merely makes the results more difficult to interpret. In my analysis shown in table 2 and figure 2, BIC reaches the lowest point at five classes, and flattens out afterward, indicating that having more than five classes does not improve the quality of the information generated. I therefore choose a five-class model for my analysis.
After initial fitting of the model, I identified three pairs of indicators that had considerable bivariate residuals, which signaled that the pairs are far from independent of each other. [End Page 55]
As it turned out, these pairs were closely related in terms of their substance as well as statistical properties. The support for Affordable Care Act (Q1) usually appeared together with the support for free college tuition (Q2); those who opposed Syrian refugees (Q3) tended to support the Muslim travel ban (Q5); and respondents who supported limits on power plants (Q6) also thought the government should do more to help blacks (Q7). The model was revised to recognize the relationships of local dependence in these indicators.
Table 3 displays the profiles of five classes, along with the overall percentage of respondents slotted into each class (overall class size) and overall response rate for the questions (overall). The conditional probabilities for response categories should be considered alongside overall response rate. In other words, 0.55 for the Affordable Care Act support in class A indicates that those in this class are likely to support Affordable Care Act (0.55) less (0.05) than the overall average (0.60) of respondents across all classes, though the number in itself is high. However, 0.27 for the support of the Muslim travel ban in class A may be not be high in itself but relative to the overall support rate (0.18) is quite substantial. In fact, class A comes in as the most likely to support the Muslim travel ban, the policy that remains generally unpopular among Asian Americans.
Among the five classes, classes D and E lend themselves to relatively straightforward categorizations. Class E, the smallest in size (0.04), [End Page 56]
features very high likelihood of answering don’t know for all seven questions, indicating that these are respondents who belong to Bourdieu’s first layer of public opinion, in which they are unable—or unwilling—to express their opinion. On the other hand, compared with A, B, and C, those in class D express considerably less support for all of the policies being discussed. One may label this as conservative but that they support the Muslim travel ban less than the average (0.16 to 0.18) puts that assessment into question. Nevertheless, class D is the only one among the five classes whose opinions consistently leans conservative. More interesting are the differences between A, B, and C, all of which express strong support for at least some of the policies. Class B is easier to understand than others: they seem to follow the standard progressive line in all of the policies, supporting active government intervention on health care, education, climate change, and racial justice, welcoming immigrants, and even espousing cultural liberalism through marijuana legalization. [End Page 57]
The profiles of A and C are less straightforward to interpret, but A displays a unique feature: strongest support for the Muslim travel ban (0.27 relative to the average of 0.18). Moreover, their opposition to the admission of Syrian refugees comes in second only to class D (0.49 and 0.93, respectively). Considering their relative strong support for the Muslim travel ban, it is reasonable to assume that at least some of those in class A are thinking of Syrian refugees and Muslims along the same line, as unwanted Others who should not be admitted to the country. In regard to other policy opinions, they seem to be following the progressive line, other than that they oppose the legalization of marijuana. The respondents in class C express moderate opinions on many issues, having slightly higher rates of don’t know than other classes except E. This pattern is most visible in the two immigration-related questions, on Syrian refugees and the Muslim travel ban, showing high probabilities of answering don’t know to these questions.
Table 4 displays conditional probabilities for covariates, standardized around the overall average. A positive value indicates that the class is more likely to overlap with the respective categorical variable. Higher education, as expected, is strongly correlated with having an opinion (not answering don’t know), though women tend toward don’t know more than men. Interestingly, classes A and C, who were more likely to oppose Syrian refugees and support the Muslim travel ban than others (except for D), are also more likely to be foreign born than native born, indicating that many of them are immigrants themselves. Party identification aligns with policy opinions: B aligns with Democrats and A and D lean Republican; C and E, which have a relatively high proportion of don’t know, are more likely to overlap with independents.
Distribution Across Racial and Ethnic Categories
Table 5 displays distribution of classes by racial and ethnic groups, using the three-step approach. The numbers indicate the absolute percentages and the plus and minus numbers in parentheses represent standardized percentages around the overall class size. For instance, 0.48(+13) for class A indicates that 48 percent of Native Hawaiian–Pacific Islanders was allocated into class A, and this is larger (+13) than [End Page 58]
the average rate (0.36).2 Therefore, a plus sign indicates that the group is relatively more likely to belong to the class, whereas minus indicates less. These results can be interpreted in various ways but the most important finding is that racial and ethnic groups are not monolithic entities with a unified opinion on policy issues. On the contrary, they are nested structures that encompass a diversity of opinions, represented by different classes (S. Kim 2019). In other words, as expected, policy opinions do not converge perfectly among Asian Americans, but at the same time do not diverge perfectly along ethnic divide either. There are dominant clusters in each group, but all five classes are present through most of the nine groups included in the analysis.
Table 6 summarizes the results, presented in a schematic fashion. Taking a step further from the specific features of each class, we are now able to witness the three axes of differentiation among five classes, each nested in one another. The first axis concerns those who have an opinion and those who do not (Bourdieu 1979), and puts class E against all other classes. Thus I name class E undecided, indicating that the absolute majority of them do not express their opinions on the issues being discussed. The second axis reveals the conventional liberal-conservative divide, between classes D and A, B, and C. Although class D features less-than-average support for the Muslim travel ban, I name the class conservative, noting their otherwise very consistent opposition against government interventions in many issues. The final axis differentiates class B from classes A and C along the lines of their opinions on, among other things, immigration—Syrian refugees and the Muslim travel ban, to be specific. Respondents in these three classes express considerable support for active government intervention in the realms of health care, education, climate change, and racial justice, but class A is relatively less enthusiastic about any form of Muslim immigration, and class C seems to maintain no strong opinion on the issue. On the contrary, individuals in class B express strong support for Muslim immigration. Hence I designate class B progressives; A immigration reservationists; and C immigration undecided. Whereas immigration undecided are reluctant to express a clear opposition to immigration, immigration reservationists come in second to conservatives in terms of their ambivalence to the Muslim immigration. I interpret their attitude as not necessarily opposed to immigration in general, but as having second thoughts about certain kinds of immigrants—Muslims, to be specific. Although it is not clear from the data whether this ambivalent attitude of immigration [End Page 59]
[End Page 60]
reservationists extends to other kinds of immigrants, it is fair to assume that their support for immigration is not as wholehearted as that of progressives. Also, given the intense anti-immigrant political agitation in contemporary politics, this ambivalent feeling could turn into a negative one.
The results from covariate analysis add another interesting dimension to these findings. As expected, foreign-born, less educated, and female respondents incline toward don’t know responses more than their counterparts. Yet immigration reservationists are slightly more likely to be foreign born than native born, indicating that immigrants are espousing an ambivalent attitude toward Muslim immigrants, even more so than progressives, who are more likely to be native born. Cambodians and Koreans are major groups in this class, with the Chinese and Filipinos following.
In summary, results from the five-class latent class model show that Muslim immigration divides an otherwise very coherent coalition of active government supporters (immigration reservationists, progressives, and immigration undecided) and that foreign-born status, education level, and ethnicity provide additional points of divergence.
Additional Analysis: Logistic Regression Models
To further explore the divergence of opinion around the issues of Syrian refugees and the Muslim travel ban, I conducted additional analysis using two sets of logistic regression models, including the two questions as dependent variables and including a series of demographic variables as independent variables, respectively. Table 7 presents the simplified results.
The response categories for the question of Syrian refugees were reverse-coded (0 = support the admission; 1 = oppose the admission), so the positive coefficients represent negative opinions on immigrants in both questions. Standard errors are presented in parentheses. ref indicates the reference category to which the coefficients of independent variable is compared. The most notable pattern concerns the age of respondents: older respondents are more likely to hold negative opinions on Muslim immigration. In addition, foreign-born respondents are significantly more likely to oppose the admission of Syrian refugees, though this was not true for the Muslim ban. Ethnic differences are also notable, with some groups (Indian, Filipino, Hmong, Japanese, and Korean for Syrian refugees; Indian for the Muslim ban) significantly less likely to support anti-immigration policies.
Interestingly, Cambodians emerge as having the strongest support of the Muslim ban, showcasing the largest (1.001) coefficient among all the variables in the analysis. This hostility may be attributable to the homeland politics in the mainland Southeast Asia. The Chams, an ethnic minority group that traces its roots to the ancient kingdom of Champa, have lived in the region for centuries and maintained a distinctive religious identity through their Muslim faith (Trankell and Ovesen 2004; Scupin 1995). Although their numbers are small—0.2 and 1.6 percent of the overall population in Vietnam and Cambodia, respectively (Pew Forum on Religion and Public Life 2009)—they have often been the target of prosecution by various regimes that governed this region. Most notably in Cambodia, the Chams were subjected to genocide under the Khmer Rouge regime. Although their social visibility dwindled after the fall of the dictatorship, recently the minority group emerged once again as the beneficiaries of Islamic internationalism as aid from the Arab world and Malaysia poured in to assist them in their religious practices and education (Bruckmayr 2006). Although it is not clear how these trends factor into the attitudes of Asian Americans from this region, it is reasonable to assume that the Muslim minority in the mainland Southeast Asia had been stigmatized in the past, and that the stigma still holds some effect among the diaspora in the United States, especially for first-generation immigrants above a certain age. The results presented in table 7 suggest this, but additional studies are needed to accurately assess the contours and origins of the anti-Muslim sentiments among Cambodians, or Asian Americans more generally.
Interestingly, the Hmong are significantly more likely to support the admission of Syrian refugees although many of them hail from the same geographical region as Cambodians. I [End Page 61]
speculate that the lasting impact of refugee experiences in the Hmong community (see E. Lee 2015, chapter 15) may have contributed to this unique pattern, although further research, most likely based on interview data, is required to unpack the dynamic beneath this “refugee consciousness” across ethnoracial lines.3 [End Page 62]
DISCUSSION: IMMIGRATION AS POTENTIAL FAULT LINE?
In the 1960s, the shared concern around war and racism led Japanese and Chinese children of immigrants to rally around Asian American identity. They envisioned a political force based on the panethnic category, which would transcend old rivalries between different immigrant groups. As William Wei (1993) narrates, however, this unity did not last long. For instance, in New York City, older, more educated Japanese Americans soon discovered that their concerns were quite different from those of their younger Chinese American comrades who grew up in Manhattan’s Chinatown. Whereas Japanese Americans wanted to focus more on antiwar activism and follow the lines of Students for a Democratic Society, young Chinese Americans were inspired by the Panthers, and focused more on the community politics around Chinatown. In New York City and other places, internal disputes like these led many Asian American organizations to dissolve in the 1970s, leading to the decline of active mobilization around the category. In other words, while the shared opinions on war and racism led to the emergence of a panethnic identity, the differences based on generation, class, and ethnicity led to differences in policy opinions, eventually resulting in the decline of activism centered on the Asian American category.
The arc of history around the Asian American movement dramatically demonstrates that categories can be made, unmade, and remade over time as political mobilization around shared concerns waxes and wanes. The results presented in tables 6 and 7 suggest several potential fault lines, many of which form a cluster without completely overlapping with each other. The different experiences of immigration, captured through nativity and age, are correlated with different levels of education and, consequently, different social positions within American society (see Alba and Nee 2003; Drouhot and Garip 2021, this issue). On one side are first-generation, older immigrants with less experience of American institutions, including educational ones; on the other side are second-and third-generation, young Asian Americans, with extensive experience. This divide maps roughly—but not entirely—onto the ethnic divide, with the newcomers (Southeast Asians) on the one side and the more established immigrants and their children (East Asians and Indians) on the other. Finally, although many Asian Americans on both sides of this divide support active government interventions on the issues such as health care, education, climate change, and racial justice, they seem to diverge on the issue of immigration, or, more precisely, on the question of Muslim immigration.
Although we cannot be sure whether the concerns about Muslim immigration will extend to Mexican immigration or immigration more generally, it is certainly possible that a significant subsection of Asian Americans will either oppose or be indifferent to future immigrants, especially when those immigrants are actively framed by politicians and media as the symbolic Other posing a threat to the established social order. We can imagine a populist-type candidate appealing to less educated Korean and Cambodian first-generation immigrant men, campaigning for active government intervention on the economy while using Trump-like rhetoric to target some immigrant groups.
Implausible as it may sound, my analysis of the NAAS indicates multiple subgroups of Asian Americans under the overall, simplified understanding of them as devout Democrats. The results from other articles in this issue also point in this direction. Van Tran and Natasha Warikoo (2021) show different levels (and even directions) of support for immigration policies among Asian Americans of different backgrounds. Most notably, whereas later generations of Asian Americans strongly support the pathway for citizenship for undocumented immigrants, they seem to be less enthusiastic about work visas and family unification policies. A strong ethnic divide is also evident, although its exact configuration differs considerably from my analysis. Rujun Yang and Maria Charles (2021) also identify divergence of opinions among Asian Americans on issues of gender politics, such as abortion, LGBTQ rights, and transgender rights. Here again, experiences of immigration emerge as an important fault line along with the differences by ethnicity. [End Page 63]
In other words, in many dimensions we are looking at the two distinctive subgroups of Asian Americans, who not only differ in their immigration experiences and socioeconomic standings but also diverge sharply in policy opinions. The stereotypical understandings of Asian Americans as model minority or ardent supporters of the Democratic party do not fully encompass these differences. Whereas second-and third-generation, young, highly educated Asian Americans of East Asian and Indian descent may fit this bill, older, first-generation immigrants of Southeast Asian origin hold very different opinions on many issues. If we hypothesize that the latter population may feel less comfortable expressing their opinions through surveys—as the high rate of don’t know in my analysis suggests—the divide between the two groups may be more substantial than we infer from available data sources. The results presented here suggest that Muslim immigration may serve as a catalyst in exposing such divide.
To be clear, I do not dispute that a strong convergence toward progressive opinions on policy exists among Asian Americans, as shown in table 1. As Janelle Wong and Sono Shah (2021, this issue) point out, an overwhelming majority of Asian Americans support active government interventions in domains such as health care and education, and many support more humane and inclusive immigration policy as well as actions for environmental and racial justice. This finding demonstrates that as of now the image of Asian Americans as devout, unified supporters of the Democratic party holds true. My analysis, on the other hand, looks toward the future—the future in which more progressive and less progressive factions of Asian Americans may take their differences seriously, and, consequently, begin to doubt whether they really belong to a unified category. The possibility seems far-fetched, but the recent leftward drift of political consciousness, especially among nonwhites, may accelerate this process of remaking and unmaking of the panethnic category. The stigma of kung flu may be a contributing factor as well.
Recently, most visible political competition occurred not in swing states between the two major parties, but in Democratic strongholds such as New York City between liberals and radicals. Considering that a majority of Asian Americans live in coastal Democratic stronghold states, it is not entirely implausible that this contest between liberals and radicals will play out within Asian American communities, if it had not begun already. My analysis preempts possible topics of debate that may emerge in those contests. In any case, my analysis does not take the Asian American category as granted and explores the potential fault lines beneath its smooth surface. Regardless of whether one stands for or against panethnic mobilization around the Asian American identity, the underlying geography of the panethnic category will be useful in thinking about the future of Asian American politics.
This analysis was severely limited by the available NAAS questions on policy opinions, which only had seven items. If we regard the antipathy toward Muslim immigration as an expression of fear against the generalized Other, we can expect the same patterns of opinion will emerge in other policy areas, especially in ones concerning minorities, such as LGBTQ rights and affirmative action. In addition, the survey items analyzed in this study only concerned domestic issues, and we do not have much information on the opinions of Asian Americans regarding international issues. In the past century, Asian Americans were generally regarded as voters who cared deeply about international issues, such as the Vietnam War, the China-Taiwan relationship, and the India-Pakistan conflict, to mention a few (Wong et al. 2011). In light of the sea change in U.S. foreign policy under the Trump administration, what is the state of opinion among Asian Americans on international issues? How are they reacting to the trade war with China? How about the protest in Hong Kong and the talks with North Korea? What do they think about the long wars in Iraq and Afghanistan? Future studies can examine how these questions map onto the fault lines described in these results.
Last, the NAAS pre-election survey was administered in 2016, and the analysis featured in this article was mostly conducted during the fall of 2019. As we are all painfully aware, the COVID-19 pandemic brought significant changes in how Asian Americans are perceived [End Page 64] in American society. To say their presence was questioned would be an understatement: in the manner reminiscent of the Chinese exclusion era, Asian Americans were stigmatized as a “disease” plaguing the body politic, both literally and figuratively. They were exposed to physical assault and political demonization, and ensuing immigration restrictions separated their families and jeopardized their careers. This article does not speak to this turn of events, at least directly. It is certainly possible that, after experiencing these attacks, Asian Americans would become even more progressive, especially on issues relating to immigration. Conversely, it is also possible that non-Chinese groups would attempt to distance themselves from the Chinese, further leading to dissolution of a coherent panethnic identity and political mobilization based on it. Future data collection efforts should pay attention to these possibilities.
CONCLUSION: NEW POLITICS FOR ASIAN AMERICANS?
Just fifty years ago, Asian American was a term uttered only on college campuses by activists, most likely San Francisco State University and University of California, Berkeley. In San Francisco’s Chinatown and Japantown, only a few miles away from these campuses, many immigrants and their children went on about their businesses, seeing themselves as Chinese and Japanese, respectively, and not much as American, let alone Asian American. The social movement based on shared concerns around the Vietnam War and domestic racism changed the situation. A new group emerged with a new identity and voice, and they were recognized by the government and became a part of common sense knowledge for the generation who came afterwards. Although active mobilization based on the identity has dwindled somewhat relative to its heyday, we live in the age where Asian American is a clearly defined, commonsensical term.
At the same time, however, the category is broad and masks internal divisions. As noted earlier, twenty-four ethnic groups are included in the Asian American category, each with its own distinctive culture and immigration history. Today, it is not clear who and how many among these newcomers identify strongly as Asian American; most likely they understand what Asian American means and would not deny that they belong to the category, but not too many of them would imbue it with cultural meanings, let alone political implications. An analysis of the 2016 post-election NAAS data revealed that different ethnic groups have different opinions on what Asian Americans share as a panethnic group. For instance, whereas 79 percent of Hmong respondents answered that Asian Americans have a common economic interest, slightly more than 50 percent of Korean and Japanese respondents had the same opinion. The answers also diverge in terms of having a common culture and political interest; and, most interestingly, opinion differs on whether Asian Americans are of a common race (Ramakrishnan et al. 2016).
In light of these findings, this analysis demonstrates two interesting patterns: first, convergence of opinion is remarkable on issues such as health care, education, climate change, and racial justice, a majority of Asian Americans supporting active government intervention on these issues; second, the divergence around the issue of Muslim immigration, which maps onto differences in immigration experience and ethnicity, is less notable but still important. Whereas popular imagination, such as embodied in Crazy Rich Asians, presents the caricature of Asian Americans as young, educated, second or third generation of East Asian origin, this analysis shows that other caricatures are equally possible, most notably of old, less educated, first-generation immigrants of Southeast Asian origin. Although these two groups have much in common in terms of policy opinions, they diverge on some issues, including Muslim immigration, undocumented immigration (Tran and Warikoo 2021, this issue), and gender politics (Yang and Charles 2021, this issue). If a new politics around Asian American identity arises, it will focus on these social divisions, which will function as a catalyst igniting different forms of mobilizations based on different identities.
Sunmin Kim is assistant professor of sociology at Dartmouth College, United States.
Esther Cho, Lucas Drouhot, Filiz Garip, Mary Grover, Josh Kaiser, Yang Lor, Kimberly Rogers, Tony Sirianni, and Emily Walton provided much support throughout this project. Jennifer Lee and Karthick Ramakrishnan as well as the participants of the RSF conference on this issue helped me pin down my ideas. I am also grateful to the three anonymous reviewers of RSF for their comments and suggestions.
1. The NAAS administrators note that the weight was constructed by comparing the survey population with the national Asian American population in the following characteristics: ethnicity, gender, age, state of residence, education, and nativity. For further specifications, including sampling procedure and margins of error for subgroups, see appendix A (Ramakrishnan et al. 2018, 41–42).
2. The numbers in parentheses do not exactly add up with class size because the probabilities for class allocation were rounded up to the second digits.
3. I thank an anonymous reviewer for bringing up this point.
Open Access Policy: RSF: The Russell Sage Foundation Journal of the Social Sciences is an open access journal. This article is published under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.