When Taiwan Identifiers Embrace the ROC: The Complexity of State Identification in Taiwan*
C. S. LIU
It remains unknown if theoretical perspectives concerning party identification can be usefully applied to the formation or consolidation of state identification. Taiwan presents a unique case for exploring the na- ture of state identification, not only because of the ambiguity concerning its official name, but because its citizens have been struggling with a hard choice between Chinese identification, an identification based on cultural heritage that is considered to be the root of its legitimacy, and an emerging national identification with an independent Taiwan which is beginning to consolidate. Theoretical work is needed to explain the nature and the volatility of state identification in the light of conventional wisdom about party identification. This study concentrates on clarifying the relation- ships between party identification, national/ethnic identification, and state identification. It will show how these concepts are interwoven and how the idea of a "state identification crisis" originated. It will also shed light on studies about conflict process and political psychology.
FRANKC. S. LIU(劉正山) is an associate professor teaching in the Institute of Political Sci- ence and the director of the Survey Center of National Sun Yat-Sen University. H is teac hing and research areas include research methods, comparative politics, public opinion, and po- litical communication. He can be reached at <firstname.lastname@example.org>.
*The author thanks John E. Transue and anonymous reviewers for their useful comments.
The project was partially funded by the N ational Science Council (98-2410-H-110-011- MY3) and the "Aim for the Top University Plan" of National Sun Yat-Sen U niversity.
©Institute of International Relations, N ational Chengchi University, Taipei, Taiwan (ROC).
KEYWORDS: national/ethnic identification; party identification; state iden- tification; Taiwan; dual identification.
* * *
The "nature" of a particular concept refers to its substantial mean- ing or its critical components. Seeking a better understanding of the nature of identification is fundamental and important to understanding politics and conflict. Unfortunately, the nature of identifica- tion in many democracies has not been well explored. Although the nature of party identification has been proved to be an "unmoved mover" of one's democratic orientations and behavior,1we do not know how far we can follow current studies on party identification to understand the nature of na- tional and state identification. Is the nature of state identification in Taiwan an unmoved mover? Or is it as volatile as a political attitude?
The nature of state identification (國家認同) and national identifica- tion (民族認同) has not been fully explored in the literature. I agree with Sapiro when she states that "national identity was not much studied by po- litical socialization scholars who focused on the United States in the early days of the field, probably because the problem was not as pressing a con- cern in the United States at that time . . . the U.S. literature has tended to focus on such matters as the development of partisanship and electoral be- havior rather than on national identity development."2
Taiwan is chosen for this study primarily because the problem of state identification there is not like that of Spain, the American Indian tribes, or Tibet. Like Spanish people who may experience dual identity, in that they see themselves as both Catalan and Spanish, for example, many Taiwanese have a "dual national identity," claiming to be both Chinese and Taiwanese.
The difference is that Spanish dual national identities are associated with regional identity, while Taiwan's dual national identity is associated with
1Richard Johnston, "Party Identification: U nmoved Mover or Sum of Preference s?" Annual Review of Political Science 9 (2006): 329-51.
2Virginia Sapiro, "Not Your Parents' Political Socialization: Introduction for a New Genera- tion," Annual Review of Political Science 7 (2004): 6.
state identification and party identification, as I shall discuss in this paper.
Taiwan represents a case where "nationality becomes integrated into the political orientations of real people," suggesting that ethnicity and political preferences are components of national identification, and where "particu- lar cohorts may be moved to change their conceptions."3 Moreover, unlike American Indians or citizens in the European Union, where sub- and super- national entities demand or depend on a sense of membership, citizenship, or other forms of identification, Taiwanese experience types of identifica- tion that overlap each other, linking the question of "who am I?" to that of
"which country do I want to belong to?" Furthermore, unlike Tibet, where Tibetans seek self-determination as a means to be freed from Beijing's con- trol, Taiwan displays little of the nationalism that would lead Taiwanese to mobilize and establish a new country.
Political scientists most often mention Taiwan because its relation- ship with the People's Republic of China (PRC) is one important aspect of Asian security. One particular issue is whether Taiwan is going to accept the relationship as defined by the PRC (unification in the long run), or whether it will go its own way (claiming independence). This paper ac- knowledges the importance of this line of research but has to bypass this body of literature on security and pay more attention to the nature of state identification.
Studying the nature of state identification conceptually will also con- tribute to a better understanding of the empirical identity "crisis" in terms of ambivalence about making consistent choices across national identifica- tion and state identification. This crisis, as I shall explain below, is usually evidenced by polls that convey contradictory messages about Taiwanese people's unification/independence orientation. For example, in a poll con- ducted by a well-known TV news station and released on June 18, 2008, 45 percent of respondents identified as Taiwanese, 4 percent claimed to be Chinese, and 45 percent claimed to be a mixture of both. When respon- dents were forced to choose between Chinese and Taiwanese, 68 percent
chose Taiwanese. However, when it came to the issue of the relationship with the PRC and the choice between unification and independence, the majority (68 percent) chose the former and 19 percent the latter (in a sepa- rate question, 58 percent said they preferred the status quo, 8 percent opted for unification, and 19 percent said they preferred independence).4
The results of this poll invite contradictory interpretations about Tai- wan's political future. Looking at the first part of the survey, some would interpret it as showing a desire for unification, while others, looking at the second part, would see it as indicating a desire for independence. This example implies that the general public, the news media, and government officials can easily confuse Taiwanese voters' state identification and na- tional/ethnic identification. In effect, there is, as yet, no clear understand- ing of the relationship between the two types of identification. Scholars of Taiwan are usually unaware of the extent to which party identification, na- tional/ethnic identification, and state identification are related to each other.
In particular, scholars of Taiwan's identification issue tend to assume that there is no difference between national and state identification, and/or they simply treat political preferences concerning independence/unification as direct measurements of state identification.5 Although there are some scholars who do separate ethnicity and party identification from national identity and discuss how ethnicity and partisanship influence national identity,6the body of literature on Taiwanese identification continues to use
"national identity" as a synonym for "state identity," and it has not directly dealt with the measurement of state identification.
4The original source is no longer available, but a video clip of this report (in Chinese) is avail- able at http://www.youtube.com/w atch?v=PgP-6-PHgJA &feature=player_embedded.
5For example: T. Y. Wang and I-Chou Liu, "Contending Ide ntities in Taiwan: Implications for Cross-Strait Relations," Asian Survey 44, no. 4 (July-August 2004): 568-90; Yu-Shan Wu, "The Evolution of the K MT's Stance on the One China Principle: National Identity in Flux," in Taiwanese Identity in the 21st Century: Domestic, Regional and Global Perspe c- tives, e d. Gunter Sc hubert and Jens Damm (N ew York: Routledge, 2011), 51-71.
6For e xample: Wen-Chun Che n, "National Identity and Democratic Consolidation in Taiwa n:
A Study of the Problem of D emocratization in a Divided Country," Issues & Studies 33, no.
4 (April 1997): 1-44; Robert Marsh, National Identity and Ethnicity in Taiwan: Som e Trends in the 1990s (Armonk, N .Y.: M. E. Sharpe, 2002).
Moreover, the key concepts identified above have not been put to- gether and studied as a whole. Therefore, we see a series of discussions about the measurement of "national identity" in terms of preferences con- cerning independence from or unification with China7rather than "state identity" in terms of psychological attachment to a state-related symbol.
This paper is devoted to this line of inquiry: understanding the substance of state identification as the basis for future studies on constructing a valid measurement of this concept. It will begin with an overview of why state identification has been a salient and important political issue in Taiwan.
This will be followed by a review of the literature on party identification.
The third section will look at theoretical studies on the nature of identifi- cation. These two sections are used to show the present state of identifi- cation studies and to shed light on the present study on state identification and the design of the survey.
Please note that the rest of this paper will consistently use the term
"identification" rather than "identity" for two reasons. First, the meaning of identification is better fitted to this inquiry. It is drawn from the term
"party identification" that is commonly used in political science literature.
Just as party identification refers to the political party with which an indi- vidual identifies, state identification refers to the national symbols with which that individual indentifies. Identity refers to a symbolic, stabilized status or label used by individuals to describe or to understand themselves and/or others, while identification refers to a more dynamic status which characterizes the process by which a person finds his or her identity. This paper will discuss this process rather than the labels themselves. Second, in Taiwan there is no significant evidence indicating that individuals have a strong and clear "identity"; instead, the majority of people are ambivalent about these labels. Therefore, instead of using the term identity, which im- plies clear-headedness about partisanship and perceptions concerning the nation and state, I shall use identification.
7For example: John F. S. Hsieh and Emerson M. S. Niou, "Measuring Taiwanese Public Opinion on Taiwanese Independence," China Quarterly 181 (March 2005): 158-68.
Identification Crisis in Taiwan
Taiwan is a unique case in that its identification problem does not fit into the conventional categories. In particular, Taiwan' identification has become a problem or even a crisis because it is composed of more than two types of identification that are attached to different symbolic systems.
Joireman identifies five types or dimensions of identification: region, reli- gion, race, language, and custom.8 Taiwanese people acknowledge Chi- nese history, custom, and language, but they are simultaneously aware of Taiwan's distinct region and custom. Hence, Taiwan's identification crisis is attributable to Taiwan having been identified in China's history as a province of China, and the identification problem on this island seems to be associated with Taiwan's willingness, or lack thereof, to seek "inde- pendence."
The case of Taiwan's identification is more complicated than a mere failure to fit the existing categories, and it has grown more and more com- plicated since China lost the Sino-Japanese War in 1895 and was forced to cede Taiwan to Japan. Since then, Taiwanese people have experienced Japanese colonial rule, the governance of the Republic of China (ROC) under Chiang Kai-shek (蔣介石) after 1945, and transition from dictator- ship to democracy. In the past, therefore, "Taiwan" has not only been re- garded as the name of a region, but also as the name of a state, particularly since the 1980s.9 Hence, in my opinion, the identification problem in Tai- wan is more about national identification (which is associated with one's ethnicity) and partisanship than geographic identification.
Note that "identification crisis" is not the same as "identity crisis."
Citizens of this island have little problem choosing a political label; in ef- fect, everyone in a democracy is free to pick any political label or symbol.
Taiwan has no identity crisis in this sense. The issue is one of an emerging
8Sandra F. Joireman, Nationalism and Political Ide ntity (Ne w York: Continuum, 2003).
9See also Yun-han Chu and Jih-wen Lin, "Political Development in 20th-Century Taiwa n:
State-Building, Regime Transformation and the Construction of National Identity," China Quarterly 165 (March 2001): 102-29.
pattern of individuals identifying with multiple symbols that seem to con- tradict each other or at least are inconsistent with their factual meanings.
For example, some people identify with the Democratic Progressive Party (民進黨, DPP) but prefer to use the state title "Republic of China"; or more commonly, some identify themselves as Taiwanese and prefer the Republic of China over Taiwan. Therefore, by saying that Taiwan has an "identifi- cation crisis" I do not mean that this crisis is an immediate threat to national security or societal stability. Instead, it implies that there is a possible or potential problem as a consequence of cross-cutting identifications.
First, the conventional wisdom among most scholars of Taiwan poli- tics is that national identification in Taiwan is primarily a reflection of sub- ethnicity: "Minnanese people (local people whose ancestors came from southeast China) tend to think of themselves as Taiwanese; mainlanders (those whose families have come to Taiwan from the mainland since 1948) tend to think themselves Chinese; and Hakka tend to waffle in between."10 As Joireman points out, every national/ethnic group is supposed to have a proper name, a myth of common ancestry, shared historical memo- ries, a common culture defined by language or religion or customs, a link with a geographic homeland, and a sense of common cause or solidarity among some members of the population.11 By this definition, however, Taiwanese cannot be seen as one nation or ethnic group. As will be ex- plained below, there are two national/ethnic identifications in Taiwan, while a significant number of people stand in between and are ambivalent (or independent) about such identifications (see table 1).12
10Tse-Min Lin, Chin-En Wu, and Feng-Yu Lee, "'Ne ighborhood' Influe nce on the Formation of Nationa l Identity in Taiwan: Spatial Regression w ith Disjoint Neighborhoods," Political Research Quarterly 59, no. 1 (M arch 2006): 35. Lin et al.'s categorization somewhat over- simplifies the situation bec ause there are Taiwanese, not necessarily mainlanders, who claim to be Chinese. Cultural primordialism suggests that people view their own cultural ba ckground as primordial, and thus it is. H ence, I think to a grea t extent people in Taiwan who claim to be Chinese display more cultural primordialism— emphasizing that culture is the critical tie that binds people together— than biological or linguistic primordialism—
emphasizing that the roots of identity lie in biological characteristics passed on by birth or language. See Joire man, Nationalism and Political Identity.
11Joire man, Nationalism and Political Identity.
12Yun-han Chu, "Taiwan's National Identity Politics and the Prospect of Cross-Strait Rela-
The second dimension that characterizes the complexity of Taiwan's national identification is partisanship. Gold argues that one of the reasons why Taiwan people's national identification is an issue has its origins in dis- trust of (or even hostility toward) the Kuomintang (國民黨, KMT), the political party that originated in China and came to rule Taiwan after World War II.13 He thinks that after years of democratization, Taiwanese society has gradually developed a historical and geographical identification that is different from that of the KMT. This further progressed into two competing
tions," Asian Survey 44, no. 4 (July-August 2004): 484-512; Cal Clark, "The Paradox of the Na tional Identity Issue in Chen Shui-bian's 2004 Presidential Campa ign: Base Consti- tuencies vs. the Moderate Middle," Issues & Studies 41, no. 1 (January 2005): 53-86.
13Thomas B. Gold, "The Status Quo is Not Static: Mainland-Taiwan Relations," Asian Sur- vey 27, no. 3 (March 1987): 300-15.
National/ethnic Identification Problems in Taiwan
Ethnic Criteria A proper name?
Myth of common anc estry?
Shared historical memories?
A common culture defined by language?
A link with a geographic homela nd?
More than one: Its official name is Republic of China (ROC), but the commonly used nation- al name is Taiwan, Taiwan (ROC), or the one used in WTO : The Separate Customs Te rri- tory of Taiwan, Penghu, Kinmen and Matsu.
Mixed: Chinese-based, also some Taiwanese stories
Diversified: both Chinese history and Ta iwan- ese history are taught in sc hools
Mixed: Ma ndarin Chinese is the official lan- guage, while "Minnan H ua" (Taiwa nese) is also commonly used in daily life.
Diversified: Constitutionally acknow ledged home land includes ma inland China, but in practic e the homeland refers to the area in- cluding the islands of Taiw an, Penghu, Kinmen, and Matsu.
identification systems: the pan-green camp (泛綠陣營), led by the DPP, the main opposition party in Taiwan, and the pan-blue camp (泛藍陣營) composed of the KMT and other smaller parties. The pan-green see China as the "guest" (we are a different country) and the pan-blue see China as the "host" (we are free Chinese in Taiwan). The alignment between this ideology and the political camps peaked during Chen Shui-bian's DPP ad- ministration.14
In sum, the problem of Taiwan's state identification is both ethnic/
national and political. As national/ethnic identification and state identifi- cation are two sides of the same coin, "with ethnicity being a benign mani- festation of identity and nationalism a politicized shared identity,"1 5 it is difficult for scholars to conceptualize, analyze, and theorize such com- plexity.1 6 Using "national/ethnic identification" and "state identification"
interchangeably has at least one serious consequence: most surveys and polls usually ask respondents to pin down their national/ethnic identifica- tion (Do you think you are Chinese, Taiwanese, or both?), while analysts, journalists, and politicians interpret the results of these polls as a profile of Taiwan people's state identification. A commonly observed mistake in news reporting and political commentary is equating those identifying themselves as Taiwanese with supporters of independence from the ROC or the People's Republic of China (PRC), or likewise equating Chinese identifiers as nationalists who are eager to see the confederation of the ROC and the PRC. One fundamental task in avoiding such misleading in- terpretations is to clarify the nature and meaning of national identification and to examine the associations among national/ethnic identification, party identification, and state identification. For example, if we take a closer
14See also Gunter Schubert, "Taiw an's Political Parties and National Identity: The Rise of an Overa rching Consensus," Asian Surve y 44, no. 4 (June 2004): 534-54; Wu, "The Evolution of the KMT's Stance," 51-71; Gunter Schubert and Stefan Braig, "How to Face a n Embrac- ing China? The DPP's Identity Politics and Cross-Strait Relations during and after the Chen Shui-bian Era," in Schubert and D amm, eds., Taiwanese Identity in the 21st Century, 72-94.
15Joireman, Nationalism and Political Identity, 9.
16See the research of Huddy and K hatib for an outstanding attempt to do this. Leonie Huddy and Nadia K hatib, "American Patriotism, National Identity, and Political Involvement,"
American Journal of Political Science 51, no. 1 (January 2007): 63-77.
look at conflicting identification (Taiwan versus ROC) or dual identifica- tion (Taiwanese and Chinese), we will see that this issue is not completely influenced by the relationship with the PRC, or identification pressure from other countries.
The Nature of Identification:
Perspectives from the Study of Party Identification
American political scientists who study party identification have focused on identification as a psychological mechanism. Since Converse, party identification has been a pillar of research on partisanship, public opinion, and political behavior.17 Some decades ago, American scholars reached a consensus that it is the "unmoved mover" of democratic charac- teristics and political behavior in countries like the United States. How- ever, the development of this research stream has left some questions about the nature of partisan identification unanswered. The focus of this section is a review of how scholars understand, perceive, and theorize the concept of identification. This review will pave the way for a discussion of the con- cept of state identification in the next section.
Studies derived from Converse's perspective agree that an individual's partisanship is his or her emotional attachment to a political party.18 Because one's belief system is composed of one's ideology and values, it is expected to be stable over time. This Michigan School view implies that partisanship is an "unmoved mover," more social-psychological than political.19
As this "unmoved mover" perspective prevailed, scholars of partisan- ship devoted most of their attention to the liability of partisanship. Some
17Philip E. Converse, "The N ature of Belief Systems in Mass Publics," in Ideology and Dis- content, ed. David Ernest Apter (New York: Free Press, 1964), 206-61.
19One important stream of research on partisanship is the measurement of this conc ept. As it is beyond the scope of this paper, intereste d reade rs may like to see Burden and Klofstad's article for a n overview. See Barry C. Burden and Casey A. Klofstad, "A ffect and Cognition in Pa rty Identification," Political Psyc hology 26, no. 6 (D ecember 2005): 869-86.
have focused on the word "mover," while others have focused on the word
"unmoved." Among students of the "mover" aspect, Cowden and McDermott employ natural experiments and confirm the implied causal links between partisanship and political preferences; that is, partisanship is proved to be an unmoved mover of political preferences.20 Goren also uses experimental methods to confirm that partisanship influences political core values, such as principles of equal opportunity, limited government, and moral tolerance, not vice versa.21 After reviewing most of the literature, Johnston concludes that partisan choices hardly ever shape or reshape one's existing partisan identification, no matter how the studies conceptualize stability.22 Even when they take measurement error into account, scholars tend to believe that partisanship, "at least in the United States, and as measured, is a mover but not entirely unmoved."23
Some other scholars focus on the "unmoved" aspect and debate whether party identification is really an unmoved object. Scholars looking at the issue from the perspective of political socialization tend to emphasize the unmoved aspect of partisanship. They see partisanship, ideology, and/
or party labels as being established through parental influence.24 Some scholars take a biological approach and find that partisanship has unmoved features, such as partisan strength and moral standards, that are embedded in one's genes.25 Moreover, political changes, historical memory, and per-
20Jonathan A. Cow den and Rose M . McDermott, "Short-Term Forces and Partisanship,"
Political Behavior 22, no. 3 (September 2000): 197-222.
21Paul Goren, "Party Identification and Core Political Values," Am erican Journal of Political Sc ience 49, no. 4 (October 2005): 881-96.
22Richard Johnston, "Party Identification: Unmoved Mover or Sum of Preferences?" Annual Review of Political Science 9 (2006): 329-51.
24Martin Kroh and Peter Selb, "Inheritance and the D ynamics of Party Identification," Politi- cal Behavior 31, no. 4 (December 2009): 559-74; Michael McDevitt, "The Partisan Child:
Developmental Provocation as a Model of Political Socialization," International Journal of Public Opinion Research 18, no. 1 (Spring 2006): 67-88; and Sapiro, "Not Your Pa rents' Political Socialization," 1-23.
25Ted Brader, and Joshua A. Tucker, "Pathways to Partisanship: Evidence from Russia,"
Post-Soviet Affairs 24, no. 3 (July-September 2008): 263-300; and Jaime E. Settle, Chris- topher T. Daw es, and James H. Fowler, "The Heritability of Partisan Atta chment," Political Research Q uarterly 62, no. 3 (September 2009): 601-13.
sonal political experiences that can arouse emotion all serve to strengthen this process.26
Instead of seeing partisanship as a psychological attachment to a po- litical party, revisionists of the Rochester School see it as a running tally of received political information, subject to situational circumstances.27 This stream of research suggests that partisanship is a not a fixed tie to an object but a flexible one that is subject to external forces. Specifically, partisan- ship is like a summary of the evaluation of past experiences with the parties and something that fluctuates in the short term, influenced by external for- ces such as political events or short-term political forces. Wolak in her study of the development of adolescent partisanship finds that it is both the political personality of adolescents and their wider political environment that determine their partisanship.2 8 It is also found that personal attributes, social networks, and political events like electoral campaigns account for partisanship tendencies.29
This line of thought is echoed by another set of studies on nonparti- sans. The mid-1960s saw a rise in political independence that suggested a rise in macropartisan volatility.30 Previous studies have identified short- term fluctuation in political independence linked to issue preferences, economic conditions, socialization, and periodic political events such as campaigns and elections.31 Moreover, recent studies suggest that the con-
26Dre w Westen, The Political Brain: The Role of Emotion in Deciding the Fate of the Nation (Ne w York: Public A ffairs, 2007).
27Rüdiger Schmitt-Beck, Stefan Weick, and Bernhard Christoph, "Shaky Attachments: Indi- vidual-Level Stability and Change of Pa rtisanship among West German Voters, 1984- 2001," European Journal of Political Rese arch 45, no. 4 (2006): 581-608; Johnston, "Party Identification," 329-51.
28Jennifer Wolak, "Expla ining Cha nge in Party Identific ation in Adolescence," Electoral Studie s 28, no. 4 (December 2009): 573-83.
29Schmitt-Beck, Weick, and Christoph, "Shaky Attachments," 581-608.
30Jack Dennis, "Political Independence in America 1: On Being an Independent Partisan Supporter," British Journal of Political Science 18, no. 1 (January 1988): 77-109.
31H arold D. Clarke and Motoshi Suzuki, "Partisan Dealignment and the D ynamics of Indepen- dence in the American Electorate, 1953-88," British Journal of Political Science 24, no. 1 (January 1994): 57-77; Harold D. Clarke and Marianne C. Stewart, "The Decline of Parties in the Minds of Citizens," Annual Review of Political Science 1 (1998): 357-78; Steven Gre ene, "The Social-Psychological Measurement of Pa rtisanship," Political Behavior 24,
cept of partisanship may not apply well to countries other than the United States, the United Kingdom, or Canada.32 Voters in new democracies may not be so attached to their partisanship.33 A study investigating a long-term panel data set (1984 to 2001) shows that only a small portion of West Ger- man voters are steadfast with regard to partisanship; to these voters, "the longer individuals feel close to a party, the less susceptible they are to ab- andon these attachments."34
Another perspective proposed by Erikson, Mackuen, and Stimson is an "online" model which emphasizes that party identification may be sim- ply a reflection of one's party support.35
Far from a psychological commitment, or even an association derived from past eva luations, this a ttachment reflects the specifics of today's decision. . . . Things will begin anew tomorrow. In politics, this part of partisanship is much close r to a party "supporter" who, w hen saying "I'm a Republican," means that he or she is supporting the Republican candidate today. When the personalities or the issues change, this person's partisan orientation w ill cha nge as well.36
As party identification has been conceptualized and confirmed to be an unmoved mover of political preferences and values, scholars of Ameri- can politics have been paying more attention to tendencies associated with this concept than to its nature. "The micro behavior literature fails to re- solve the issue of the nature of citizen attachment to party, in particular, whether it is fundamentally based upon rational appraisal of outcomes or on 'identification' in its strongest sense, a self-image of commitment to one
no. 3 (September 2002): 171-97; Michael B. MacKuen, Robert S. Erikson, and James A.
Stimson, "M acropartisanship," Am erican Political Scie nce Revie w 83, no. 4 (December 1989): 1125-42; Richard G. N iemi, Stephen C. Craig, and Franco M attei, "Measuring In- ternal Political Efficacy in the 1988 National Election Study," American Political Science Review 85, no. 4 (Decembe r 1991): 1407-13; and D avid O. Sears and N icholas A. Valen- tino, "Politics Matters: Political Events as Catalysts for Preadult Socialization," Am erican Political Science Review 91, no. 1 (March 1997): 45-65.
32See also Johnston, "Party Identification."
33Clarke and Stewart, "The D ecline of Parties."
side."37 Hence, the substance or the core of "unmoved mover" remains unclear in the literature. At least, both the Michigan model (seeing parti- sanship as permanent commitment) and the Rochester model (seeing parti- sanship as standing choice) suggest that if it is true that one's partisanship is developed and strengthened through mechanisms such as political soci- alization or selective processes, there must exist a "core value" to which the layers gained from one's daily experience can be added.
While there is no consensus about the nature of partisan identification and each of the approaches introduced above attracts support, I will further investigate the nature of national identification from the perspective of social identity theory.
The Nature of Identification:
Perspectives from Social Identity Theories
If the core of one's partisanship is not one's "belief system,"3 8as com- monly used in political science, then it can best be expressed as the differ- entiation between groups. As Greene argues, "When we think about what partisanship is, we need to consider not just attitudes toward, but group belonging with, a political party" (emphasis in original).39
Instead of seeing partisanship as a self-growing entity, social identity theorists often see it as group membership. The term "social identity" as employed in social psychology theories refers to the label used by a group of people to differentiate themselves from others regarding personal or group characteristics, such as age, occupation, or gender. Specifically, it is a part of an individual's self-concept that derives from his or her knowledge of his or her membership of a group (or groups) together with the value
37Robert S. Erikson, M ichae l B. Mackuen, and James A. Stimson, The Macro Polity (New York: Cambridge University Press, 2002), 15.
38Converse, "The Nature of Belief Systems in Mass Publics."
39Steven Greene , "Social Identity Theory and Party Identification," Social Science Quarterly 85, no. 1 (March 2004): 136-53.
and emotional significance attached to that membership.40 Social identity theory suggests that (1) people associate themselves with "the social cate- gories, attributes, or components of the self-concept that are shared with others and therefore define individuals as being similar to others"4 1and that (2) individuals attempt to maximize differences between the in-group and the out-group and thus perceive greater differences between the two types of groups than actually exist and show favoritism toward in-group mem- bers, while such in-group favoritism and out-group derogation need not co-occur.42 Greene summarizes that social identity theory explains the formation of individuals' party identification in the United States, Great Britain, Scotland, and Australia because "these social identifications are not based on any formal group membership, but rather self-perceived mem- bership in a particular group" and because "the group nature of partisanship should naturally create a bipolar partisanship where individuals charac- terize the political parties into us and them and exaggerate perceived dif- ferences to favor their own group."44 Therefore, party identification and national identification can be seen as two types of social identification;
social identification is broader in its conceptual scope than party identi- fication and national identification.
Hsu applies this social psychological perspective to explain Taiwan's identification crisis in terms of national identification.45 He argues that the "problem" is based on socialized perception about an external threat (China) and common fate (Taiwan's prosperity). That is, Taiwanese iden- tification has been developed as a national identity in contrast to Chinese identity. "National identity is a constant contrast between 'us' and 'them',
41Kriste n R. Monroe, James Hankin, and Renee B. van Vechten, "The Psychological Foun- da tions of Identity Politics," Annual Review of Political Science 3 (2000): 421.
42Greene, "Social Ide ntity Theory and Party Identification."
45Yung-Ming Hsu, "Making Se nse of Political Learning: The Case of Forming Identities in Taiwa n," Proc eedings of the National Scie nce Council, Re public of China - Part C:
Humanities and Social Sc iences 11, no. 4 (2001): 371-89, http://nr.stpi.org.tw/e journal/
existing over a substantial period to underlie most individual political atti- tudes. In this way, the individual's self-categorization sorts out their inter- est between contesting identities, in this case Chinese versus Taiwanese."46
"The emergence of this 'Taiwanese' versus 'Chinese' division not only redrew boundaries between 'Bensheng' (people from this island) and 'Waisheng' (those from other provinces of China), but also confirmed the rational learning behavior during this nationalizing process."47
Unlike the Michigan School, which emphasizes the stability of iden- tification, it is not assumed in social psychology theories that one's identity should grow more stable over time. "The formation of identities, including 'nationality', is neither reared promptly, inherited naturally, nor granted automatically . . . the formation of new identity politics is an aggregate contingency that young constituencies confront with striking and lasting political realignments."4 8 That is, socialization and other external forces, such as neighborhood influence,49 can increase, but not necessarily de- crease, tendencies concerning identity.
It is worth mentioning here a distinction between "group identity"
and "group consciousness" that supplements this perspective. As Hsu puts it, group identity is fluid and consciousness is contingent.50 "The presence of an out-group symbol is sufficient to arouse consciousness of group salience but is not necessarily an indispensable condition for group iden- tity."51 Hence, "national identity emerges as a particular form of group con- sciousness due to a constant contrast between a new 'us' identity and a foreign 'them'."5 2 This further implies that a concept that is associated with
47Ibid., 386. By the late 1990s, most people in Taiwan felt that conflicts betw een differe nt places of origin were no longer much of a problem. The origin line ha s been replaced by the inconsistency betw een Taiwanese nationalism and Chinese nationalism, such as "Tai- wanese on Taiwan," opposed to "Chinese in mainland China." See Chen, "National Ident- ity and Democratic Consolidation."
49See Lin, Wu, and Lee, "'Ne ighborhood' Influence."
50Hsu, "Making Sense of Political Learning."
this "them vs. us" can be seen as a good fit with social identity theories.
Unlike an action of choosing between names or labels, words or actions that can arouse conscious and emotional differentiation between groups can be seen as a cause of identification formation.
Although social psychology theories provide a more flexible view of how identification forms, they fall short in providing more insights for understanding how it forms outside the United States. These theories ex- plain American voters better because "in America both parties represent broad cross-sections of the public, the only area in which the parties can truly set themselves apart as distinct is ideology."53 Additionally, they fall short in explaining those whose identification is weak and those who live without any identification with a group at all. As Greene states about par- tisans, leaners seem to "identify with both their preferred party and other independents" (emphasis in the original).54
Party, National, and State Identifications in Taiwan
The above review of the nature of identification suggests that party identification, national identification, and state identification in Taiwan are strongly related with each other, if not mutually embedded. Scholars with- out a basic understanding of such differences may have used these concepts interchangeably, incorrectly interpreted the results of opinion polls, and/or overstated their policy implications.
In order to study these three concepts it is necessary to properly oper- ationalize them. Standardized survey questions about party identification are easy to find. But it is a challenge to find proper measurements for the concepts of national identification and state identification separately, since they have been used as one concept and measured by the commonly used survey question: "Do you consider yourself as Taiwanese, Chinese, or both?"
53Greene, "Social Ide ntity Theory and Party Identification," 149.
The problem with this question is the ambiguous way it dangles be- tween state and national identification. If one wants to continue to use this question, which concept should it be linked to? Following the above dis- cussion on social identity theories, in which the main point is that words or actions that can arouse conscious and emotional differentiation between groups can be seen as a cause of identification formation, as this question is very likely to have the effect of emotional differentiation, I judge that it is more associated with national identification than state identification.
Regarding state identification, preferences about unification with or independence from China (or mainland China) are the most commonly used measurement that can be found in survey practice and the literature.55 The problem with this measurement is that the wording of the question al- ways leaves room for respondents to imagine or interpret the meaning of
"our state" (is it the Republic of Taiwan, the Republic of China, or any other), or simply to use Taiwan as the state name. Such a question cannot stimulate a respondent to identify state symbols and choose between them.
Hence, a better alternative for measuring state identity is to bring state symbols back into the survey question, and the simplest way of doing this is to ask respondents to choose which state names they like to identify themselves with.
Respondents may give their answers to the state name question too quickly in a telephone survey, or give ones that seem to them to be politi- cally correct or safe to choose. To avoid such problems, I employ a deliberative-poll style of wording. I ask interviewers to read clearly and slowly an unwanted consequence of each choice (see the next section for the wording), forcing respondents to think directly about the state image with which they identify and then to choose carefully between "Taiwan"
and "Republic of China."
55The typical question wording, ac cording to Taiw an's Election and Democratization Study (TEDS) for the 2008 presidential election survey project (http://www.tedsnet.org), is the following: "Concerning the relationship be tween Taiwan and mainland China, which of the following six positions do you a gree with: 1) immediate unification, 2) immediate indepen- dence, 3) maintain the status quo, move toward unification in the future, 4) maintain the status quo, move towa rd inde pendence in the future, 5) maintain the status quo, decide either unification or independence in the future, 6) maintain the status quo forever."
The following analysis presents the relationships between these con- cepts. The cross tabulate analysis is based on data collected in a nationwide telephone survey conducted by a survey center at a research university in Taiwan. The survey period was December 17-25, 2008, nine months after the 2008 presidential election. One advantage of conducting a survey at that time is that interviewees were no longer mobilized by the presidential election campaigns. The survey used the Computer Assisted Telephone Interview System— CATI. Interviewers called out between 6:30 p.m. and 10:00 p.m. and the interview time was about twenty minutes. The random- ly drawn interviewees were eligible voters aged 20 years or older. The number of the sample was 1,098. Sampling error is 95 percent, 3 percent more or less under level of confidence. Besides the timing of the survey, this data set is unique on account of its inclusion of three items addressing party identification, national identification, and state identification.
The analysis is composed of two parts: descriptive analysis and re- gression analysis. I will first use cross-tabulation and graphs to show the relationships between the three identifications. Next, each of the concepts will be regressed on the other two. This will help explore the causal mech- anism behind the relationships.
The question about national identification is:56
In Taiwan, some people think they are Taiwanese. There are also some people who think that they are Chinese. Do you consider your- self as Taiwanese, Chinese, or both?
The majority (53.2 percent or 584) said they are Taiwanese, 36.2 per- cent said "both," and only 5.2 percent said they are Chinese; 59 respondents said "don't know" or refused to answer. This distribution is consistent with
56This variable is also called "ethnic identity" or "self-identity," see Yun-han Chu, "Naviga- ting between China and the United States: Taiw an's Politics of Identity," in Schubert and Damm, eds., Taiwanese Identity in the 21st Century, 133-54.
all other similar surveys adopting this question. Although the exact figures may differ across surveys, the pattern that the majority claims to be either Taiwanese or both is the same as in other surveys.
The question used for probing party identification is the same as the one adopted in most surveys:
Could you tell me which political party you tend to support? I will read the party names. Please simply tell me the number over the phone. 1. DPP; 2. KMT; 3. Taiwan Solidarity Union (TSU); 4. People First Party (PFP); 5. New Party (NP); 6. Taiwan Independence Party (TIP); 7. Others; 8. Not supporting any party.
Small parties have been decreasing in both number and influence in Taiwan. Individuals' political preference and their political identity has been less about a specific political party than about choosing between the pan-green camp (DPP, TSU, and TIP) and the pan-blue camp (KMT, NP, and PFP).57 To avoid the noise of a small number of respondents identify- ing with small parties and to make more sense of this partisan variable in the following analysis, I recoded this variable to "political camp identifi- cation" where 18.7 percent (205) are blue camp identifiers, and 10.4 per- cent (114) are green camp identifiers. Notably, the majority (68.1 percent) claimed no partisan affiliation, reflecting the fact that most Taiwanese psy- chologically want to avoid being labeled.
The question about state identification is a long one, as it is aimed at obtaining thoughtful answers instead of pop-up answers containing a mix- ture of other concepts:
This question is a bit long and not so easy to answer. Therefore please listen carefully and tell me what you think. As far as nation name is concerned, many people believe that using a new country name will help us get out of the mire of our ambiguous identity in the world and
57See also Schubert, "Taiwan's Politic al Parties and N ational Identity."
therefore get more international support. However, it will also pro- voke military conflicts in the cross-Strait relationship so we run the risk of the destruction of our economy and even the lives of our friends and relatives because of war. On the other hand, some people believe that by keeping the name ROC, we have more room to nego- tiate in cross-Strait relations so that our economy and constitution will endure and peace will prevail. However, in this case, the main- land will keep using its one China foreign policy to prevent our future generations from participating in international affairs with our na- tional name and even force us to abandon our own national identity on international occasions. Therefore, if you could decide Taiwan's future, would you support (1) own nation name for Taiwan (including Penghu, Kinmen, and Matsu)? Or, (2) the current name of ROC?
The interviewers were careful to read the question slowly and clearly.
About one-fifth of respondents lost patience or refused to make a choice.
But the rejection rate was lower than expected. The majority of respon- dents (56.6 percent) chose ROC while 24.4 percent chose Taiwan.
Visualized Cross-Tabulate Analysis
The first step in examining the relationship between the three con- cepts is to contrast these variables with each other using cross-tabulation.58 All of the chi-square tests of independence indicate strong correlations for the three pairs, even taking into account the cases with missing values (all significant at the .001 level).
The following three cross-tables of the three variables are analyzed using the Pearson residuals, which refer to standardized deviations of ob- served from expected values. In the upcoming analysis, the signs of Pear- son residuals are visualized using shadings (blue and red) to indicate the
58The tool for this analysis is "visualizing c ategorical data" (vcd) provided by David Meyer, Achim Zeileis, and Kurt Hornik. For more information, see http://cran.r-proje ct.org/web/
pa ckages/vcd/index.html, http://w ww.math.yorku.ca/SCS/Online/mosaics/about.html, and http://www.math.yorku.ca/SCS/Courses/VCD/vcd-tutorial.pdf.
size of the residuals of an independence model: very colorful for large residuals (> 4), less colorful for medium sized residuals (< 4 and > 2), grey/
white for small residuals (< 2). Each colored residual violates the null hy- potheses of independence. The result of a significance test can be visual- ized by the amount of grey in the colors. If significant, a colorful palette is used, if not, the amount of color is reduced. In other words, the more colorful a cell is, the higher the likelihood of rejection of the null hypothesis of independence. Observations with a deviance residual in excess of two may indicate lack of fit.59
Figure 1 shows the contrast between political camp identification—
the green camp (Green), the blue camp (Blue), no partisan affiliation (NoID), and no answer (N/A, including "don't know," "it depends," and re- fuse to answer)— and national/ethnic identification (Taiwanese, Chinese, and both).
The tiles shaded deep blue correspond to two cells, (Green, Tai- wanese) and (Blue, both), whose residuals are greater than +4, indicating much greater frequency in those cells than would be found if Political Camp Identification and National/Ethnic Identification were independent.
The three tiles shaded deep red, (Green, both), (Blue, Taiwanese), and (Green, N/A), indicate this combination is extremely rare under the hypo- thesis of independence. This pattern suggests that in Taiwan, party politics is connected to ethnicity, even though there are a significant number of individuals whose partisan orientation cannot be probed by this survey method.
Notably, green-camp identifiers, compared to their blue camp counterparts, are less likely to claim that they are "both" Taiwanese and Chinese, while blue-camp identifiers are more likely to claim dual ethnic
59Pearson Chi-square statistic is the sum of squares of Pearson residuals. The Pearson re- siduals are calculated a s (n - m) / sqrt (m), where n is the counts of a va riable and m is the expected frequency. Standardized residuals which exceed the values 2 and 4 in absolute value highlight cells whose residuals are individually significant at approximately the .05 and .0001 level, respectively. See Da vid Meyer, Achim Zeileis, and Kurt H ornik, "The Strucplot Framework: Visualizing Multi-way Contingency Tables with vcd," Journal of Statistical Software 17, no. 3 (October 2006): 27.
identification instead of simply claiming to be Chinese. One explanation is that over the three decades of democratization in Taiwan, Chinese iden- tifiers have been self-reflecting about their identification. It is possible that embracing dual national/ethnic identification is a better way of distin- guishing themselves from those from mainland China while retaining their ethnic or cultural identity with an imagined China. However, again, this could be primarily caused by the vagueness of the wording of this survey question.
Another interesting pattern is that "Chinese" as a national/ethnic label lacks power to explain political camp identification. This suggests that some individuals who see themselves as Chinese identify with the green camp; similarly, it would be premature to say that pan-blue supporters are Chinese identifiers.
Figure 2 shows the contrast between political camp identification and state identification— new country name (Taiwan), Republic of China Figure 1
Contrasting Political Camp Identification with National Identification
Political Camp Identity Green
-4.00 -5.46 p-value =
(ROC), and reluctant to answer (N/A). It shows a clear and strong rela- tionship between the two variables, evidenced by high positive residuals in the blue cells, (Green, TW) and (Blue, ROC), as well as by high negative residuals in the red cells, (Blue, TW) and (Green, ROC). The red cells show that the number of pan-green camp supporters who dislike ROC ex- ceeds that of pan-blue camp supporters who dislike changing the national status quo. It is worth noting that blue-camp identifiers are less likely to avoid the national identification question, although it is not clear if green- camp identifiers would do the same.
Figure 3 shows the contrast between national/ethnic identification and state identification. The pattern that Taiwanese is consistent with those shown in figure 1, suggesting that the connection between national/ethnic identification and state identification is similar to that between national/
ethnic identification and political camp identity, particularly the cells (Tai- wanese, TW) and (both, ROC). What figure 1 does not show, but can be ascertained from figure 3, are two points. First, as expected, those who Figure 2
Contrasting Political Camp Identification with State Identification
< 2.22e-16 Pearson residuals:
-4.55 -2.00 0.00 2.00 4.00 7.81
N/A ROC TW
Political Camp Identity
claim to be Chinese are more likely to choose ROC rather than a new name.
Second, as the bottom left red cell suggests, individuals who do not state their national/ethnic identification are not likely to be those seeking a change in the status quo. In other words, individuals who seek change in the nation's name are likely to express their opinions in the survey. Individuals who have avoided answering the national/ethnic identification question are likely to remain reluctant to answer the national identity question.
In sum, the three cross-tabulate graphics show that the three identity questions are strongly related with each other. The findings confirm that individuals who identify with Taiwan as an ethnic name or a national name tend to be pan-green supporters. However, the following findings give us a new insight into the thoughts of Taiwanese voters: (1) Chinese identifiers are more likely to embrace dual identities (figures 1 and 3); (2) Chinese identifiers are not necessarily pan-blue supporters and can be pan-green supporters (figure 1); (3) pan-blue supporters (with respect to political camp identity) and Taiwanese identifiers (with respect to national/ethnic Figure 3
Contrasting National Identification with State Identification
< 2.22e-16 -5.59 -4.00 -2.00 0.00 2.00 4.00 6.57
identification) are strongly associated with willingness to express their ex- pectations about the national name (figures 2 and 3).
The above analysis shows the correlations between the three con- cepts— national/ethnic identification, political camp identity, and national identity. The following regression analysis employs multinomial log-linear regression to explore the extent to which these variables are endogenous to each other. Each of the three variables is regressed on the other two.
Results are shown in tables 2, 3, and 4.
Ethnic identification can be influenced by political camp identifica- tion, if it can be identified, and national identification. As table 2 shows, pan-blue supporters are more likely than pan-green supporters to identify themselves as Chinese or embrace dual ethnic identification. Additionally, those who have internalized the Republic of China's legitimacy are more likely than those who are concerned about that legitimacy and international status to identify as Chinese or both Chinese and Taiwanese. Moreover, individuals who dislike the pan-blue parties tend to reject being seen as Chinese and to claim that they are not simply Chinese but also Taiwanese.
Next, consistent with the above analysis, regression results suggest that one's political camp identification can be influenced by one's ethnic Table 2
Multinomial Model of National Identification (Base = Taiwanese) Identity Type
Political Camp (Ba se = pan-green) Blue camp O the r parties No affiliation Nationa l Na me (Base = N ew N ame) ROC Intercept
Beta S.E. Beta S.E.
0.80 0.00007 0.76
0.47 0.91 0.44
0.20 0.44 Note: N = 842; *p < .05; **p < .01; ***p < .001.
identification and national identification. Chinese identifiers are more likely to identify with the pan-blue camp but not other smaller parties.
Those who claim dual national/ethnic identification are more likely to be pan-blue supporters. Their flexibility in national/ethnic identification can lead them to support other smaller parties or to reject any partisan label.
Moreover, unlike the influence of ethnicity on political camp identity, the influence of national identity is more straightforward: individuals choosing the status quo are likely to support the pan-blue camp. If they do not like the blue camp, they will reject partisan labels. However, it is not clear if they will turn to other smaller political parties; the relationship between ac- knowledging the ROC and identifying with other smaller political parties is not statistically significant (see table 3).
Finally, both national/ethnic identification and political camp identi- fication can be movers of national identification. People who claim to be ethnically Chinese or both Chinese and Taiwanese, as well as people who are blue-camp supporters, tend to support or defend the status quo. It is worth noting that individuals who concealed their partisanship, the major- ity of respondents in this survey sample, are also concerned about adopting a new name for their country, even though they are aware of the risk of de- parting from the status quo (see table 4).
In sum, national/ethnic identification, political camp identification, and state identification are endogenous concepts in Taiwan. Setting aside Table 3
Multinomial Model of Party Identification (Base = Green-Camp) Identity Type
Both National Name
Blue Camp Other Parties No Affiliation
Beta S.E. Beta S.E. Beta S.E.
0.80 0.47 0.35 0.29
0.00005 0.91 0.91 0.77
0.76 0.44 0.25 0.15 Note : N = 842; *p < .05; **p < .01; ***p < .001.
the few individuals who identify with smaller parties and those who con- ceal their partisanship, the findings above suggest that Taiwanese people's political identity is simultaneously influenced by their reflections about ethnicity, feelings about political camps, and concerns about the country's status quo.
Conclusion and Discussion
The nature and the formation of national identification is an emerging topic in political science, particularly in political psychology and area studies of newly developing democracies. Prompted by concern about the divergence of opinions with respect to national identity in Taiwan, this paper addresses the complex relationships between national/ethnic identi- fication, party identification, and state identification. One major contribu- tion of this paper is the finding that, consistent with the Rochester School's revisionist perspectives, party identification, measured by Political Camp Identification, is not completely fixed in Taiwan, which suggests that fur- ther investigation is needed concerning whether partisanship is really a static, "unmoved mover" of political orientation.
Binomial Model of State Identification (Base = New Name) Identity Type
Chinese Both Political Camp
Blue camp Others No affiliation Intercept
0.63 0.20 0.35 0.91 0.25 0.23 Note: N = 842; ***p < .001.
Because political camp identification is strongly associated with state identification and national/ethnic identification in Taiwan, it can influence the other two types of identification and be influenced by them as well.
Therefore, I argue that state identification in Taiwan is a dynamic concept, subject to influence from political camp identification and national/ethnic identification. This scenario gives us a channel for understanding the for- mation of identity: through this dynamic process individuals with consis- tent identities in one or two areas are expected to be more stable in all other areas of identity as well. Others whose national/ethnic identification or political camp identification has been reshaped by their social network or by external events will experience changes in their state identification, if not ultimately in their state identity.
The findings of the present study also shed light on the developing discussion about dual identification. I find that individuals in Taiwan who claim dual national/ethnic identifications (both Chinese and Taiwanese) are not those who truly embrace one of two state identifications, i.e., Republic of China or "Republic of Taiwan." Instead, it is very likely that in effect they are Chinese identifiers who prefer the ROC's status quo. Therefore, this paper suggests that dual identification may not be a consequence of ambivalence as much as an attempt to avoid cognitive dissonance. Rather, it is more likely to be (1) a result of thoughtful reflection on one's identi- fication as a pre-stage of switching state identification, or (2) a superficial response to the mainstream opinion that using "Taiwan" was politically correct at the time of the survey, while using ROC may not have been. This pattern of behavior requires further inspection, explanation, and explora- tion.
This paper also contributes to current studies on Taiwan's efforts to be "independent." These studies mostly focus on macro-level, long-term changes in the percentages of individuals claiming to be "Chinese," "Tai- wanese," or "both." This study, based on a more micro perspective, sug- gests that this current way of measuring state identification may be over- simplified, if not misleading. In fact, this measurement is more about national/ethnic identification than state identification. The findings of this paper further suggest that because this conventionally used question is