Asian American Candidate Preferences: Evidence from California

The diversity of the Asian American population presents challenges for theories of bloc voting, partisan voting, and descriptive representation. What cues (if any) do Asian American voters rely on? How informative are racial and partisan cues to Asian American voters. This article looks at the candidate preferences of Asian American voters in the 2018 election. I look at elections where an Asian American candidate was on the ballot and compare outcomes within district to the gubernatorial race (a race with no minorities on the ballot). I use surname-coded voter registration records and precinct-level vote returns to estimate Asian American candidate preferences as a racial group and by national-origin. I find strong evidence of national-origin preferences (i.e. Vietnamese for a Vietnamese candidate) among Asian American voters. In instances where the national-origin of the candidate and the national-origin of the voter did not align, voters seem to rely on partisan cues. National-origin preferences are sufficient enough that in one instance voters switched parties within the same election to vote for a candidate of the same national-origin. These findings have implications for theories of minority vote choice and challenges the existing literature on the strength of partisan cues (replication data can be found at: https://sites.google.com/view/vivienleung/research).


Introduction
Theories of bloc voting often start from the assumption that voters prefer in-group candidates and that voters will make inferences about a candidate using descriptive traits such as race (McDermott 1998) and gender (Koch 2002). The symbolic importance of visually seeing in-group candidates achieve political success also has downstream effects of increasing voter turnout and trust in government (Tate 2001(Tate , 2004Stokes-Brown 2016). The emphasis on race however, is complicated when applied to Asian Americans due to the rich diversity and heterogeneity of this population.
For Asian Americans, their racial group may not be as salient as their nationalorigin group since the racial moniker is an imposed category. Asian Americans have very disparate histories in the United States and do not share a singular unifying language, custom, religion, or phenotype. We might assume, for example, that a Korean American would strongly support a Korean American candidate, 1 but it is not clear if non-Korean Asian Americans will automatically assume that the Korean candidate will best represent their interests. It is unclear if an Asian American candidate provides a shared group cue to Asian American voters of all national-origins or only to those of the same national-origin group.
This study aims to assess two things. The first is whether Asian Americans have preferences for candidates of the same national-origin group. For example, do Chinese American voters prefer Chinese American candidates? Secondly, this study also aims to assess whether Asian American candidates have intra-racial appeal. Do Asian American candidates appeal to Asian American voters across national-origin lines?
Asian American vote preferences present a tougher 'test' of theories of racial bloc voting and descriptive representation because of the added layer of nationalorigin differences. Following from the descriptive representation literature, a voter can potentially surmise that a candidate of the same national-origin can best descriptively represent them due to similarities in customs, language, and religion. The same cannot be assumed for candidates that are a different national-origin from the voter. For example, an Indian American candidate has very different customs, religion, language, and phenotype from, say a hypothetical Cambodian American voter. The only thing that the candidate and voter share in this scenario is the same racial designation as Asian. The Cambodian American voter in this scenario may have nothing descriptively in common with the Indian American candidate, as they both speak different languages, ascribe to different religious customs, different cultural customs, and also look very differently from one another. At surface level, there are no cross-cutting traits that Asian Americans share aside from being labeled as "Asian American" and treated as the same in American society.
If Asian Americans do have racial candidate preferences that transcend nationalorigin boundaries, this would demonstrate that Asian American racial ties among voters are much stronger than previously established in the literature. On a campaign and mobilization level, this could also mean that Asian Americans can be more easily engaged by a descriptively representative candidate. Although work has been done on how descriptive candidates mobilize turnout, I look instead at an estimate for vote choice. I find that in general, the greatest support for an Asian American candidate tends to be from his or her national-origin group. The evidence for intraracial appeal among Asian American candidates is mixed.
In this study I look at Asian American candidate preferences in the 2018 midterm election in California where there is a large enough Asian American population to disaggregate results by national-origin. 2 I compare these results to those of the gubernatorial election, an election between two white male candidates, in order to establish whether or not there is a greater preference for an Asian candidate relative to a non-Asian candidate. The gubernatorial race estimates allow me to estimate Asian American baseline partisan preferences in election matchups where there is no racial or national-origin cue. I can then compare estimates between the gubernatorial race and the Asian candidate race. 3 This paper is organized as follows. First, I provide a review of partisan and nonpartisan vote cues for minority voters. Then, drawing from the literature, I generate a theoretical framework for how Asian Americans may disrupt conventional theories about minority voters, and discuss my hypotheses. This is then followed by an overview of my data, methods, and findings. Afterwards, I offer a discussion of my findings. In the final section, I look at the implications for these findings, and provide suggestions for future research.

Literature
Studies of racial and ethnic voting tend to focus primarily on African Americans and Latinos (Bullock 1984;Stokes-Brown 2016;Soto 2007;Barreto et al. 2009). The presence of racial and ethnic minority political representatives has been found to have a variety of important effects such as efficaciousness (Merolla et al. 2013), higher levels of political participation (Abramson and Aldrich 1982), increased contact with legislators and better perceptions of government (Bobo and Gilliam 1990). Legislators of color have also been found to be more cognizant of issues that affect 2 Due to space limitations there are some State Senate races and a Congressional race that I do not explicitly discuss. The analysis and breakdown for those races can be found in the Online Appendix. A summary table of the other State Senate races can be found in Table 9. 3 I also provide an additional estimate of baseline party preference using 2018 senatorial elections in districts with high Asian populations that do not have an Asian candidate on the ballot. I do not refer to the other senatorial vote in the main body of the text, but the estimates are provided as an additional robustness check for baseline party preference. See Online Appendix Table 10 and Figs. 10-14. Other State Senatorial race estimates can be found in Table 9 and Figs. 7, 8 and 9.
1 3 communities of color (Cobb and Jenkins 2016) and this descriptive representation also comes with substantive benefits (Grose 2005).
Among African Americans and Latinos, there is a clear preference for co-ethnic candidates (Barreto 2007(Barreto , 2010; Barreto et al. 2005;Tate 1994). Latino and African American voters routinely use a candidate's racial and ethnic cues to help inform them of their vote choices. Although the race of a candidate is important, in situations when party and racial cues conflict voters will use party cues supersede ethnic or racial signals (Michelson 2005).
Studies of in-group voting also tend to focus on whether or not minority candidates can mobilize minority voters. Fraga (2016) finds that co-ethnic candidates do not increase minority turnout, rather as the proportion of co-ethnics in the district increases, minority turnout increases. Using the 2012-2016 California State Senate races, Sadhwani finds that Asian Americans who live in places that are 15-30% have higher voter turnout. Turnout in these districts is also increased an Asian candidate is on the ballot (2019). Although turnout is increased when an Asian American is on the ballot, it is unknown whether or not Asian voters also have a preference for candidates of the same race or candidates of the same national-origin. There is limited research on how Asian Americans use racial and national-origin cues (if any) in their vote choice. Tam notes that assuming a 'pan-Asian' hypothesis (i.e. a unified Asian preference for policy and candidates), is problematic (1995). In an analysis of candidate support for Mike Honda and Ro Khanna, Sadhwani finds that Honda, a Japanese American, received more support from Vietnamese. His opponent, an Indian American named Ro Khanna, received more support for Chinese and Indian voters in that district (2017). Asian Americans should not be viewed as a monolithic community, and there needs to be a greater examination of the cues that Asian American voters use.
Asian Americans have weaker ties to the two major political parties and have historically opted out of identifying with the major parties, preferring instead to be label themselves as 'independent' or 'moderate' (Wong et al. 2011). The overwhelming majority of Asian Americans are foreign born (~ 65%) and do not arrive in the United States with entrenched partisan preferences. It is unknown if Asian Americans have clear partisan preferences, or if Asian Americans are true moderates or Independents. 4 The same holds true for second-generation Asian Americans, who then do not experience the same parent-child partisan socialization that African Americans and white children experience (Cho et al. 2006;Hughes et al. 2006;Raychaudhuri 2018). The proclivity towards identifying as moderate also could be a signal that Asian Americans may not respond as strongly to party cues at the ballot box. Party cues are prefaced on strong party attachment and identification-the high instances of Asian Americans identifying themselves as 'moderate' or 'Independent' may indicate weaker party preferences compared to African Americans and whites. Hajnal and Lee argue that Asian Americans may be "true Independents" with no significant attachment to either political parties. Parties also have very limited contact with Asian Americans and may be missing out on opportunities to socialize and engage immigrant populations such as Asian Americans and Latinos (Wong 2006). Analyses of large-N surveys such as the 2016 Collaborative Multi-racial Survey also find that Asian Americans support progressive causes that are typically associated with the Democratic platform such as the Affordable Care Act. 5 Having no explicit partisan attachment could mean that Asian American voters may be more reliant on racial or national-origin cues. Due to a variety of cultural differences, there is also no prototypical 'Asian' candidate that can speak to the entire group. Unlike Latinos, Asian American last names vary wildly by national-origin-seeing the last name "Wang" may send a signal to Chinese and Taiwanese voters, but this cue may not be as useful for non-Chinese and non-Taiwanese voters. Voters also believe that samerace candidates will have the same policy preferences or beliefs that reflect the racial group (McDermott 1998;Stokes-Brown 2016).
While the panethnic label of 'Asian American' is important for exerting political power as a larger coalition (Le Espiritu 1992), studies of Asian American coalitions and identity has found that as whole, Asian Americans are loosely attached to one another and that attachment is situational (Okamoto 2006;Park 2008). Fewer than a third of Asian Americans identify with the panethnic label of 'Asian American' instead preferring the national-origin label to describe oneself (Wong et al. 2011). Studies of Asian American candidate preference are often limited by a lack of largescale data, lack of candidates, or are only focused on one national-origin group. Collet (2005) and Tam (1995) find some support for panethnic voting in scenarios where there is an Asian candidate versus a non-Asian candidate, but due to data and case limitations, are unable to make inferences about larger Asian American voter preferences as a whole, and the extent to which race matters relative to national-origin.
Although there is variation by national-origin, Asian Americans have voted for the Democratic presidential candidate in the last few election cycles (Masuoka et al. 2018). U.S. born Asian Americans tend to align more closely with the group identity, citing common experiences as Asians in the U.S. (Park 2008). Given the long road from immigration to naturalization to the ballot box (Masuoka et al. 2018), second-generation Asian Americans are more likely to be registered and to vote. U.S. born Asian Americans are also born and raised in a context where the label "Asian American" is salient and shared experiences of discrimination and marginalization can increase racial turnout (Wong et al. 2011;Okamoto and Ebert 2010). Furthermore, extended length of time in the U.S., also leads individuals to be more accepting of the pan-ethnic/racial identifier (Lee 2019). Therefore, pan-ethnicity might still matter for Asian American candidates and serve as a cue for Asian American voters. 6 Finally, studies of descriptive representation have also looked at other descriptive markers such as gender. Although women who run tend to be elected just as often as men, the prevailing scholarship has not found that merely being descriptively female is enough to garner a "female vote". This is most clear in the 2016 election results, with roughly 52% of white women voting for Trump over Clinton (Junn 2017;Tien 2017). Women candidates can get a gender 'boost' from women of the same party, especially if a female Democrat is running against a male Republican (Brians 2005). Although gender does matter to a certain extent, voters first make evaluations of candidates on the basis of their party affiliation and incumbent status (Dolan 2004;Sanbonmatsu and Dolan 2009;Stokes-Brown and Dolan 2010). 7

Theory and Hypotheses
Overall, existing scholarship finds that although voters prefer candidates of the same race/national-origin, these descriptive preferences such as gender or race are secondary to party preferences. Asian Americans however, may not respond to partisan cues as strongly due to their status as a predominantly first-generation immigrant community and lack of explicit party preferences. This could indicate that Asian American voters might less responsive to partisan cues and more responsive to race and national-origin cues. I investigate more deeply the added layer that national-origin attachment plays because for immigrant populations, the 'racial' identifier may not be as salient. Asian American racial identity attachment has been found to vary according to context and circumstances (Junn and Masuoka 2008;Masuoka 2008). Asian American racial identity attachment also may not be as strong as nationalorigin attachment (Lien et al. 2001). Therefore, I theorize that the national-origin cue should be the strongest for Asian American voters. Although there is a smaller contingent of U.S. born Asian American voters who may have a different relationship with the racial identity, in general Asian Americans should prefer candidates of the same national-origin. The ideal scenario then, for an average Asian American voter is one where the national-origin and partisan identification of the candidate are congruent, and do not present conflicting cues. In the event that the national-origin and party of the candidate are incongruent, race may actually be the more salient cue for an Asian American voter because it offers more relevant information compared to party. H1: Asian Americans have a preference for candidates who share the same nationalorigin background.
In the event that an Asian American candidate of the same national-origin group is present, Asian American voters should then exhibit a racial preference for other Asian American candidates as the next 'closest' possibility. This leads me to my second hypothesis.
H2: Asian Americans have a general preference for candidates of the same race, but it is not as strong as candidates of the same national-origin background. For example, a Vietnamese voter should prefer other Asian candidates relative to non-Asians.
A preference for a same national-origin candidate should not supersede a preference for Democratic candidates. The extant literature in American politics finds that partisan identity is one of the most salient identities that an individual possesses. This explains why Black voters do not suddenly change parties when a Black Republican is on the ballot because the party cue is more salient and important than the racial one. Despite Asian American non-partisanship, evidence across elections show a general preference for Democrats and policies that Democrats support. Thus, we can expect that the partisan cue should be more effective than a racial one in situations where the racial/national-origin cue and partisan cue are conflicting.

H3:
In the event that there is an Asian candidate that does not align with the party preference (i.e. an Indian American Republican), the party cue should take precedence over race.

Data and Methods
The main method of analysis for this project is using ecological inference (Barreto et al. 2019;King 1997) to estimate national-origin voting among Asian Americans. Ecological inference is a method of inferring discrete individual behavior from aggregate group-level (ecological) data. This inference method is based on a combination of King's method of ecological inference 8 and Goodman's inference (1959). The standard model when calculating vote choice is as follows: we can observe the total number of votes for Candidate A and Candidate B, the proportion of Black and white voters in the precinct, and the percentage of votes that went to each candidate We do not observe the percentage of Black voters who chose Candidate A or B and the percentage of white voters who voted for Candidate A or B. The information is then fed through an iterative Bayesian regression model. It has been used to estimate African American voter preferences (Gay 2001) and has also been used in legal cases when redrawing districts to maximize minority districts (Barreto et al. 2019) and in research by the American Constitution Society for Law and Policy for voting rights research on minority voting patterns. 9 I use this method to estimate the percentage of votes that went to each candidate by national-origin group using the R package 'eicompare' (Collingwood et al. 2016). 10 Voter ballots are also confidential-although voting records in California are publicly available, it only tells us whether or not someone voted and not who they voted for.
I use California as the selected case study because California is home to the largest Asian American population in the United States. Asian Americans account for about 14% of California's population. There is also a lot of heterogeneity within the California Asian American community with Chinese, Korean, Japanese, Filipino, Vietnamese, and Indian making up the largest Asian American populations in the state. Asian Americans in this state live in both national-origin specific enclaves such as Koreatown and less diverse places such as Central California. In particular, the districts I draw from are a mix of both large diverse Asian dense districts such as District 7 in Alameda and Santa Clara County, to District 27 which is a more homogeneous district comprised of mostly Chinese Americans. 11 I focus on Congressional and State Senate races in this study due to their increased visibility on the ballot as opposed to local municipal elections. The 2018 election was also a nonpresidential year which tends to decrease municipal turnout (Anzia 2013;Hajnal and Lewis 2003).
Cases were selected based on whether or not an Asian American candidate ran in that Congressional district in the 2018 general election with an Asian American population above 15%. In 2018 this includes Judy Chu, Young Kim, Ro Khanna, and Cristina Osmeña. Although Janet Nguyen (R-34) is a State Senator, her race was selected due to the demographics of her district (largely Vietnamese) and the fact that she was a long-term incumbent who was unseated by a Democratic challenger, Tom Umberg. 12 The race was also hotly contested and garnered significant media attention. Although there are other state senate races in locations where there are more than 15% Asian Americans, this particular State Senate race offers insight into how Vietnamese Americans, a more conservative national-origin group, behave in the aggregate. 13 The five cases total include three Democrats and two Republicans (Tables 1 and 2). Political Behavior (2022)  Although I would have more cases to work with if I use prior elections, I opt to only use 2018 in this study in order to control for electoral context. I use Congressional and State Senate races as opposed to other elections 14 because these races have higher turnout rates compared to municipal elections, and thus are more comparable to the gubernatorial 'baseline' election. 15 I also use the 2018 California elections as the locale of choice due to open primary system that California has enacted following the passage of Prop 14 16 which was put into place 2012.
The data for this project comes from the California Statewide Database. The Statewide Database collects the Statements of Vote and Statements of Registration for each precinct in California for all elections (general, primary, and special). The Asian American national-origin demographic data comes from two sources. The first is the American Community Survey 5-year estimates which estimates the Citizen Voting Age Population (CVAP) by ethnicity and citizenship per Census block. This includes estimates for non-Hispanic whites, Asians, African Americans, Hispanics or Latinos, among others.
The disaggregated demographic data by national-origin comes from the California Statement of Registration. This statement is precinct level and contains information about the number of individuals registered by party and by nationalorigin. This information is then broken down by national-origin using surname matching into the following categories-Chinese, Vietnamese, Filipino, Japanese, Indian, and Korean-and country of birth matching using Social Security Administration records (Lauderdale and Kestenbaum 2000). The surname matched voter registration rolls do not differentiate between white and black voters, only Jewish, Latino, and Asian subgroups. A potential detriment to this is if a person with an Asian, Latino, or Jewish surname changes their surname due to marriage or other reasons (i.e. Chen to Smith), they will no longer be marked by that national-origin. Asian American last names which do not fit into the above-mentioned subgroups are put into the "other" category. For example, an individual with a Hmong or Cambodian last name will not be counted in the Asian categories. This means that this surname-matching procedure is a conservative estimate of how Asian American voters there may be. Despite these shortcomings however, this remains one of the few ways to get accurate matched voter registration files broken down by Asian nationalorigin as the voter registration card does not ask this information. 17 In order to establish if in-group voting and/or national-origin voting has occurred, I compare the estimates from each Congressional district race to the estimated vote share for California governor. The Democratic gubernatorial candidate in this election is Gavin Newsom and the Republican gubernatorial candidate is John Cox. The gubernatorial election is an open race following the term limitation of former governor Jerry Brown. 18,19 Analysis I divide my analysis by electoral race and compare the national-origin preference in each race relative to the gubernatorial race. Within each race I look at how the candidate fared among his or her respective national-origin group, and how he or she fared among the other national-origin groups. First, I look at Indian American support for Ro Khanna in CA-7, the largest Asian American district outside of Hawaii, and home to a diverse Asian American community. In my second case, I look at Vietnamese American support for Janet Nguyen in State Senate District 34, home to the largest Vietnamese enclave in the United States. In my third case, I look at Filipino American support for Cristina Osmeña in District 14. Then, I look at Chinese American support for Judy Chu in CA-27, a predominantly Chinese American district. The final case in this analysis looks at Korean American support for Young Kim in CA-39, a district with a rapidly growing Asian and Latino population. Party affiliation by national-origin can also be seen in Fig. 1. The cases are ordered as follows because in case 1, the race of the Democratic candidate varies from the baseline gubernatorial race. In case 2, the race of the republican candidate varies from the baseline. In case 3, the race of the Republican candidate varies from the baseline race. In case 4, both party and race of the candidate vary from the baseline. Finally, in case 5, the race and partisan affiliation of both the candidates varies from the baseline election. The cases are presented as such to offer a more logical flow of how cases vary away from the baseline gubernatorial race (least varied on party identification/race to more varied in party identification/race).

Case 1: Indian American Support for Ro Khanna (CA-17)
Ro Khanna, an Indian American Democrat, has represented the district since 2016, when he beat long-term incumbent Mike Honda, a Japanese American Democrat, for the seat. 20 District 17 is located in the San Francisco Bay Area and is home to major Silicon Valley tech firms such as Apple and eBay. It includes parts of Alameda County and Santa Clara County. The district is the only majority Asian American district in the United States outside of Hawaii; according to Census estimates, Asian Americans comprise of roughly 48.9% of the district. It is heavily Democratic and is a reliably blue district. This district is home to a substantial Indian and Chinese population. The district is also relatively affluent, but with an estimated median income of about $120,000 (Fig. 2).
In 2018, all Asian Americans in the district overwhelmingly preferred Khanna to his opponent, Republican Ron Cohen. Khanna won by a landslide, securing 62.0% of the overall vote compared to Cohen at 22.9%. Indian American voters strongly preferred Khanna, with roughly 85.9% of Indian American voters supporting him, second highest of all Asian national-origin group support. Most Indian Americans in the district are independent (45%) or Democrat (47%) and Khanna seems to have secured both independent Indian Americans and Democrat Indian Americans in this election. What is surprising is that the estimated Indian American support for Gavin Newsom is substantially higher (97.41%) than their support for Ro Khanna (85.9%). Even taking into account the margins of error, Newsom enjoyed greater support among Indian Americans than Khanna did (~ 12%). Based on this analysis, Khanna did not receive a significant national-origin support bump from Indian American voters in his district.
In contrast, Khanna enjoyed a slight support bump from Chinese voters in the district. An estimated 75.64% of Chinese voters cast a ballot in support of Khanna Table 11 and Fig. 15 in the Online Appendix. Turnout in this district was also lower during this year (211k votes) compared to 2016 (233k votes). Given that Indian American voters comprise of about 13.4% of voters in that district, what appears to be a decline in support for Ro Khanna by Indian Americans may be attributed to lower turnout and the fact that he ran against a Republican in 2018 (D-27 is an extremely safe Democratic district). There were also slightly more votes cast in the gubernatorial (214k) compared to the senatorial (211k) which may also cause Indian support to appear lower for Khanna relative to Newsom.  22 Latinos comprise of about 47% of the district, followed by 29% white, and 21% Asian American. According to the 2011 American Community Survey, the cities of Westminster and Garden Grove are home to the largest concentrations of Vietnamese Americans in the United States, at 37.1% and 31% respectively. District 34 is comprised of parts of Los Angeles County and Orange County and includes long standing major Vietnamese enclaves such as Westminster, Little Saigon, and Santa Ana. Vietnamese have traditionally been the more conservative Asian American national-origin group, having come to the U.S. during the Vietnam war as refugees under then-President Ronald Reagan. Although recent estimates of Vietnamese Americans estimate a shift away from the Republican party, Vietnamese Americans still remain strong supporters of the GOP. Janet Nguyen a Republican who has represented the district since 2014, was defeated by challenger Tom Umberg, a white Democrat in 2018 (Fig. 3). Ecological inference estimates that about 86% of Vietnamese voted for Nguyen, the Republican state senate candidate. The Vietnamese vote for Cox, the Republican gubernatorial candidate, was not as strong as it was for Nguyen, at 58.7% for Cox to 86% for Nguyen. Taking into account the standard error, we can see that the Vietnamese vote preference was much stronger than what their regular partisan vote for governor would predict. Only about 34% of Vietnamese Americans in the district are registered Republicans, suggesting that the Independent Vietnamese Americans in this district lean more conservative. In contrast, other groups such as Filipinos and Indians voted for the Democratic State Senate candidate Umberg at similar rates to their vote for Gavin Newsom, indicating that Janet Nguyen did not have crossparty or cross-racial appeal for other Asian Americans. This race is a particularly salient example of both national-origin and party congruence working in favor to give Nguyen a 'bump' among Vietnamese voters.

Case 3: Filipina/o Support for Cristina Osmeña (CA-14)
California's 14th Congressional district is located in Northern California and comprises of San Mateo county and a small portion of San Francisco county. The district is relatively diverse, with about 37% whites, 32% Asian Americans, 24% Latinos, and 4% Black. CA-14 is considered a safe Democratic district. According to the 2011 American Community Survey, CA-14 is home to one of the largest concentrations of Filipino Americans in the United States, with a substantial number of Filipinos residing in Daly City, where they make up 33.2% of the population of that city. Filipino Americans are traditionally more socially conservative and more likely to affiliate with the Republican Party. 23 The 2012 National Asian American Survey estimates that about 27% of Filipinos are Republicans, 24% are Democratic, and 45% are Independent. 24 Filipino Americans voted Democratic for the first time in 2008 for Barack Obama and 65% of Filipino Americans voted for Hilary Clinton in 2016. Cristina Osmeña (GOP) ran against Jackie Speier (D), the six-term incumbent who has served in the House since 1979. Osmeña has no prior political experience (Fig. 4).
Although CA-14 has a substantial Asian American population, Filipinos only make up about 3.7% of the registered voters in the district. As such, the estimates for the Filipino vote are difficult to estimate. Ecological inference estimates that Osmeña received about 48% of the Filipino vote. Filipino support for Speier hovered at about 53%. The Filipino vote for Gavin Newsom is roughly 57%. John Cox received roughly 39% of the Filipino vote. Due to the smaller number of registered Filipino voters in the district, the standard errors for these estimates are high, I cannot say definitively that Filipinos preferred Osmeña. However, it does appear that Osmeña enjoyed larger share of the Republican vote compared to John Cox. Based on the estimates of this election, it appears that Filipino American voters do not explicitly have a Democratic or Republican preference in this district. However, given that only 10.9% of Filipino American voters in this district are Republican, and Osmeña definitely outperformed and garnered more of the vote share among Filipino Independents.

Case 4: Chinese American Support for Judy Chu (CA-27)
Congresswoman Judy Chu is the incumbent Representative for California's District 27 and is the first Chinese American woman elected to Congress. The 27th district is situated in the San Gabriel Valley and is home to one of the largest Chinese enclaves in the country. It is home to the cities such as San Gabriel and Monterey Park. After the lifting of national-origin quotas in U.S. immigration in 1965, large numbers of Asian Americans began moving to the area and it became known as "the Chinese Beverly Hills" (Saito 1998). The San Gabriel Valley area that Chu represents has garnered a reputation for being a hub for Chinese and Vietnamese food and shopping. White flight during the 1970's pushed whites into the northern part of the district into cities such as Pasadena. In general, Asian Americans make up about 37% of the district, Latinos about 28%, and whites about 30%. Chu has represented the district since 2009 when she was elected in a special run-off election. Since her time in office, Chu has garnered a reputation of being very liberal and has generally supported progressive policies (Fig. 5).
Chinese American voter support for Chu was overwhelming, my analysis finds that her estimated vote share to be about 92.6% among Chinese voters (see Fig. 5). Chu received a greater percentage of the vote than the amount of registered Chinese Democrats in the area (~ 27%). In contrast, the Chinese American vote for Gavin Newsom hovered at about 51.6% (see Fig. 5). This means that there was an overwhelming national-origin preference for Chu that was far greater than what the partisan vote would predict (almost a 40-point increase). Although there are fewer Koreans, Japanese, and Vietnamese in the community (hence the larger margins of error), support for Chu was still higher relative to their vote for Gavin Newsom. However, given the amount of error, it is difficult to say whether that is mostly a party cue or a racial one for the non-Chinese Asian Americans. In the case of Judy Chu, there is a clear preference among Chinese voters for her candidacy, one that surpasses a typical Democrat/Republican match up. Although non-Chinese in that district also supported Chu, the margins of support are not large enough that we can call it a clear racial preference in addition to partisan preference. Chinese Americans have historically trended slightly Democratic, but the presence of a national-origin candidate seems to dramatically increase Chinese American support for the Democratic party. In District 39, the former incumbent Ed Royce, a Republican, opted not to run for reelection. Gil Cisneros (Democrat) and Young Kim (Republican) advanced through the primaries to the general election. Cisneros is a Latino businessman with no prior political experience. Kim, a Korean-American, was Assemblywoman of that district at the time. The 39th district is composed of parts of Los Angeles County, Orange County, and San Bernardino County. In 2018, this district was considered a swing district, having voted for Romney in 2012 and Clinton in 2016. The district is roughly 34% white, 28.5% Asian, 32.6% Latino, and 2.3% Black. The fastest growing demographic groups in this area are Asians (predominantly consisting of Chinese and Koreans) and Latinos. After a close election which included an absentee ballot tally, Gil Cisneros was declared winner, garnering 51.6% of the vote to Young Kim's 48.4% (Fig. 6).
The election between Kim and Cisneros presents an interesting test for the racial bloc voting hypothesis. Young Kim is a first-generation Korean American with somewhat moderate views on immigration and boasts more political experience than Cisneros, having served in the office of the former incumbent Ed Royce as well as having served in the California Assembly. Nationwide, roughly 40% of Korean Americans identify as Democrats, with an additional 26% as independent. 25 Although a quarter of Koreans identify as independents, a 2016 exit poll estimated that 84% of Koreans voted for Hillary Clinton (Wang 2017) and about 78% of Koreans voted for Barack Obama's re-election, 26 indicating a steady lean towards the Democratic party.
Korean Americans are also underrepresented in Congress, had Young Kim won, she would have been the first Korean American woman to serve in Congress. 27 Although Kim lost the election to Cisneros, she received an estimated 59.58% of the Korean vote. Surprisingly, 63% of Koreans also voted for the Democratic gubernatorial candidate Gavin Newsom. Young Kim's estimated vote share among Koreans goes beyond the estimated share of Korean Republicans (who make up approximately 34% of the Korean American voter population in that district). In this election, about a third of Korean American voters crossed party lines in the same election to vote for a co-ethnic candidate of the Republican party while simultaneously supporting a Democratic candidate for governor. This result is startling, especially given the strong Democratic support in previous years for Obama and Clinton.
Ticket splitting, where a voter splits their ballot between two parties, within an election is also purported to be rare, and individuals typically engage in straight ticket voting for one party (Beck et al. 2013). Therefore, the Korean voter 'switch' from Democrat to Republican in the same ballot is rare. This could be an indication of a not-so-strong attachment to the Democratic party such that a more moderate, descriptively representative Republican such as Young Kim could persuade some Koreans to ticket split.
In contrast, Chinese and Latinos in the district voted along the same lines as their gubernatorial vote would predict. Roughly 55.24% of Chinese voters in the district voted for the Democrat Gil Cisneros and 57.05% of Chinese voted for Gavin Newsom (although less than 25% of Chinese American voters in this district are registered Democrats). This indicates that the Chinese vote in this district fell largely within party lines, and Young Kim did not receive any sort of boost among Chinese voters for being Asian American. A racial cue does not seem to supplant a partisan one, hence the similar levels of support among Chinese for both the Democratic candidates. About half of Chinese voters are registered Democrats, and these election results for Kim and Newsom are reflective in the estimated share of registered Chinese Democrats. Relative to the Korean results, Chinese voters did not respond strongly to Kim being Asian American. Since the district is comprised of mostly white, Chinese, Korean and Latinos, estimates for Indians, Japanese, Vietnamese, and Filipinos have much higher standard errors, and therefore are not discussed in this analysis.

Other Congressional Races
Due to the relatively smaller share of Asian Americans in other districts, disaggregation of estimated vote share by national-origin is prone to large margins of error. As such, I present the aggregated estimated Asian vote share for other Asian American candidates relative to Latinos and others. In smaller locales, Asian Americans generally preferred Asian American candidates. Due to the smaller share of Asian Americans in these districts, the margin of error in the aggregate vote is still fairly high. In general, Democratic Asian American candidates enjoyed a relatively high vote share from Asian American voters in these districts. This aggregate vote share for Democratic Asian American candidates is relatively the same as the estimated aggregate vote share for the Democratic gubernatorial candidate Gavin Newsom. Generally, Asian Americans support Asian candidates but there is no indication that Asian American candidates receive greater support among Asian American voters beyond party (Tables 3, 4, 5, 6, 7 and 8).

Discussion
Based on the cases selected, I conclude that Asian Americans have strong nationalorigin preferences. In cases where there is an Asian candidate of a differing nationalorigin, party takes preference (with one exception). This has significant implications as the Asian American population moves beyond traditional destinations such as California, Hawaii, and New York and grows in battleground states such as Virginia and Nevada. I began this project with the hypothesis that Asian Americans in general feel closest to their own national-origin group and this was reflected in all of the cases (Chinese for Judy Chu, Indians for Ro Khanna, Koreans for Young Kim, Vietnamese for Janet Nguyen) 28,29 (Figs. 7,8 and 9) (Table 9). Due to a smaller population size, I cannot make too many claims about Filipino preference for the GOP candidate Cristina Osmeña, although it appears, she received a larger share of the Filipino Republican vote (47.6%) compared to the Republican gubernatorial candidate John Cox (39.1%).
I find no evidence to support my second hypothesis. Asian American voters do not seem to have particularly strong preferences for Asian candidates outside of their national-origin group. In most cases, the estimated vote share for Democratic governor was roughly the same as the estimated vote share for the Democratic representative. This indicates that in cases where the national-origin of the voter and candidate are not the same, the voter defaults to a partisan cue 30 and Asian candidates do not seem to enjoy a 'racial' vote boost. Despite lower numbers of explicit Democratic party identification, I also find that in general most Asian American independents leaned Democrat.
In Cases 2 and 5, a Republican Asian American candidate was on the ballot. Although Asian Americans typically lean Democratic, the presence of Republican Young Kim swayed Korean American voters to split the ticket, voting for both her and Democrat Newsom in the same election. Korean voters have been fairly Democratic in the last few presidential cycles, so this result is especially notable. Based on the evidence presented, Kim out-performed among Korean Americans compared to gubernatorial Republican candidate Cox by roughly 22.8 points. Out of the Korean voters for Newsom, almost one-third of Korean voters crossed party lines to vote for a Republican Korean American candidate. For all other national-origin groups in that district, the partisan cue took effect, and other Asian voters in the district did not change their support from Democrat to Republican. The same is also true in the fourth case. Vietnamese Americans, a Republican leaning group, supported the Republican Janet Nguyen. Other national-origin groups in the area followed the partisan cue (as is reflected in the vote share for Newsom and the Democratic candidate for State Senator, Umberg). 30 Analysis of districts with non-Asian candidates (but with high Asian population) also supports this (see Table 10 and Figs. 10-14). In districts without a co-ethnic candidate, there is less variation. In general, all Asian American national-origin groups voted for Democratic representatives. The only exception in District 47 where Vietnamese preferred the Republican candidate, however this is in line with general Vietnamese voting behavior which leans more conservative. There is little variation in these districts between the vote for governor and the vote for representative. See Figs. 10-14 in the Online Appendix. 28 A supplemental analysis of other State Senate races also finds support for the national-origin support hypothesis. See Table 9 and Figs. 7, 8 and 9. 29 I also find this national-origin support in State Senate races for SS-22 (see Fig. 8) and SS-24 (see Fig. 9), both of which did not contain a partisan cue. In SS-22, Chinese voters had a clear preference for Eng, a Chinese candidate. In SS-24, Korean voters preferred Choi, a Korean American.
Political Behavior (2022Behavior ( ) 44:1759Behavior ( -1788   I find mixed support for H3, in the presence of a conflict between party and race, Asian voters utilize the party cue rather than a racial one. Based on the cases presented, national-origin cues and party cues are more important for Asian voters. 31 I find evidence that Asian American candidates enjoy some intra-racial support so long as there is a shared party, however there does not seem to be an added racial 'bump' or 'benefit' for candidates. In a situation where party and national-origin conflict, Asian voters may be more responsive to the national-origin cue, given that Asian Americans have generally weaker partisan attachments or, as Tajfel and Turner speculate, may be true 'true Independents'. This preference for nationalorigin may be due to stronger in-group identification and weaker attachment to the racial identity of Asian Americans. In the case of Korean-Americans in the Los Angeles and Orange County region (a traditional Republican stronghold), a strong Asian Republican candidate may be very appealing. A standout in these cases is the election of Gil Cisneros vs Young Kim in District 39. I find that Korean Americans preferred the co-ethnic candidate Young Kim even though she is the Republican candidate. In the same election, Korean Americans voted for a Democrat, Gavin Newsom, and a Republican, Young Kim. In this particular case, about 22.8% of Korean Americans did not engage in straight ticket voting for Democratic candidates, and switched partisan preferences mid-ballot to vote for a co-ethnic candidate. In all other cases, the vote for a coethnic did not conflict with the group's partisan preferences. This race is indicative of strong Korean American solidarity post the Los Angeles Riots and the continued strength of Christian Korean churches that reinforce national-origin ties.

Conclusion
My case studies find little evidence that Asian Americans feel particularly strongly about Asian candidates outside of their own national-origin. In general, most Asian American candidates are Democrats, so there tends to be no conflict between race and party. I find that national-origin cues tend to be the most important, as voters will lend greater support to candidates that share the same national-origin, supporting said candidates at rates greater than what is accounted for by party. In all cases in 2018, a candidate enjoyed strong support from other co-ethnics when there was no partisan conflict. However, merely having an Asian American name on the ballot did not predict stronger support among non-coethnic Asian Americans. In these cases, the votes of the other Asian Americans more generally reflected their partisan preferences. This study also demonstrates stronger Asian American preferences for the Democratic party in California, echoing a trend to what other scholars in Asian 31 Although I cannot direct test whether or not this preference for national-origin candidates is due to the presence of national-origin identity, future work may look into this as well as the salience of Asian American panethnicity or situations in elections where panethnicity may become salient.
Political Behavior (2022Behavior ( ) 44:1759Behavior ( -1788 American politics have found (with the exception of Vietnamese Americans). 32 Future work on Asian American strength of partisanship and campaign mobilization could help to explain this case.
In American politics, partisanship is said to be the most important when it comes to vote choice. I find that Asian American candidates do not enjoy greater support from other Asian Americans that are not in the same national-origin group. This means that at the ballot box, a Cambodian candidate can feasibly rely on Cambodian voters to support him or her at a rate greater than what their party preference would predict but cannot reasonably expect non-Cambodian Asians to support him or her beyond what their party preference would predict. It may be the case that Asian American candidates may be better off appealing to other Asian Americans in more traditional partybased platform messaging rather than making pan-racial race-based appeals to garner support from Asian Americans outside of their national-origin group.
As the United States continues to diversify and the electoral landscape continues to change, the question of national-origin and race versus party preferences for Asian Americans will only get stronger. This study finds that Asian Americans follow party and national-origin cues but do not necessarily respond to racial ones. I also find that there is some fluidity with Asian American party attachment in 2018. A limitation of this study is that is this is a study of aggregate behavior and does not delve into individual motivations for party cues versus national-origin ones. A potential avenue for future work could look at how Asian Americans how Asian Americans conceptualize pan-ethnicity and its political ramifications.