2 .xwasmw we . . ”Hakka“? , .A ‘ ram...» . a fab»... “2.39:. . .315... . . .r. . .? “mania“? $3.1 . H ‘ a; in, tuuwMA 1L5udh| ~17,” ._. UL . . LC ‘ .. . . . . A .— THESIS ‘1 «,1 /" 1. . .A 1' t - 1/ "I LIIRARY Michigan State University This is to certify that the p p dissertation entitled ‘ The FO‘“"‘+‘°"‘ oc th‘a ldLn‘L‘Fi cdw»: (Li/mt} ‘HWC Elet+c>icfl€ in T“““"“"; \ng +0 ”76 presented by Lu .. hue; C he v1 has been accepted towards fulfillment of the requirements for 00(‘i'of' Phihbup‘hg’ degree in _ / m fl M Major professor r pawl fl. llbi‘t‘xmscifi \1’ ‘ 3 “1'11 Date MSU is an Affirmatiw Action/Equal Opportunity Institution 0-12771 PLACE IN RETURN BOX to remove this checkout from your record. To AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE JUN] 15 Zia? JUMZQLZWT 9&1 0 9 2007 ‘7 (a. &' c9 6/01 c2/CIRC/DateDuo.p65~p.15 THE FORMATION OF PARTY IDENTIFICATION AMONG THE ELECTORATE IN TAIWAN, 1987 TO 1996 By Lu-huei Chen A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Political Science 1999 ABSRTACT THE FORMATION OF PARTY IDENTIFICATION AMONG THE ELECTORATE IN TAIWAN, 1987 TO 1996 By Lu-huei Chen In this study, the formation of party identification among the electorate in Taiwan between 1987 and 1996 was examined. The social bases of support for the three major parties were shown, and the functions of party identification were investigated. As the opposition parties were established and the elections became competitive, the electorates had chances to formulate their party identification to deal with the complexity of politics in Taiwan. However, as discussed in this study, second-generation voters were more likely than voters in the other two generations to have stable partisan preference, and they employed their partisan preferences to determine their national identity. Therefore, second-generation voters might formulate their party identification and employ that identification to determine their political attitudes. Because of the limitation of available data, it was not clear that whether voters in the first and the third generations formulated a long—term commitment toward a particular party and employed it as a cue to understand politics. For my wife Elaine and my parents iii ACKNOWLEDGEMENTS I wish to thank Professors Paul R. Abramson and Tse-min Lin for their guidance in conducting this research and in writing of this dissertation. Professors Darren Davis, Jim Granato, and Brian Silver also provide valuable suggestions and constructive criticism for this study. I learn a lot from all of them, not only about their fields of specialization but also their views about finding research questions and developing solid research. I also wish to thank Professor Angela Merti g for her kindness in serving as an outside examiner on my dissertation committee. During my study at Michigan State University, many teachers give me constructive support. I wish to express my appreciation to Professors Ada Finifter, Scott Gates, Tom Hammond, Larry Heimann, Gretchen Hower, David Rohde, and Ken Williams. Many friends also provided their kindness and warmth during the time I was at MSU. Especially, Dave Lektzian provides warm friendship and solid support when we took classes and worked together. In addition, I want to express my thanks to Sara Benesh, Ching-Lun Woody Chang, Tsai—Lun Chian g, JenHei Chen, Hsin-Yen Chung, Erik Herron, Dave Jaye, Shun-Jie J i, Shu Keng, Alex Lee, Junhan Lee, Cheng-Hung Lin, Wendy Martinek, Misa Nishikawa, Ching-Ping Tang, Louisa Tseng, J ong-Tian Wang, Jonathan Wu, Wan-Ying Yang, and Chia-N an Yeah. I also appreciate Sue Cooley Miller’s expertise in editing my dissertation. Many Professors and friends in Taiwan also provide their generous helps during my study at MSU. I would like to express my thanks to Yih-yan Chen, Su-feng Cheng, Teh-fu Huang, Yun-tai Hung, I-chou Liu, Shing-yuan Sheng, Meng-hsi Tsai, and Ching- hsin Yu. I also want to express my appreciation to the Electoral Behavior Research Group at National Taiwan University and the Election Study Center at National Chengchi University for providing me with data sets. Finally, I am deeply indebted to my parents and my brothers. They were very supportive when I went abroad and pursued my Ph. D. at MSU. Special thanks are extended to my wife. She came with me to the United States my first year, and then returned to work in Taiwan. She is always the one who supports me the most. Without her, I could not have completed this dissertation. TABLE OF CONTENTS LIST OF TABLES vii LIST OF FIGURES xi CHAPTER 1. INTRODUCTION 1 Research Questions ......................................................................... 1 Social Forces, Issue Preferences, and Party Identification in Taiwan .................................................................................... 2 Outline of the Dissertation .............................................................. 7 2. THEORETICAL ELABORATION 10 Literature Review .......................................................................... 11 The Application of Party Identification in Taiwan ....................... 30 3. METHODOLOGY 32 Research Design ............................................................................ 32 Data and Measurement .................................................................. 34 The Research Hypotheses ............................................................. 38 4. GENERATIONS, ETHNIC GROUPS, AND PARTY IDENTIFICATION 44 Generation Effect, Life-Cycle Effect, and Party Identification 45 The Stability of Party Identification Among the Electorate in Taiwan .................................................................................. 48 Political Generation and Partisan Preference ................................ 52 Ethnicity and Partisan Preference ................................................. 58 The Emergence of the New Party ................................................. 66 Conclusion ..................................................................................... 72 vi 5. SOCIAL BASES OF PARTISAN SUPPORT 75 Social Cleavage and Partisan Support ........................................... 75 Social Bases of Support for the KMT ........................................... 78 Social Bases of Support for the DPP ............................................. 90 Social Bases of Support for the NP ............................................. 101 Conclusion ................................................................................... 1 11 6. ISSUE PREFERENCES, NATIONAL IDENTITY, AND PARTY IDENTIFICATION 114 National Identity, Political Values, and Partisan Preference ...... 115 Determinants of Partisan Preference ........................................... 122 Conclusion ................................................................................... 134 7. THE IDENTIFICATION PROBLEM: WHETHER PARTY IDENTIFICATION AFFECTS NATIONAL IDENTITY ............ 137 The Functions of Party Identification ......................................... 137 National Identity and Electoral Politics in Taiwan ..................... 138 Distribution of National Identity Among the Electorate ............. 142 How Partisan Preference Affects National Identity .................... 154 Simultaneity, Consistent Estimation, and Efficiency .................. 162 Conclusion ................................................................................... 164 8. CONCLUSION 168 Research Findings and Discussion .............................................. 168 Suggestions for Future Research ................................................. 174 APPENDIX A 177 B 178 C 183 D 187 vii BIBLIOGRAPHY 190 viii LIST OF TABLES Tables 1-1. Representatives of the National Assembly Elections in Taiwan, 1972 to 1996 ............................................................................................... 5 1-2. Legislative Elections in Taiwan, 1972 tol998 ................................................. 6 4-1. Percentage of KMT Partisans Among two Ethnic Groups, by Generation, 1987 to 1996 ................................................................................ 64 4-2. Two Measurements of Partisan Preference, by Generation, 1996 ................. 67 4-3. Two Measurements of Partisan Preference among Taiwanese, by Generation, 1996 .......................................................................... 69 4-4. Two Measurements of Partisan Preference among Mainlanders, by Generation, 1996 .......................................................................... 71 5-1. Percentages of KMT Partisans, by Gender, 1987 to 1996 ............................. 79 5-2. Percentages of KMT Partisans, by Level of Education, 1987 to 1996 .......... 83 5-3. Percentages of KMT Partisans, by Vocations, 1987 to 1996 ........................ 86 54. Percentages of KMT Partisans, by Subjective Socioeconomic Status, 1987 to 1996 ................................................................................ 88 5-5. Percentages of DPP Partisans, by Gender, 1987 to 1996 .............................. 92 5-6. Percentages of DPP Partisans, by Level of Education, 1987 to 1996 ............ 95 5-7. Percentages of DPP Partisans, by Vocations, 1987 to 1996 .......................... 97 5-8. Percentages of DPP Partisans, by Subjective Socioeconomic Status, 1987 to 1996 .............................................................................. 100 5-9. Two Measurements of Partisan Preference, by Gender, 1996 ..................... 103 5-10. Two Measurements of Partisan Preference, by Level of Education, 1996 ........................................................................................... 106 5-11. Two Measurements of Partisan Preference, by Vocation, 1996 ................ 108 5-12. Two Measurements of Partisan Preference, by Subjective Socioeconomic Status, 1996 ...................................................... 110 6-1. Determinants of Partisan Preferences Among the Electorate, 1993 and 1996 ..................................................................................... 123 6-2. Determinants of Partisan Preferences (DPP versus KMT), by Generation, 1993 ........................................................................................... 127 6-3. Determinants of Partisan Preferences (DPP versus KMT), by Generation, 1996 ........................................................................ 129 6-4. Determinants of Partisan Preferences (DPP versus NP), by Generation, 1993 ........................................................................ 132 6-5. Determinants of Partisan Preferences (DPP versus NP), by Generation, 1996 ........................................................................ 133 7-1. National Identity Among the Electorate, by Generation, 1993 and 1996.... 144 7-2. National Identity Among Taiwanese, by Generation, 1993 and 1996 ......... 148 7-3. National Identity Among Mainlanders, by Generation, 1993 and 1996 ........................................................................................... 150 7-4. Determinants of National Identity, 1993 and 1996 ...................................... 156 7-5. Determinants of National Identity, by Generation, 1993 ............................. 158 7-6. Determinants of National Identity, by Generation, 1996 ............................. 160 A-l. Great Events in Taiwan ............................................................................... 177 8-1. Distribution of Partisan Preferences, 1987 to 1996 .................................... 178 B-2. Distribution of Partisan Preferences, by Generation, 1987 to 1996 ............ 179 B-3. Partisan Distribution Among Taiwanese, by Generation, 1987 to 1996 ..... 180 B4. Partisan Distribution Among Mainlanders, by Generation, 1987 to 1996 ........................................................................................... 181 B-S. Distribution of the Electorate, by Political Generations ............................. 182 C-1. Party Preference Index, by Gender, 1987 to 1996 ...................................... 183 C-2. Party Preference Index, by Level of Education, 1987 to 1996 ................... 184 xi C-3. Party Preference Index, by Vocation, 1987 to 1996 ................................... 185 C-4. Party Preference Index, by Subjective Socioeconomic Status, 1987 to 1996 ........................................................................................... 186 xii LIST OF FIGURES FIGURES 2-1. Party Affiliation in the United State, 1952 to 1996 ....................................... 15 3-1. Division of Taiwan’s Six Political Generations ............................................. 39 3-2. The Relationship Between Partisan Preference and Other Variables ............ 43 4-1. Distribution of Partisan Preferences: 1987 to 1996 ....................................... 51 4-2. Percentages of KMT Partisans, 1987 to 1996 ................................................ 55 4-3. Percentages of DPP Partisans, 1987 to 1996 ................................................. 56 4-4. Percentages of KMT Partisans Among Taiwanese, 1987 to 1996 ................. 59 4-5. Percentages of DPP Partisans Among Taiwanese, 1987 tol996 ................... 60 4-6. Percentages of KMT Partisans Among Mainlanders, 1987 tol996 ............... 62 4-7. Percentages of DPP Partisans Among Mainlanders, 1987 to 1996 ............... 63 xiii CHAPTER 1 INTRODUCATION Research Questions Party identification is a long-term psychological attachment, and most voters in the United State have their own partisan loyalty. However, I wanted to discover whether party identification is also a useful concept to explain the political participation of people in Taiwan. Because the party system is not well established in this newly democratic country, it is of interest to understand the social bases of party support and the functions served by party identification in Taiwan. In this dissertation, I examined the social bases of party support in Taiwan to show how these social forces affect the electorate’s partisan preferences. As Lipset (1981) pointed out, people’s social characteristics, such as ethnicity, generation, gender, religion, educational level, vocation, and socioeconomic status, might affect their partisan support. Thus, I examined how these factors determine the partisan support of people in Taiwan. Also, people’s issue preferences might play an important role in their partisan preferences, so I investigated how people’s issue preferences might affect their partisan preferences. Party identification determines people’s political attitudes. According to the theory of party identification (Abramson 1983, Campbell et al. 1960), party identification plays an important role in shaping people’s opinion, influencing individual voting behavior and political participation, and increasing psychological involvement in politics. Thus, one can expect that party identification helps people construct their political views. A citizen with strong psychological attachment to a particular party is more likely to have a positive evaluation of that party, evidence greater interest in political affairs, and be more likely to participate in politics. In this research, I examined whether voters in Taiwan formulated their party identification and whether they employed their party identification to form their political attitudes. Measurements of partisan preferences employed in Taiwan might indicate people’s short-term evaluation of parties. However, those measurements might also capture people’s long-term psychological commitment toward political parties. One way to examine whether these measurements indicate respondents’ psychological attachment to political parties is to examine whether people employ their partisan preferences to determine their issue preferences. When people employ their partisan preferences to determine their stances on specific issues, we might consider that these measurements capture people’s long-term psychological attachment to political parties. Social Forces, Issue Preferences, and Party Identification in Taiwan To understand how the electorate in Taiwan forms its partisan preference, it is necessary to mention some specific historical events and different ethnic groups relating to this development. The Republic of China (ROC) was established in 1912, but Japan controlled Taiwan then. Taiwan was returned to the National government’s control in 1945, after Japan was defeated in World War 11. However, the National government retreated to Taiwan after the civil war in 1949, and most mainlanders came with Chiang Kai-shek to Taiwan at that time. Minority mainlanders (about 12 percent to 14.3 percent of the total population in Taiwan) ruled two other major ethnic groups, Min-nan-jun (Southern Fukienese) (72.5 percent) and Hakkanese (12.5 percent),1 and most officers in the central government were mainlanders. The ruling party, the National Party or the Kuomintang (KMT), still propagandized that it would stage a roll-back to mainland China, and kept educating its people as Chinese but not Taiwanese. To legitimate its power, however, the KMT provided local elections beginning in 1950, and it began to recruit Taiwanese into the central government when Chiang Ching-kuo, the son and successor of Chiang Kai-shek, became the Premier in 1972 (Chu 1998). In 1987, all 21 mayors and county magistrates were Taiwanese, but 80 percent of the Cabinet ministers were mainlanders (Tien 1989). There was only one major party--the KMT-- in Taiwan before 19862, and it dominated most national and local elections. In the National Assembly elections, the KMT won about 65 percent of the votes and 81.4 percent of the seats between 1972 and 1986 (see Table l-l). As shown in Table 1-2, the KMT also got about 73 percent of the 1 Both Min-nan-jun and Hakkanese used to be called Taiwanese in Taiwan. The Min-nan jun speak a Fukien dialect and the Hakkanese speak a different dialect. 2 Two small but legal satellite parties, the Young China Party (YCP) and the Democratic Socialist Party (DSP), were formed in China before 1949. However, neither party challenged KMT rule, and their leaders were aging mainlanders. They did not win any local or national elections in Taiwan, and their combined popular vote in elections totaled about 1 percent. See Tien (1989, chapter 4). votes and 83 percent of the seats in the legislative elections between 1972 and 1983. However, a new party, the Democratic Progressive Party (DPP), emerged before the National Assembly election and the legislative election in 1986 and began to challenge the KMT in the elections3. Two major platforms of the DPP were to advocate Taiwan’s independence and to promote democratic reforms in the 1986 elections (Hu and Chu 1992). Combining national identity and ethnic identity issues, the DPP successfully attracted supporters in the elections. In 1986, the DPP won 20 percent of the votes in the National Assembly election and 25 percent of the votes in the legislative election, and its share of votes increased to one third of the total votes in the 1995 legislative election. The DPP’s success in the national and local elections forced the KMT to undertake more democratic reforms and adopt a more ambiguous stance on the issue of unification with mainland China versus Taiwan independence (UM-TI issue). Therefore, President Lee Teng—hui, succeeding Chiang Ching-kuo as the first Taiwanese president in 1986 and the chairman of the KMT, recruited more Taiwanese into the power circle and undertook more rapid democratic reforms (Chu 1998). For example, the proportion of Taiwanese on the KMT Central Committee was 6.1 percent under Chiang Ching—kuo’s chairmanship during the Tenth Party Congress (1969-1976), but it increased to 53.3 percent under Lee Teng-hui’s chairmanship in 1995 (Hung 1996). In addition, during the constitutional reform, Lee favored popular election of the president instead of indirect election, and he transformed the political system from a parliamentary system into a presidential one. These measures gave the mainlander elite, the nonmainstream faction in 3The DPP were established in September 1986 by the opposition movement known as Tangwai (outside the [KMT] party). Most of its leaders are native Taiwanese. See Chu (1992) and Huang (1992). Table 1-1. Representatives of the National Assembly Elections in Taiwan, 1972 to 1996 Year Seats KMT TAN GWAl/DPP NP OTHER Vote Seat % Vote Seat % Vote Seat % Vote Seat % % (N) % (N) % (N) % (N) 19720 53 64-0 81.l( 43) -- -- -- -- 36.0 18.9(10) 1980“ 76 66.3 81.6( 62) 8.6 3.9( 3) -- -- 25,1 14,5 (11) 1986“ 84 64.2 81.0( 68) 19.9 13.1 (11) -- -- 16,9 6.0( 5) 1991 325 68.8 78.2 (254) 23.6 20.3 (66) -- -- 7.6 1.5( 5) 1996 334 49.7 54.8 (183) 29.9 29.6 (99) 13.7 13.8 (46) 6.8 1.8( 6) Source: Huang 1997: 146, Table3. a Vote statistics do not include those of aboriginal and occupational groups. Statistics on seats do not include those of overseas Chinese groups. Note: Parties: KMT, Kuomintang; DPP, Democratic Progressive Party; NP, New Party. the KMT, excuses to charge him with willingness to seek Taiwan’s independence (Lin, Chu, and Hinich 1996). To challenge President Lee's leadership in the KMT, in 1993, some KMT legislators defected from the KMT to form the New Party (NP) before the magistrate elections. Although the NP did not win any district in the 1993 election, it won 13 percent of the seats in the legislative election of 1995 and some other local councilors' elections. On the UM-TI issue, the NP favors the unification of Taiwan with mainland China, but the DPP prefers a separate identity for Taiwan. Because mainlanders are more likely to support unification with mainland China, one significant difference in the social bases of support of the DPP and the NP is that few mainlanders support the DPP, whereas half of the NP's votes come from mainlanders (Huang 1994). Table 1-2. Legislative Elections in Taiwan, 1972 to 1998 Year Seats KMT TANGWAI/DPP NP OTHER Vote Seat % Vote Seat % Vote Seat % Vote Seat % % (N) % (N) % (N) % (N) 19720 36 73.1 83.3( 30) -- -- -- -- 26.9 16.7( 6) 19750 37 77.6 81.1( 30) -- -- -- -- 22,4 13.9( 7) 1980“ 70 71.9 80.0( 56) 13.0 11.4( 8) -- -- 15,1 8,6( 6) 1983“ 71 69.4 87.3( 62) 18.9 8.5( 6) -- -- 11.7 4.2( 3) 1986“ 73 66.7 80.8( 59) 24.6 16.4(12) -- -- 8,7 2,7( 2) 1989 101 59.2 71.3( 72) 29.9 20.8 (21) -- -- 10.9 7.9( 8) 1992 161 52.7 58.4( 94) 31.4 31.7 (51) -- -- 15.9 9.9 (16) 1995 164 46.1 51.8( 85) 33.2 32.9 (54) 13.0 12.8 (21) 7.8 2.4( 4) 1998 225 46.4 54.7 (123) 29.6 31.1 (70) 7.1 4.9 (11) 16.9 9.3 (21) Sources: 1. Huang 1997: 146, Table 4. 2. The author used newspaper reports in calculating the data on the 1998 election. a Vote statistics do not include those of aboriginal and occupational groups. Statistics on seats do not include those of overseas Chinese groups. Note: Parties: KMT, Kuomintang; DPP, Democratic Progressive Party; NP, New Party. Therefore, the relationship among party loyalty, national identity, and ethnicity is very important in understanding the political participation in Taiwan. From a comparative perspective, I examined party identification among the electorate in this newly democratic nation. In addition, I examined the functions played by party identification and tested whether party identification is a useful concept in Taiwan. The change and continuity of party loyalty of voters in Taiwan also provides a great laboratory to examine how issues shape, or are shaped by, the individual’s party affiliation. In this research, by employing survey data, I investigated the social bases of party support. Also, I examined the relationship between individual partisan preference and their national identity. Outline of the Dissertation There are eight chapters in this dissertation. Literature on party identification in the United States and the application of party identification outside America is reviewed in Chapter 2. The discussion focuses on the origins of partisanship, the stability of party identification among individuals and electorates, the best way to measure partisan loyalties, the dimension of party identification and the meaning of Independents, the relationship between party identification and issue preferences, and party identification as a cross-nation concept. I also consider how to apply the research results to Taiwan. In Chapter 3, the methodology used in this study is explained. I employed survey data collected by the Electoral Behavior Research Group at National Taiwan University (NTU) between 1987 and 1996, and I conducted secondary data analyses to answer the research questions. Because the DPP was eStablished before the 1986 elections and the legislative elections are national elections in Taiwan, I examined surveys conducted after the legislative elections were held. I include the sampling process used in the surveys and sample sizes of these surveys. Further, I demonstrate different measurements of partisan preferences in these data sets, and present my research strategy by using these data sets. Because no panel studies employing islandwide samples are available in Taiwan, I did not plan to review the stability of partisan preferences of individual voters. The research strategy was to examine the aggregate level of partisan stability between 1987 and 1996, and to examine generational differences in the stability of partisan preferences. In Chapter 3, I also define political generations in Taiwan and discuss other variables used in this research. After elaborating on theory and methodology, I examine generational differences in partisan support in Taiwan. According to the theory of party identification, such identification is a long-term psychological attachment to political parties. However, different political generations in Taiwan might have different political experiences. In addition, different ethnic groups also have different political socialization experiences from their parents. In Chapter 4 I examine the differences between ethnic groups and among generations in regard to their party preferences. In Chapter 5, I show the social bases of party support and discuss how different sociological backgrounds affect the partisan support of the electorate. I focus on several demographic variables, such as people’s gender, educational level, vocation, and subjective socioeconomic status, in an attempt to discover how these social forces might affect people’s partisan preferences. I also employ people’s political generation as a mediator variable to detect generational differences along with differences based on other social characteristics. Further, I examine how the emergence of the NP might affect the social bases of the two major parties. There are a number of controversies regarding the relationship between party identification and issue preference. In Chapter 6, I examine the relationship between issue preference and partisan preference of voters in Taiwan. I demonstrate how people’s preferences regarding two major issues, i.e., "concerns about money politics" and democratic values, affect their partisan preferences. I include people’s national identity in this model to explain their partisan preferences. Because of the complicated relationship between Taiwan and mainland China, people’s national identity plays a crucial role in determining their partisan support. Because measurements of partisan preferences employed by the NTU might capture not only people’s short-term partisan preferences but also their long-term commitment to political parties, I examined whether people employ their partisan preferences to determine their national identity. This is discussed in Chapter 7. Chapter 8 is the conclusion of this dissertation. In this chapter, I discuss the research findings and present suggestions for future research. CHAPTER 2 THEORETICAL ELABORATION "Party identification is an attitudinal variable that measures an individual’s sense of attachment to a political reference group" (Abramson 1983: 71). It is one of the most important variables in the political attitude and voting behavior research. Party identification "is a psychological identification, which can persist without legal recognition or evidence of formal membership and even without a consistent record of party support" (Campbell et a1. 1960: 121). Two elements have been central to the whole notion of party identification: "an extended time horizon and some engagement of partisan feelings with self-identity" (Converse and Pierce 1985: 144). There are several characteristics of this long-term psychological attachment. First of all, party identification is highly stable. Second, party identification develops early in life. Third, it increases in strength during one's lifetime4 (Campbell 1960). The importance of party identification rests not only in its direct effect on the individual's vote, but also in its effect on other major determinants of the vote (Niemi and Weisberg 1993a: 210). In addition, most Americans identify with a party, and the individual's party ‘ Some controversy exists about the presence of life-cycle effects in the development of partisan strength. For a summary of this debate, see Abramson (1989). 10 identification is relatively stable, so party identification plays a crucial role in explaining the individual’s vote choice and in maintaining the political order. This chapter is divided into two parts. In the first part, I review literature relating to several important issues regarding the concept of party identification. In the second part, I discuss the application of party identification in Taiwan. Literature Review In this section, I review literature relating to several important issues regarding the concept of party identification. First, I discuss the origins of party identification. Second, I examine the stability of party identification among individuals and electorates. Third, I discuss whether there is a proper way to measure partisan loyalty. In addition, I explore the extent to which party identification shapes, or is shaped by, policy preferences. Then, I examine the dimensionality of the party identification measure and the meaning of partisan independence. Finally, I discuss the application of party identification to cross-nation research. The Origins of Partisanship Socialization research suggests that "most persons develop party loyalties before they develop positions on the issues" (Hyman 1959: 74-81; cited in Abramson 1983: 75). As Abramson (1983: 86) pointed out, one "of the earliest findings of political- socialization research is that most Americans learn their parents’ party loyalties and that these loyalties are learned early in life." Campbell and his colleagues (1960) also 11 demonstrated the consistency of respondents’ party identifications with those of their parents. However, early research depended on respondents’ recall of their parents’ party identification, so one must be cautious about accepting the validity of early research. We need direct information from both the respondents and their parents to examine how parents transmit their party loyalties to their offspring. To explore the process of socialization, the Survey Research Center (SRC) of the University of Michigan conducted a student-parent study in spring 1965. Respondents were interviewed first in 1965, and the same respondents were re-interviewed during 1973 and 1982. These data sets provided more direct evidence for understanding the process of socialization between parents and their offspring. Niemi and Jennings (1991) showed that parents’ party identification exercised a great influence on their offspring’s party identification during their high school years. As these offspring reached their mid-208, the effect of parental party identification was still significant, but its importance decreased. At the same time, the offspring’s own stances on issues began to influence their party identification. When the offspring were in their mid-308, their parents’ party identification still had a similar magnitude of influence on their party identification, but the offspring’s own issue stances increased in importance in shaping their party identification. Therefore, parents’ party identification may be one of the resources through which offspring gain their party loyalties, but still other factors affect the genesis and duration of individuals’ party identification. Because there were different historical experiences and socialization processes for different generations of various ethnic groups in Taiwan (Liu 1995a), it is beneficial to examine the difference in 12 partisan preference between different generations of various ethnic groups in that country. The Stability of Party Identification Among Individuals and Electorates The importance of party identification rests on two facts. First, a majority of people have an attachment to a specific party. Therefore, Campbell and his colleagues (1960: 121) asserted that there are few "factors of greater importance for our national elections than the lasting attachment of tens of millions of Americans to one of the parties. These loyalties establish a basic division of electoral strength within which the competition of particular campaigns takes place. And they are an important factor in assuring the stability of the party system itself. " Second, individuals’ party identification is relatively stable. "A general observation about the political behavior of Americans is that their partisan preferences show great stability between elections" (Campbell et al. 1960: 120). Therefore, Campbell and his associates (1960: 166) claimed that "party identification is typically a life-long commitment," and it can be "changed on the national scale only by major social cataclysms" (see also Miller and Shanks 1996: 132). As shown in Figure 2-1, the aggregate distribution of party identification was relative stable from 1952 to 1964, and more than three out of four respondents identified with one major party. However, the proportion of partisan identifiers was declined nationally since 1964, and fewer citizens have steady and strong psychological attachment with a party than in the past. The effect of party identification on people’s electoral choice and political evaluations also has declined (Nie, Verba, and Petrocik 1979). 13 Although party identification in the aggregate has declined since 1964, there is some evidence that party identification is still the most stable attitude at the individual level. Two three-wave panel studies conducted by the SRC in 1956-1958-1960 and 1972- 1974-1976 showed that citizens rarely cross the boundary between identification and nonidentification (Converse 1964; Converse and Markus 1979). The SRC student-parent three-way panel studies also confirmed this finding (Jennings and Markus 1984). Therefore, party identification is relatively stable at the individual level. Another issue relating to the concept of party identification is whether it increases in strength during one’s lifetime. As Campbell and his colleagues (1960: 161) claimed, Whenever we encounter relationships of this sort involving age, we are forced to choose between two competing interpretations. First, age may mark an historical epoch in which the person has matured or undergone some special variety of experiences that has left an imprint on his attitudes and behaviors... .The second interpretation presumes instead that older people will always feel stronger bonds with a political party than will newer members of the electorate. Therefore, the generational or/and life-cycle effect might play an important role in developing party identification. Before examining these effects, we need to define several terms. The generational or cohort effect "is produced by influences associates with birth cohort membership" (Glenn 1977: l 1). The life-cycle or aging effect is produced by influences associated with the process of growing older. And the period effect is "produced by influences associated with each period of time" (Glenn 1977: 11). "If a life- cycle explanation for the weak party loyalties of the young were valid, we would expect their feelings of party identification to grow stronger with age. . .A generational explanation. . .assumes that there is a formative period during which fairly enduring attitudes are learned; after that, attitudes tend to become relatively stable" (Abramson 1983: 110). 14 e2: 3 $3 "83.... 3:5 2.. 5 553.54.. 3:.— .2 2:3..— .Ewwioaz we basicb 05 mo 3:55 EVE—om c8 5:50 05 use 5:50 cacaomom aogw 05 3 883:8 £25m Humane.» T:%:&%E a. .553 + 535.5— 613? IT :35an wuehm lol 08— 32 No9 82 $2 owe vwE $3 82 wha 2.9 :9 N52 03— meg — _ _ _ L! L 1F _ h p _ _ _ _ 83 39 N03 82 $2 83 $3 N3— . p p — — h — m— vdm v.96. rem 15 Campbell and his colleagues (1960:163) asserted that " group identification is a function of the proportion of a person’s life he has been associated with the group. The longer a person thinks of himself as belonging to a party, the stronger his sense of loyalty to it will become." They argued that partisan strength increases with age. Converse (1969) analyzed the survey data from the five-nation studies conducted by Almond and Verba, and he concluded that the individual ’s partisan affiliation increases with aging. However, it is not easy to distinguish the generational effect from the life-cycle effect if one analyzes only the cross-sectional data (Abramson 1983). As longitudinal data become available, one might be able to separate these two effects. From the student-parent panel studies, it was found that parents had stronger party identification than their offspring did (Niemi and Markus 1984). N iemi and Markus (1984: 1015) concluded that "the high school class of 1965. . .began their adult political lives strikingly less committed to political parties than were their parents, and they have remained less committed for nearly two decades" (See also Abramson 1983; Jennings and Niemi 1981). Norpoth and Rusk (1982) used the SRC-CPS surveys from 1964 to 1976 to test whether a life-cycle effect or generational replacement contributed to the decline of partisanship. They showed that generational replacement explained the lion's share in the decline of party identification between 1964 and 1976. Abramson (1983:115) examined SRC-CPS election surveys (1952-1980), Gallup data (1945-1971), and the Michigan student-parent data (1965 and 1973), and he concluded that "all [these data sets] provide compelling evidence that partisan strength has not increased with age" (1983: 115). 16 As discussed above, in the case of the United States, it has been found that young citizens enter the electorate with a weak link to political parties, and the strength of party identification does not increase as they age because they have a different political experience from their parents. On the other hand, if the young enter the electorate when the elections are more competitive or the regime is more democratic, they might have a higher level of partisan preference than the older generation does. Therefore, examining a newly democratic country, such as Taiwan, provides a good opportunity to determine the generational effect on the intensity of party identification. The Best Way to Measure Partisan Loyalties Converse and Pierce (1985: 134) admitted that "there is no ’right’ way to measure partisanship which somehow is optimal for all times, places, and research questions." However, the concept has two major components: it is a long-term and psychological attachment to political groups (Campbell et al. 1960, Converse and Pierce 1985). In the following discussion, 1 demonstrate how different measures affect our understanding of the stability of partisan loyalty and American politics. There are at least two measures of party affiliation. The first is employed in the Gallup polls. The question used in the Gallup polls reads: "In politics, as of today, do you consider yourself a Republican, a Democrat, or an Independent?" The other measure was developed by the Survey Research Center in Michigan. The basic Michigan SRC question reads: "Generally speaking, do you usually think of yourself as a Republican, a Democrat, an Independent, or what?" These two different measures might capture different components of the individual ’s party affiliation. Converse (1976: 35) argued: l7 The ’general’ and ’usually’ qualifiers in the SRC question were originally intended to broaden the time reference and properly classify the long-term identifier who is momentarily piqued at his own party, or tempted to defect temporarily to vote for a charismatic candidate of another party. A verb like ’consider’ in the Gallup question has somewhat parallel, if perhaps weaker, overtones; but the ’as of today’ invites in the baldest way a very transient frame of reference. Abramson and Ostrom (1994: 23) also argued that "the Gallup party affiliation question taps a different aspect of partisanship than the SRC question. On their faces, the SRC ('generally speaking'. ..'usually') and the Gallup ('as of today') items appear to have a different time referent, with the Gallup item having a short-term focus." Therefore, when "short-term forces are weak, the Gallup and SRC questions are likely to yield similar results--they will be low levels of variability. When the political context is volatile, both measures will change, but the degree of change will be different" (Abramson and Ostrom 1994: 25). If Converse‘s and Abramson and Ostrom's arguments are correct, one can expect that the fluctuation of the Gallup measure will be greater than the SRC measure, and the Gallup measure will be more sensitive to short-term forces, such as economic condition and political evaluation, than the SRC measure. MacKuen, Erickson, and Stimson (1989) employed the Gallup measure to demonstrate the dynamic of macropartisanship. They (1989: 1127) argued that "party identification may be treated as a continuous macro phenomenon measured through time." They used the Democratic percentage of the major party identifiers as the measure of macropartisanship to explore the relationship between macropartisanship and other economic and political evaluations. First, MacKuen and his colleagues demonstrated the variability of macropartisanship and presented the great fluctuation in macropartisanship since World War H. The authors also showed the electoral consequences of macropartisanship. They (1989:1129) find that "a one-point shift in partisanship yields a 18 three-seat gain in House election (R2 = .38)." In addition, MacKuen and his colleagues argued that macropartisanship is shaped by, but does not shape, economic evaluation and the presidential approval. Therefore, they (1989: 1139) asserted, Knowing that the public’s partisanship is subject to considerable variation forces us to consider the standard view of party systems and realignment theory....Indeed, we discover that the partisan balance varies according to the political and economic performance of various govemments.. . .The mid-range dynamics we highlight are of tangible importance. They yield partisan movements of realignment magnitude (though not realignment duration) that require neither miracles nor catastrophes but instead arise from the routine success and failure of ordinary politics. MacKuen et al. demonstrated presented statistical evidence of the dynamic of macropartisanship. However, they did not mention whether the difference between the Gallup measure and the SRC measure matters. Abramson and Ostrom (1991) argued that difference between the questions really matters. They compared the Gallup measure with the SRC measure and found that the Gallup measure of party identification evidenced larger variability than the SRC measure. They also replicated MacKuen and his colleagues' model (1989), and found that when the Gallup measure of party affiliation was employed, even with fewer cases, macropartisanship was influenced by both the index of consumer sentiment and presidential approval. However, when they used the SRC measure, they found that macropartisanship was not shaped by the index of consumer sentiment or presidential approval. Abramson and Ostrom (1991 : 190-1) concluded, The SRC measure ...has substantially less volatility than the Gallup measure. Moreover, results using the NES and GSS surveys show less total variation than the results using the Gallup survey. The NES results are not strongly related to congressional election results. Finally, a time series analysis employing the NES surveys suggests that the SRC measure is not strongly driven by short-term economic and political evaluation. 19 They (1991:82-3) argued, "Compared with the Gallup measures, the NES measure of partisanship is not strongly related to short-term electoral outcomes and does not appear to be driven by short-term economic and political evaluations." MacKuen et a1. (1992) also presented evidence employing the CBS News and the New York Times (CBS-Times) measure of party affiliation to respond to Abramson and Ostrom’s challenge. However, they did not reply to Abramson and Ostrom’s (1991) argument about the effect of different question wordings. Therefore, Abramson and Ostrom (1994) used an experimental research design to examine whether these two different measures really made a difference. To clarify the effects of different question wordings, Abramson and Ostrom conducted six statewide telephone interviews to examine the effect, if any of difference in wording. Because the Gallup measure is more sensitive to short-term forces than is the SRC measure, it was expected that the relationship between short-term evaluations found using the Gallup measure would be stronger than that using the SCR measure. Abramson and Ostrom found that the zero- order correlation between the Gallup measure and short-term indicators was higher than that between the SRC measure and such indicators. "If the Michigan SRC measure taps long-term partisan loyalties compared with the Gallup question, responses to the SRC question should be more stable over time than responses to the Gallup question" (Abramson and Ostrom 1994: 40). Abramson and Ostrom made two-wave, three-wave, and four-wave comparisons between these two measures to examine over-time stability of partisanship across the four-wave panel studies. Because the fourth wave in the panel studies was "conducted only two weeks before the election, [we can expect that it was] at the point of maximum 20 political volatility" (Abramson and Ostrom 1994: 42). As they expected, in "six of the seven comparisons including the fourth wave, respondents who are always asked the SRC question over time are at least 10 percentage points more likely to give consistent responses than are respondents who are always asked the Gallup question" (Abramson and Ostrom 1994: 42). Green, Palmquist, and Schickler (1998) replicated the work of MacKuen et al. (1989) by extending the observations of time periods and applying different statistical models. They found that consumer sentiment and presidential approval had smaller influence on macropartisanship, but partisanship was strongly affected by its own past. From the above arguments, one can see that different measures might capture different components of party affiliation. The Gallup measure is more volatile and sensitive to short-term evaluations, whereas the SRC measure is more stable and less likely to be affected by short-term political and economic evaluations. Because party identification contains long-term commitment and affective factors, the SRC measure seems to capture these components better than the Gallup measure. However, it is hard to guarantee that the SRC measure was properly translated into Chinese to tap the long-term psychological component of this concept in Taiwan. Also, panel studies are available in Taiwan. Therefore, a better research strategy for this study was to employ longitudinal data sets to examine whether the aggregate-level distribution of partisan preferences is stable over time. 21 Dimension of Party Identification and The Meaning of Independent As to the measure of party identification, scholars also focused on two other issues. One is the meaning of Independents, and the other is the dimension of party identification. Both issues relate to the meaning and the measure of partisan Independents. The ideal independent citizen is well informed, pays attention to public affairs, knows what the government is doing, and votes based on judging the different platforms of various parties. However, the myth of independent voters seems to be unrealistic. Far from being more attentive, interested, and informed, Independents tend as a group to be somewhat less involved in politics. They have somewhat poorer knowledge of the issues, their image of the candidates is fainter, their interest in the campaign is less, their concern over the outcome is relatively slight, and their choice between competing candidates. . ..seems much less to spring from discoverable evaluation of the elements of national politics. (Campbell et al. 1960: 143) By employing the SRC measure, one can separate respondents into four categories: independents, leaners, weak identifiers, and strong identifiers. Some scholars have argued that the learner has higher political involvement than the weak partisan does. Petrocik (1974) tested whether the above mentioned four-point strength-of—party- identification scale is monotonically related to attitudinal variables. He found that only two out of ten variables met the theoretical expectations. Leaners are more interested in politics and the present campaign than were weak identifiers. Leaners also were more concerned about the election outcomes and more likely to vote than weak identifiers. Weak identifiers received lower scores than learners on three indexes: political efficacy, political participation, and use of the mass media to follow the campaign. Abramson (1983: 80, Table 5.1) analyzed eight presidential elections between 1952 and 1980. He discovered that "Independents who leaned Democratic were as likely 22 as, or more likely than, weak Democrats to vote Democratic. Likewise, in most of these elections, Independents who leaned Republican were as likely as, or more likely than, weak Republicans to vote Republican."5 However, as Abramson (1983: 80) pointed out, "Leaning Independents may be true Independents, but when pressed to say which party they feel ’close’ to, they define that closeness according to the way they plan to vote for president" (see also Miller and Shanks 1996; Shively 1980). Because party identification is an attitudinal variable, not a behavior record, and it contains an affective component, "it seems reasonable to consider leaning Independents to be more partisan than Independents with no partisan leaning, but to be less partisan than weak identifiers" (Abramson 1983: 81). The SRC measure of party identification assumes a unidimensional scale with strong Democrats at the left end, strong Republicans at the other end, and Independents in the middle. As Campbell and his colleagues asserted, the SRC measure of party identification invites "the individual to state the direction and strength of his partisan orientation. ...[It] permits us to place each person in these samples on a continuum of partisanship extending from strongly Republican to strongly Democratic" (1964: 122- 123). In the 1980 election study, CPS introduced some new measures to examine the meaning of partisanship. Weisberg (1980) employed these new measures to examine the validity of the SRC measure of party identification. First, he constructed a "party 5 For turnout, Abramson and his colleagues (1994: 118, Table 4-4, 1998: 83, Table 4-4) demonstrated that "there are no consistent differences in reported turnout between weak partisans and independents who leaned toward a party" among Whites in the 1992 and 1996 elections. Abramson (1983: 295, Table 16.1) also demonstrated similar results among Whites from 1952 to 1980. In five out of eight elections, weak partisans had lower turnout rates than did those independents who leaned toward a party. 23 difference" index by subtracting the feeling thermometer for Democrats from that for Republicans. Second, he (1980: 48) used the following question to construct a "party support" scale: "In your own mind, do you think of yourself as a support of one of the political parties, or not? (If yes,) which political party do you support? On this scale from 1 to 7 where 1 means ’not very strongly’ and 7 means ’very strongly,’ please choose the number that describe how strongly you support the (Republican/Democratic) Party." The distribution of "party support" was from -7 for strong Republican supporters to +7 for strong Democratic supporters, with 0 for those not supporting either party. Another variable is "party independent, " which Weisberg (1980: 50) measured using the following question: "Do you ever think of yourself as a political independent, or not? (If yes,) on this scale from 1 to 7 (where 1 means ’not very strongly,’ and 7 means ’very strongly’), please choose the number that describes how strongly independent in politics you feel." Weisberg (1980) argued that these new measures can be used to capture the facet of Independents in party identification. From his analysis (W eisberg 1980: 40, Table 2), he demonstrated that about one in six respondents considered himself or herself to be both partisan and Independent simultaneously. Weisberg (1980: 46) found that the "party difference" measure had a higher correlation than the seven-point party identification scale with the individual ’s vote. Weisberg also showed that those who neither supported any party nor considered themselves Independents were more likely to fall-into the weak partisan category of the traditional measure of party identification (see Weisberg 1980: 52). 24 A better way to examine differences between different measures is to conduct panel studies by employing both measurements with the same respondents at different times. However, no data are available from panel studies employing different measurements. Therefore, I focused on the stability of partisan preferences among the electorate in Taiwan. Party Identification Shapes, or is Shaped by, Policy Preferences The standard view is that party identification, as a long-term psychological attachment, affects the individual’s evaluations of issues and candidates, but it in turn is largely unaffected by such evaluations, except in dramatic circumstances (Abramson 1983; Abramson et al. 1998; Campbell et a1. 1960). In recent years, there have been challenged to the claim that party identification is unaffected by policy preferences. Fiorina ( 1981) argued that partisan identification is a "running tally" of past experiences. MacKuen, Erikson, and Stimson (1989) demonstrated that quarterly measurements of macropartisanship will vary in response to changes in presidential approval and in public evaluation of the state of the national economy. Two different views are discussed in the following section. Because national politics are remote and complex, party identification can serve as a political cue to help citizens organize their political world (Campbell et al. 1960; Miller 1976). Campbell and his colleagues demonstrated that "party has a profound influence across the full range of political objects to which the individual voter responds. The strength of relationship between party identification and the dimension of partisan attitude suggests that responses to each element of national politics are deeply affected by 25 the individual’s enduring party attachment" (Campbell et al. 1960: 128). In addition, "the influence of party identification on attitudes toward the perceived elements of politics has been far more important than the influence of these attitudes on party identification itself" (Campbell et al. 1960: 135). Therefore, Campbell and his colleagues (1960) believe that the individual ’8 party identification shapes his/her policy preference. Abramson (1983: 72) said that Campbell and his colleagues advanced four major claims about the functions performed by party identification: Party identification contributes to opinion formation. Party identification influences voting behavior. Party identification enhances psychological involvement in politics. High levels of party identification among the electorate provide a check against new party movements and contribute to the established party system. PPP!‘ Abramson and his colleagues (1998: 174-185) presented four examples to illustrate the association between the individual ’8 party identification and his/her other political attitudes. First, an identifier tends to give more positive evaluations to his or her own party’s candidate. Second, there is a strong relationship between the individual ’s party identification and his or her evaluation of presidential performance. Third, a partisan identifier tends to have close stances with his or her party’s nominee. Finally, an identifier is more likely to give positive evaluations to his or her own party on a summary measurement of retrospective evaluations during the last six general elections (1976 to 1996). The distribution of party identification has declined since 1964, so some scholars have begun to question the stability and the function of party identification. Some scholars have begun to examine whether the individual ’s policy preference affects his or her party identification. Fiorina (1981: 84) argued that party identification is a "running 26 tally of retrospective evaluations of party promises and performance." Analyzing the SRC and CPS panel studies in 1956-60 and 1972-76, Fiorina found that about 15 to 20 percent of respondents changed their partisan categories (Democrat, Independent, Republican) during the two-year periods (1956-58, 1958-60, 1972-74, 1974-76). Fiorina (1981: 96) argued that "retrospective evaluations can play a major role in moving individuals up and down the party identification scale." However, he (1981: 96) also admitted that "the more partisan an individual, the less responsive his or her reported partisanship will be to retrospective evaluations," but, "as the latter cumulate over time, even strong identifiers may eventually cross the threshold of their category." Hence, Fiorina (1981: 102) asserted that "there is an inertial element in voting behavior that cannot be ignored, but that inertial element has an experiential basis; it is not something learned at mommy’s knee and never questioned thereafter. " Page and Jones (1979: 1078) also argued, "It seems quite plausible to us that policy preferences ...may be both causes and consequences of party loyalties. ...[And] we must consider the possibility that voters' party loyalties may both affect and be affected by comparative candidate evaluations or intended votes during a particular Campaign period." From their model, they demonstrated that an individual's policy preferences have strong effects on his or her party attachment. Jackson and Franklin ( 1 983: 968) showed that "party identifications are subject to change as individual preferences change, and the change comes from shifts in people's perceived party 'Proximities relative to their own preference." Luskin and his colleagues (1989) used a CFoss-sectional survey in 1976 to examine the intergenerational transformation of partisanship from parents to their offspring. They found that offspring deviated from their 27 parents’ partisanship as a function of issue proximities. Niemi and Jennings (1991) used the student-parent, three-way panel studies to demonstrate that issues play a role in determining partisan preferences in the offspring. When applying the concept of party identification outside the United States, one needs to know the historical and political context of that country. Older generations in Taiwan did not experience competitive elections until 1986, but their stances on the issue of unification with mainland China versus Taiwan independence might be a part of their "belief system." However, the future relationship between Taiwan and mainland China is remote to the younger generation, but they have experienced and/or supported the growth of the opposition parties during the past two decades. Therefore, in examining the relationship between party identification and issue preferences in Taiwan, one also must take generational differences into account. Party Identification as a Cross-nation Concept The concept of party identification is developed in the United States, and it plays a critical role in voting behavior among the American electorate. Whenever this concept i 8 applied to other countries, the differences in the historical contexts and party systems of those countries must be considered. Many scholars have applied the concept of party identification to Western democracies, and have demonstrated that party identification helps to understand political participation outside the United States. However, Butler and Stokes (1974) showed that Voters tended to change their partisan preference to accompany vote switching. Campbell and Valen (1966) demonstrated that party identification was an excellent predictor of 28 individuals’ voting behavior in Norway. Converse and Dupeux (1962) studied party identification in France. They brought scholars to focus on whether one’s party identification is a long-term psychological commitment or whether it will change in response to short-term factors, such as one’s voting behavior, evaluations of party leaders, or perceptions of parties’ positions on issues (For suspicions about the application of party identification, see Budge and Farlie 1976; Clarke and Stewart 1987; Crewe 1976; Kaase 1976; LeDuc et al. 1984; Shively 1972; Stewart and Clarke 1998; Thomassen 1976. For defenses of the application of party identification, see Cain and Ferejohn 1981; Johnston 1992; Norpoth 1978; Richardson 1991; Schickler and Green 1997.) One problem with applying party identification outside the United States is finding an equivalent measurement of this concept (Converse and Pierce 1985, Dalton 1998, Kaase 1976). As Norpoth (1978: 42) argued, "The literal translation of the American question would confound party membership and subjective identity" in Germany. Johnston (1992) also showed that different question wordings and question orders affected the distribution of nonpartisans and the stability of party identification in the Anglo-American Democracies. Schickler and Green (1997) argued that a measurement error is included in the estimation, party identification is very stable across time and countries. In addition, different institutional settings and party systems across countries need to be considered. Lipset and Rokkan (1967) showed that party systems and voter alignment in Europe have evolved out of a deep social cleavage. "In a party system having a close relationship between the parties and the social classes, it is difficult to isolate the independent influence which party identification itself has on the electorate" (Campbell and Valen 1966: 268). Therefore, Miller (1976:30) contended, "It is not 29 necessary nor important to demonstrate some type of primacy for party identification in the phenomenology of the voter. It is important to understand the social psychology of individual political behaviour, and that involves understanding many of the functions attributed to party identification." Therefore, I focused on the functions of party identification in Taiwan. The Application of Party Identification in Taiwan When party identification is applied to Taiwan, the specific historical events and particular social and political settings need to be considered. Because most mainlanders followed the leader of the KMT, Chia Kai-shek, to Taiwan in 1949, we must realize that these historical events and personal experiences affected partisan preferences among mainlanders. In addition, composing major proportion of the population, Taiwanese were ruled by minor mainlanders through authoritarian control from 1949 to 1987, so these political experiences also influenced the formation of their partisan preferences. I examined electors from different generations to see whether these historical events played an important role in the formulation of their partisan preferences. I also examined partisan support of different ethnic groups to see how the socialization process in different ethnic families might have affected people’s partisan preferences. The opening of electoral competition in the national elections of 1972 and the formation of the DPP in the 1986 elections provided a different political experience for the electorate. Open elections provide excellent chances for political parties to organize their supporters and promote their platforms. Therefore, I examined the social bases of 30 partisan support to examine how people’s social characteristics might affect their partisan preferences. Also, because the NP was established in 1993, I examined whether its emergence affected the social bases of partisan support of two other major parties. Other political attitudes might affect people’s partisan preferences as well. I investigated how people’s democratic values, "concerns about money politics," and national identity might affect their partisan preferences. Finally, the importance of party identification is its influence on other political attitudes. I examined how people’s partisan preferences might affect theirs national identity. If voters have stable partisan preferences across time and they employ those preferences to determine their national identity, we might consider that measuring partisan preference can tap people’s long—term commitment toward political parties. 31 CHAPTER 3 METHODOLOGY The best way to understand how a voter acquires or defects from his or her partisan affiliation is to examine that voter’s own attitudes. Therefore, survey research was the best way to answer the research questions posed in this study. In this chapter, the research design, data and measurement, and the research process are described. The research hypotheses are presented at the end of the chapter. Research Design This research was focused on three major issues. First was whether the distribution of partisan preferences among the electorate is stable in Taiwan. Second was the social bases of partisan support in Taiwan. Third was the relationship between partisan preference and issue preference in Taiwan. Party identification is a long-term psychological attachment. To determine whether an instrument captures the theoretical components of party identification, panel data are needed to test it. However, islandwide survey in Taiwan has used this research design. Thus, my strategy was to examine the stability of partisan preferences among the 32 electorate by analyzing several cross-sectional islandwide surveys in Taiwan. Voters from different generations might have had varying political or/and social experiences during their early adulthood, and these experiences might have had a lasting effect on their partisan preferences. Therefore, I examined the stability of partisan preferences among voters of different generations. If the distribution of partisan preferences among voters is stable, one can consider that the measurements of partisan preference might tap people’s long-term commitment to political parties. Election studies have been concerned with the relationship between partisan support and social cleavage (Berelson et al. 1954, Lazarsfeld et al. 1944, Lipset 1981). However, individuals' attitudes also play important roles in their voting choice (Campbell et a1. 1960). Thus, I examined the social bases of party support to show how these sociological factors affect voters’ partisan preferences. I employed survey data to show the relationship between party identification and other social characteristics, such as ethnicity, generation, gender, educational level, occupation, and socioeconomic status, to discover the social bases of partisan support in Taiwan. Also of interest is the relationship between voters’ partisan preference and their national identity. However, it is not easy to distinguish the relationship between these two variables in the non—experimental setting of empirical research. One might suggest that it would be better to employ a simultaneous equation model to examine the relationship between these two variables. However, because the correlation between the instrumental variables and the endogenous variable is very low in this case, it might be better to use an ordinary least squares (OLS) estimator to examine the relationship between these two variables. Therefore, I employed OLS to demonstrate the determinants of partisan 33 preference. In addition, I examined whether partisan preference affected national identity. I also discuss advantages and disadvantages of using an OLS estimator in this research, and whether OLS might bias the research findings. Data and Measurement To shed light on the formation of party identification among the electorate in Taiwan, I employed data four cross-sectional surveys. Because the DPP was established and began to challenge the KMT in the 1986 elections, I employed data from four surveys conducted after that year’s legislative elections. Another reason for using these four surveys is that the legislative elections are national elections, so I could examine the trend of partisan support as well as changes in and continuity of partisan preferences among voters with different social characteristics during last decade. The Electoral Behavior Research Group at National Taiwan University (NTU) conducted the above-mentioned four surveys between 1987 to 1996. These surveys were conducted in 1987, 1990, 1993, and 1996 after the legislative elections, and all of them were islandwide face-to-face surveys. The numbers of respondents in each survey were 1,430 (1987), 1,310 (1990), 1,398 (1993), and 1,383 (1996). All respondents were at least 20 years old and were eligible to vote in the legislative elections. The samples for these surveys were selected from official voter registers or official household registration data. From those official records, research staff could find the name, gender, age, and address of all eligible respondents. Multi-stage stratified sampling was used. The population was divided into four strata: Taipei City, Kaohsiung City, Provincial Cities, and Counties. 34 Based on the proportion of eligible voters in the legislative election from each stratum, quotas for each stratum were decided and then selected from city/county, district/town (or village), and precinct using probabilities proportional to size (PPS) sampling". Because surveys were conducted at least six months after the legislative elections, it seemed unlikely that voters employed their vote choices as their partisan preferences. At least two problems are inherent in the application of party identification in Taiwan. One is that respondents might not have wanted to indicate their partisan preference because it is a sensitive matter to state a Tangwai/DPP preference during the 19808 in Taiwan. The other problem is that respondents might have confused party membership with subjective identity (Liu 1987). Therefore, we can expect the proportion of DPP identifiers to be fewer than its votes during the 1980s. The NTU research team used different question wordings to measure partisan preference. In 1987, the wording was: "Between the KMT and the DPP, which one do you prefer?" (Party 1D_A measurement) Response choices ranged from "strongly prefer the DPP" to "strongly prefer the KMT." On the same survey, respondents were asked the extent to which they liked the KMT and the DPP (like/dislike measurement). In 1990, the survey only asked respondents whether they liked/disliked the KMT and the DPP. The NT U research team changed these questions in the 1993. Respondents were asked: "In our country, there are three major parties: the KMT, the DPP, and the Chinese Social Democratic Party (CSD), and they have different political ideologies and positions. Among these three parties, which one’s ideology and positions is close to yours?" (Party ID_B measurement) ‘5 For a detailed description of the sampling process, Hu et al. (1993, Chapter 2). 35 If a respondent chose one of these three parties, the follow-up question was: "How close are you to this party?" The respondent could answer either "very close, close," or "somewhat close." However, if a respondent did not choose any party, there was no follow-up question to elicit his or her partisan tendency. In the same survey, a respondent was also been asked: "In our country, there are two major parties: one is the KMT, the other is the DPP. Which party do you tend to accept emotionally?" (Party ID_C measurement) There was a follow-up question to explore the respondent’s intensity of partisan preference, but no question for Independents. Three types of questions were posed to measure partisan preference in the 1996 survey. The first was a "like/dislike" question for the three major parties separately. The second was Party ID_B measurement: "Among these three parties, which one’s ideology and positions are close to yours?" A follow-up question was asked to measure the intensity of the partisan identifiers, but no follow-up question was put to Independents. Another question was: "These three parties have their own history, ideas, spirit, and supporters. Someone firmly stands aside with one party, and believes himself or herself to be a loyal supporter or partisan of this party. How about you?" (Party ID_D measurement) If the respondent was an identifier, he or she was not asked the intensity of the partisan preference. However, if he or she was an Independent, there was a follow-up question to ask the partisan tendency. Therefore, including the "like/dislike" question, the NTU research team employed at least five different measurements to capture the theoretic components of party identification in the 1987 to 1996 surveys. One possible advantage of employing these data sets is that I could use the different question wordings to examine whether differences in wordings affected people’s responses. However, the NTU research team 36 did not employ all of the different measurements in each survey, so it was hard for me to make comparisons. In this research, therefore, I had to make a compromise and use the most similar measurements across different years to describe the trend in partisan preference. I employed Party ID_A measurement in 1987, combined two like/dislike questions in 1990, used Party ID_C measurement in 1993, and used like/dislike questions in 1996 to present the overall trend in partisan preferences between 1987 and 1996. Demographic variables include the individual ’s ethnicity, political generation, gender, educational level, and occupation. Some attitudinal variables are also important to this research. I included the respondent’s subjective perception of socioeconomic status, democratic values, concerns about “money politics,” and national identity. One variable that needs to be emphasized is political generation. Because distinctive political generations are more likely to be formed in periods of rapid social change (Abramson 1983), I applied some significant events in Taiwan to classify the political generations.7 I divided the electorate into three generations, each of which had its specific experiences and had faced different social conditions. As shown in Figure 3-1, most first-generation and second-generation citizens experienced great social disorder and an economically poor environment. Second-generation voters witnessed the emergence of the opposition movements, and Tangwai began to challenge the KMT in the elections. They also received their elementary education under the KMT's authoritarian control since 1949. As to the third-generation voters, they rode on the economic take-off of Taiwan, and they experienced political openness during their early adulthood. 7 Great events that occurred in Taiwan between 1945 and 1993 are listed in Taiwan in Table A-1, Appendix A. 37 The Research Hypotheses After reviewing research outside the United States, I believe that party identification is a useful cross-nation concept. However, whether voters in Taiwan formulate a party identification is an interesting question. One way to examine whether people formulate a long-term commitment to political parties is to employ panel data to examine the stability of partisan preferences among individual voters. However, as no panel data are available in Taiwan, I examined the stability of partisan preferences among the electorate between 1987 and 1996. When the distribution of partisan preferences among the electorate is stable, these measurements of partisan preference might be considered to capture people’s long-term psychological attachment to political parties. In addition, according to the theory of party identification, people can employ their party identification to determine their political attitude. Therefore, when voters have stable partisan preferences over time, and they might employ those partisan preferences to determine their political attitude, the voters can be considered to have formulated their party identification. I also wanted to explore the social bases of party support in Taiwan, so I examined how different sociological backgrounds might affect voters’ partisan preferences. According to the theory of political socialization, parents transfer their party loyalty to their children, and this maintains the stability of the party system. However, different political generations also experience different political events during their early adulthood, so generational effect also plays an important role in the formation of one's party identification. Because Taiwanese had fewer chance to enter the power circle 38 First-generation Second-generation Third-generation mainlanders mainlanders mainlanders l l r r 1942 1961 First-generation Second-generation Third-generation Taiwanese Taiwanese Taiwanese F irst-generation mainlanders (1942 and before): They were born on mainland China and began their elementary education before 1942. First-generation Taiwanese (1942 and before): They experienced Japanese education, and only a few of them continued to receive education under KMT rule after 1942. Second-generation mainlanders (1943 to 1960): They began their elementary education in Taiwan after 1949, and most of them grew up when Taiwan was economically poor; they also experienced the electoral competition between the DPP/Tangwai and the KMT. Second-generation Taiwanese (1943 to 1960): They began their elementary education under the KMT’s authoritarian control after 1949, and most of them grew up when Taiwan was economically poor; they also experienced the electoral competition between the DPP/Tangwai and the KMT. Third-generation mainlanders (1961 and after): They grew up in an economically prosperous Taiwan and had the right to vote in the competitive elections; their parents might be first-generation mainlanders because some first-generation mainlanders got married late. Third-generation Taiwanese (1964 and after): They grew up in an economically prosperous Taiwan and had the right to vote in the competitive elections. Source: Liu 1995: 91, Figure 1; but the author changed the classification of generations. Figure 3-1. Division of Taiwan’s six Political Generations 39 before the 1990s, it is reasonable to expect that first-generation Taiwanese were less likely to be involved in politics. On the other hand, first-generation mainlanders had a strong attachment to the KMT because they were followers of the KMT leaders and mainlanders held most important positions in the central government before the 1990s. Therefore, first-generation voters were socialized in a one-party dominant regime, so they were either loyal to the ruling party or did not challenge the KMT in order to keep themselves safe. Also, first-generation mainlanders were more likely to support the KMT than were first-generation Taiwanese. As Table 1-2 indicates, the ruling party won more than 70 percent of the votes in legislative elections before 1983. Therefore, without open and competitive elections, the ruling party is more likely to dominate the elections, and it is hard for voters to build a psychological attachment to other parties. Following political openness and electoral competition, second-generation and third-generation Taiwanese found that more and more Taiwanese entered the central government and the KMT power circle, and the DPP offered a different political choice. So they were more likely to get involved in politics. As to mainlanders, their parents, i.e., first-generation mainlanders, followed Chiang Kai-shek to Taiwan, and the parents tended to have higher political involvement. We can thus expect that intergenerational transmission of party identification is more likely to happen in mainlander families (Beck 1974). Therefore, I hypothesized that first- and second-generation mainlanders were more likely to be partisan identifiers than were first- and second-generation Taiwanese. In addition, I hypothesized that third-generation mainlanders would maintain a higher level of partisan preference, but the strength would be weaker than that of their parents because they faced 4O political openness and electoral competition among major parties. Moreover, third- generation Taiwanese were more likely to have partisan preferences because they faced democratic reforms and competitive elections during their early adulthood. I also hypothesized that Taiwanese were more likely to support the DPP, whereas mainlanders were more likely to support the KMT or the NP. Lipset (1980) argued that the dominant class tends to vote for the ruling party. Since the KMT maintains the economic prosperity, I hypothesized that people with higher socioeconomic status were more likely to support the KMT. In addition, the KMT is still the ruling party, so state employees are more likely to support it. Abramson and his colleagues (1998) found that males were more likely than females to support Republican presidential candidates in the 1980, 1984, and 1988 presidential elections. In 1996, Clinton gained 54 percent of white females’ votes, whereas he received only 42 percent of white males’ votes (Abramson et al. 1998). In Taiwan, females face more cultural and traditional hurdles than females in advanced industrial countries, in term of their political participation. Their access to political activities and political information is limited, and it makes their political views more conservative than those of males. Therefore, I hypothesized that females were more likely to support the KMT because they wanted to maintain the status quo. I also examined the relationship between party identification and other attitudinal variables. Because one of the major platforms of the DPP was adopting more rapid democratic reforms, I hypothesized that people preferring democratic reforms were more likely to support the DPP. Another key issue, unification with mainland China versus Taiwan independence, was also critical in understanding the formation of party loyalty in 41 Taiwan. As mentioned above, the DPP promoted Taiwan’s independence, the NP preferred unification with mainland China, and the KMT took a more ambiguous stance on this issue during the 19908. Because the KMT used to take a pro-unification stance on this issue, it was of interest to observe how this issue might have changed the social base of support of the KMT. I hypothesized that supporters of the NP were more likely to have Chinese identity, but that the DPP’s partisans had Taiwanese identity. In addition, the KMT attracted fewer people with Chinese identity because it adjusted its stance to the middle on this issue. Because the KMT adopted a position similar to the DPP on the national identity issue, I hypothesized that national identity became less important in distinguishing between these two parties. Another important issue is people’s concern about "money politics." During the late 1980s and early 19908, skyrocketing real estate and stock prices heightened people’s concerns about the fair distribution of wealth. President Lee had to seek support of the business community and local factions to win the power struggle with the so-called nonmainstream mainlander elite within the KMT. Therefore, a new cleavage in the socioeconomic justice issue, i.e., concerns about money politics, was salient to the electorate in Taiwan (Lin et al. 1998). Thus, I predicted that people who were concerned about "money politics" were less likely to prefer the KMT. The framework of this research is illustrated in Figure 3-2. Political generation was a mediator variable in this study. In Chapter 4, the stability of partisan preferences among three political generations and two major ethnic groups is examined. In Chapter 5, I discuss the social bases of party support and explore how social forces such as gender, education, vocation, and subjective SES affect people’s partisan preferences. I also 42 Subjective SES Concern About "Money Politics" National Identity Partisan Preference —> +__ Democratic Values Figure 3-2. The Relationship Between Partisan Preference and Other Variables. examine the relationship between partisan preference and national identity in Chapters 6 and 7. 43 CHAPTER 4 GENERATIONS, ETHNIC GROUPS, AND PARTY IDENTIFICATION Party identification has several characteristics. First of all, party identification is highly stable. Second, party identification develops early in life. Third, it increases in strength during one’s lifetime8 (Campbell 1960). However, an individual’s party identification is relatively stable compared to many other political attitudes, although it is also mutable (Fiorina 1981; Markus and Converse 1979; Jennings and Niemi 1978; Page and Jones 1979). In this chapter, I examine the stability of party identification among the electorate in Taiwan. As mentioned in previous chapters, voters in Taiwan have experienced great social and political change during the last twenty years. Because traumatic events might influence the formative socialization of electors who entered the electorate during different historical periods (Abramson 1989), I examined generational differences in their party identification. In addition, different ethnic groups have had varying political- socialization experiences, and those experiences might also affect their party identification. Thus, I examined differences between Taiwanese and mainlanders in term of their party identification. 3 Some controversy exists about the presence of life-cycle effects in the development of partisan strength. For a summary of this debate, see Abramson (1989). 44 Generational Effect, Life-cycle Effect, and Party Identification Jennings and Niemi (1978) presented four basic models of political socialization. The first one is the life-long persistence model, which indicates that "What is learned early is asserted to endure; resistance to change is asserted to increase as the individual becomes accustomed to and comfortable with a given set of orientations" (Jennings and N iemi 1978: 334). The second model is "the life-long openness model. This model assumes the existence of few or no residues from pre-adult learning" (Jennings and Niemi 1978: 334). The third model, the life-cycle model, "holds that, while persistence is the rule, certain orientations are very amenable to alteration at given life stages" (Jennings and Niemi 1978: 334). The last model, the generational model, "also posits strong persistence in general, but it allows for considerable new socialization and/or resocialization with lasting effects during the formative, impressionable years" (Jennings and Niemi 1978: 335). These models focus on whether people’s attitudes will change or be persistent after their pre-adult or/and early-adult leanings. Compared to other political attitudes, party identification is relatively stable (Converse 1964; Converse and Markus 1979; Jennings and Niemi 1978), so it was of interest to know how people develop their partisan preferences, whether they change their partisan preferences, and why they change their partisan preferences. As Campbell and his colleagues (1960: 161) wrote, Whenever we encounter relationships of this sort involving age, we are forced to choose between two competing interpretations. First, age may mark an historical epoch in which the person has matured or undergone some special variety of experiences that has left an imprint on his attitudes and behaviors. ...The second interpretation presumes instead that older people will always feel stronger bonds with a political party than will newer members of the electorate. 45 Therefore, generational and/or life-cycle effect can play an important role in developing of party identification. 9 Several problems arise when cohort analysis is applied to explain the change in or continuity of partisan preferences among the electorate (Abramson 1989: 551-553; Glenn 1977:12-17). First, the cells of a cohort table contain sampling errors. Therefore, if one wants to make inferences from cohort tables, sampling errors must be considered. Second, cohort analysts assume that compositions within cohorts remain unchanged. Therefore, one must ensure that the compositions of the electorate are relatively constant. Most mainlanders followed Chiang Kai-shek to Taiwan in 1949, and there have been no mass immigrations or emigrations since then. Therefore, the population in Taiwan is relatively closed. Third, it is not easy to distinguish basic effects (generational, life-cycle, and period) from one another. Campbell and his colleagues argued that partisan strength increases with age. They asserted that "group identification is a function of the proportion of a person’s life he has been associated with the group. The longer a person thinks of himself as belonging to a party, the stronger his sense of loyalty to it will become" (Campbell et al. 1960:161). Converse (1969, 1970) also believed that the individual’s partisan affiliation increases with age (See also Shively 1979a, 1979b, 1979c). On the other hand, from the student- parent panel studies, Jennings and Markus (1984: 1015) concluded that The high school class of 1965 ...began their adult political lives strikingly less committed to political parties than were their parents, and they have remained less committed for nearly two decades. This result is most sensibly interpreted as a 9 The generational or cohort effect "is produced by influences associated with birth cohort membership." The life-cycle or aging effect is produced by influences associated with the process of growing older. And the period effect is "produced by influences associated with each period of time" (Glenn 1977: 11). 46 generational distinction, one that arose from the unique sociopolitical milieu surrounding the Vietnam-era cohort’s entry into adulthood and one that is likely to persist, although with diminishing sharpness, with passage of time. After examining SRC—CPS National Election Studies (1952-1980), Gallup data (1945- 1971), and the Michigan student-parent data (1965 and 1973), Abramson (1983:115) concluded that "all [these data sets] provide compelling evidence that partisan strength has not increased with age." He argued that generational effect rather than life-cycle effect is a better explanation for the development of party identification. (See also Abramson 1979a, 1979b; Claggett 1981; Glenn 1972, 1977; Glen and Hefner 1972; Jennings and Niemi 1975, 1978, 1981; Inglehart and Hochstein 1987; Markus 1983; Norpoth and Rusk 1982.) Further, Abramson (1989) thought that generational replacement has contributed to the decline of partisan strength in the American electorate. 10 Besides the generational effect, I also examined differences between ethnic groups in term of their party identification. As mentioned earlier, minority mainlanders have controlled most central officers in Taiwan since 1949, so mainlanders were expected to have had a higher level of political efficacy and political involvement than Taiwanese. Because the family is an important agent for political socialization, it was expected that parents from different ethnic groups would exercise different levels of influence on their offspring. Niemi and Jennings (1991) showed that parental party identification exercised a great impact on the offspring’s party identification during their high school years. As these offspring reached their mid-208, the effect of parental party identification was still significant, but its importance decreased. At the same time, the ‘° For the case of Sweden, see Holmberg (1994). 47 offspring’s own issue stances began to influence the offspring’s party identification. When the offspring reached their mid-308, their parents’ party identification still had the same degree of influence on their party identification, but the offspring’s own issue stances increased in importance in shaping the offspring’s party identification. Therefore, parents’ party identification may be one of the resources through which offspring gain their party loyalties, but still other factors affect the genesis and duration of individuals’ party identification. Because the older generation in Taiwan experienced authoritarian control whereas the younger generation experienced political openness and party competition, I examined how these experiences affected individuals’ partisan preferences. An effective way to examine the stability of partisan preferences of individual voters is to employ three-way panel data to see whether individual voters’ partisan preferences are stable. However, no panel data were available in Taiwan, so I examined the distribution of partisan preferences at the aggregate level to see whether the electorate had stable partisan preferences. The Stability of Party Identification Among the Electorate in Taiwan Compared to other political attitudes, party identification is relatively stable. I examined the distribution of party identification among the electorate in Taiwan between 1987 and 1996 to see whether it was stable. I employed data from surveys conducted by the Electoral Behavior Research Group at the National Taiwan University. 11 The ” For the sampling process and case numbers of these surveys, please see Chapter 48 question in 1987 was: "Between the KMT and the DPP, which one do you prefer?" (Party ID_A measurement). Respondent were asked to specify "strongly prefer the DPP" to "strongly prefer the KMT." In 1990 and 1996, ’2 the survey asked respondents only whether they liked/disliked the KMT and the DPP. In the 1993 survey, respondents were asked, "In our countries, there are two major parties: one is the KMT and the other is the DPP. Which party do you tend to accept emotionally?" (Party ID_C measurement). There was a follow-up question to explore respondents’ intensity of partisan preference, but there was no follow-up question for the Independents. These questions were very similar, although they might have captured both long-term psychological attachment and short-term electoral effect. Because no other longitudinal and islandwide survey data sets were available for this period,13 these data sets were the most appropriate ones with which to test the hypotheses for this study. These variables were coded into three categories--prefenin g the KMT, preferring the DPP, and indifference--to examine voters’ partisan preferences toward two major parties. If the aggregate level of party identification is found to stable, it might indicate that these measures captured the electorate’s long-term commitment to those parties. '2 In the 1996 survey, respondents also were asked to express whether they liked/disliked the NP. ‘3 The Election Study Center (ESC) at National Chengchi University (NCCU) conducted face-to-face surveys after the legislative elections in 1986, 1989, 1992, and 1995. However, on two of these surveys, only voters in Taipei Municipality and Taipei County were interviewed. 49 The distribution of partisan preference among the electorate between 1987 and 1996 is shown in Figure 4-1 (see also Table B-1 in Appendix B). There was a decline in the proportion of KMT partisans between 1987 and 1996. The percentage of KMT partisans varied from 31.8 percent to 47.3 percent, and the range was 15.5 percent. More than four out of ten voters were KMT partisans between 1987 and 1990. The percentage declined to 38.3 percent in 1993 and continued to decline until 1996, at which time just under one voter in three preferred the KMT. Martial law was lifted in 1987, and the KMT faced challenges from opposition parties thereafter. Hence, the proportion of KMT partisans has declined dramatically in the last decade. This issue is discussed further in the next section. From Figure 4-1, one can see that the distribution of DPP partisans was more stable. In addition, there was an increase in DPP partisans between 1987 and 1996. The proportion of DPP partisans varied from 8.2 percent to 15.5 percent, and the range was 7.3 percent. The proportion of DPP partisans increased by about 0.7 percent every year between 1987 and 1996. As mentioned in Chapter 1, the proportion of the DPP’s vote in the legislative elections increased monotonically between 1987 and 1995. Therefore, by constructing affective ties with voters, parties are able to increase their votes in the elections. As Figure 4-1 indicates, the relative strength of partisan forces of the two major parties changed in the last decade. The ratio of KMT partisans to DPP partisans was 5.4 to 1 in 1987, 4.6 to 1 in 1990, and 3 to 1 in 1993; it declined still further, to 2 to 1, in 1996. The proportion of KMT partisans declined dramatically between 1987 and 1996, but the increase in DPP partisans was relatively slow. Therefore, the proportion of KMT 50 55 - 52.7 47.2 20 r 15.5 12.9 15 r 10.2 1987 1990 1993 1996 -+— Prefer KMT —I— Prefer DPP + Incifferent Source: Surveys conducted by the Electoral Behavior Research Group at the National Taiwan University (NTU). Note: For the numbers of cases on which these percentages are based, please see Table B-1, Appendix B. Figure 4-1. Distribution of Partisan Preferences: 1987 to 1996. partisans was still 16 percent greater than the proportion of DPP partisans in 1996. The social bases of partisan support were the focus of this research. One might argue that, as shown in Figure 4-1, about half of the total sample was dropped out in this study because only partisans were examined. However, this research focused on the formation of partisan preferences among the electorate, so it was inevitable that partisans were my research interest. One can cite many examples to show that eliminating cases from the research is not a problem when one intends to make valid inferences from his or her research. It is well known that overall voter turnout declined after 1960 in the United States. When one focuses on explaining voters’ choices in the United States after 1960, one has to dropped half or more of the cases from the total samples in his or her data 51 analysis. However, it will not damage the validity of the research because one is interested in making inferences regarding voters’ behavior. Therefore, it seems reasonable to focus on differences among partisans in various social groups when one wants to examine the social bases of party support. From Figure 4-1, It can be seen that the proportion of KMT supporters declined between 1987 and 1996, whereas the proportion preferring the DPP increased. However, I am not sure whether the KMT lost its support from all generations or whether voters in the younger generation became less likely to support the KMT. By the same token, I also wanted to examine which generation became more likely to support the DPP. Because different generations experienced different historical events, these experiences might have had a long-term effect on people’s partisan preferences. In addition, I wanted to explore whether or not partisan attitudes among different generations in Taiwan were stable. Political Generation and Partisan Preference The concept of generation can be viewed as both biological and sociological (Mannheim [1928] 1952). Generations are produced through the "biological rhythm in human existence," but "individuals who belong to the same generation, who share the same year of birth, are endowed, to that extent, with a common location in the historical dimension of the social process" (Mannheim [1928] 1952: 290). Manheim argued that historical experiences are most likely to influence persons during their late adolescence and early adulthood. 52 Taiwan presents a good setting in which to examine the generational effect on people’s partisan preferences for several reasons. First, during the past twenty years, Taiwan’s party system has been substantially transformed. It was a one-party-dominant system before 1987, a two-party system between 1987 and 1993, and a three-party system since 1993. Second, Taiwan has experienced dramatic political changes and democratic reform that might have influenced the formative socialization of citizens who entered the electorate during different historical periods. Third, different ethnic groups might have had different socialization experiences that led to differences in their partisan preferences. I defined political generations according to great social or political events, and divided the electorate into three different generations. The first generation was born before 1943, the second generation was born between 1943 and 1960, and the third generation was born after 1960.14 Voters in the first generation lived under the KMT’s authoritarian control; thus they can be expected to have a positive affective attitude toward the KMT or to be less likely to be involved in politics. Voters in the second generation faced diplomatic setbacks and dramatic social and political changes, such as Taiwan’s loss of its seat in the United Nations in 1971, the first legislative election in 1972, the Chungli Incident in 1977, and the United States’ building of a formal diplomatic relationship with the People’s Republic of China (PRC) in 1978. Thus they have begun to support the Tangwai/DPP candidates in the elections. Voters in the third generation experienced the emergence of two new parties in 1986 and 1993, during their “ For 1987, the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. For great events that happened between 1945 and 1993 in Taiwan, see Appendix A, Table A-1. 53 late adolescence and early adulthood, so their partisan preferences can be expected to be changeable and volatile. First, I examined the partisan preferences of these three generations toward the KMT between 1987 and 1996. As shown in Figure 4-2, between 1987 and 1996, partisan preferences toward the KMT held by first-generation (born in 1942 and before) and second-generation (born between 1943 and 1960) voters were more stable than those of third-generation (born after 1960) voters. A8 to first-generation voters, the percentage of KMT partisans varied from 34.9 percent to 46.4 percent; the range was 11.5 percent. First-generation voters became more likely to support the KMT over time; four out of nine first-generation voters preferred the KMT in 1996. The distribution of KMT partisans in the second generation was less stable than that in the first generation. It varied from 28.4 percent to 45.8 percent; the range was 17.4 percent. It seems that, since 1990, second-generation voters have been less likely to prefer the KMT than voters in the other two generations. The prOportion of KMT partisans in the third generation varied from 29.1 percent to 58.2 percent; it was the most volatile among the three generations. Almost six out of ten third—generation voters preferred the KMT in 1987, and this group was the most loyal to the KMT at that time. However, its loyalty toward the KMT declined monotonically between 1987 and 1996, and fewer than three out of ten third- generation voters preferred the KMT in 1996.15 As mentioned above, there was an increase in DPP partisans between 1987 and 1996. One can examine generational differences in preferences toward the DPP. '5 I also examined the cohort born after 1962. This cohort experienced the establishment of the DPP and voted after the ending of martial law. However, its partisan preference was similar to that of the third generation in my research. 54 6O 1 55 + 50 - 45 ‘ 43.8 40 —« 35 1 29.1 30 i 1987 1990 1993 1996 + 1961 andafter + 1943 to 1960 + 1942 andbefore Source: Surveys conducted by the Electoral Behavior Research Group at the National Taiwan University (NTU). Note: For the numbers of cases upon which these percentage are based, see Table B-2, Appendix B. For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. Figure 4-2. Percentages of KMT Partisans: 1987 to 1996. However, I must point out that there were too few first-generation DPP partisans to make confident inferences in 1987, 1990, and 1996 (see Table B-2, Appendix B). Therefore, one must be cautious when making inferences about DPP partisans among first- generation voters. As shown in Figure 4—3, partisan preference toward the DPP was more volatile in the third generation between 1987 and 1996. The proportion of DPP partisans varied from 9 percent to 18.6 percent; the range was 9.6. The proportion of DPP partisans in the second generation started at 9.6 percent in 1987, about 12 percent second- generation voters preferred the DPP in 1990, and about 15 percent of second-generation voters preferred the DPP between 1993 and 1996. The range of DPP partisans in the second generation was 5.4 percent between 1987 and 1996. The first generation had the 55 most stable partisan preference toward the DPP, and first-generation voters were less likely to support the DPP than voters in the two younger generations between 1987 and 1996. The proportion of DPP partisans in this generation varied from 5.5 percent to 10.8 percent; the range was 5.3 percent. Therefore, the increase in DPP partisans resulted from the support of voters in the two younger generations. Examining Figures 4-2 and 4-3, one can see that partisan preferences for the two major parties among different generations changed between 1987 and 1996. Let us examine the KMT ’8 supporters first. In 1987, a greater proportion of third-generation voters as compared to first- and second— generation voters preferred the KMT. Almost six out of ten third-generation voters preferred the KMT. At the same time, only 34.9 percent 20 — 18.6 15.0 /4’5 15 ‘ 12.3 A 10 — ' 10.8 5 _. 5.5 O F I I l 1987 1990 1993 1996 +1961 andafter +1943 to 1960 +1942 andbefore Source: Surveys conducted by the Electoral Behavior Research Group at the National Taiwan University (NTU). Note: For the numbers of cases upon which these percentage are based, see Table B-2, Appendix B. For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. Figure 4-3. Percentages of DPP Partisans: 1987 to 1996. 56 of first-generation voters preferred the KMT. However, this situation was reversed in 1996. First-generation voters became the group most loyal to the KMT, whereas since 1990 the two younger generations became less likely to prefer the KMT. Between 1987 and 1996, the proportion of third-generation voters preferring the KMT dropped by 29.1 percent, but the percentage of first-generation voters preferring the KMT increased by 8.9 percent. Therefore, third-generation voters contributed to the decline of KMT partisans between 1987 and 1996. Electors in the second generation also have become less likely to support the KMT since 1990; the proportion of second-generation voters preferring the KMT declined by 17.4 percent between 1990 and 1996. On the other hand, first- generation voters evidenced relatively stable support for the KMT between 1990 and 1996. The distribution of DPP partisans was relatively more stable than the distribution of KMT partisans. However, because there were too few cases in some cells, one can not make confident inferences about the distribution of DPP partisans in the first generation (see Table B-2, Appendix B). However, as shown in Figure 4-3, the older the generation was, the less likely it was to be attracted by the DPP in 1996. Therefore, the distributions of partisan preferences among the two older generations were more stable than those of the youngest one. Voters in the youngest generation were more mutable in their partisan preferences than were voters in the two older generations, and they were more likely to be attracted by the DPP. Therefore, older citizens’ party preferences were relatively stable. Although these citizens faced the emergence of the DPP and competitive elections, their loyalty to the KMT was relatively 57 stable. On the other hand, as a new party, the DPP has attracted more young citizens in the last decade. However, as mentioned above, different ethnic groups might have had different political and social experiences during the formative socialization of their early adulthood. That issue is discussed in the following section. Ethnicity and Partisan Preference As mentioned in Chapter 3, mainlanders and Taiwanese experienced different political-socialization processes during their early adulthood. Generally speaking, most first-generation mainlanders followed Chiang Chia-shek to Taiwan in 1949, so they were more likely to support the ruling party. In addition, most first-generation mainlanders experienced the Chinese Civil War, so their positive attitude toward the KMT was relatively strong and they were more likely than Taiwanese to transmit their strong partisan preference to their offspring. Therefore, mainlanders were more likely to support the KMT (Chen 1992; Chu 1992; Chu and Lin 1996; Lin 1988; Lin et al. 1998; Shyu 1991, 1993; Tsai 1997). On the other hand, first-generation Taiwanese experienced Japanese control and/or authoritarian political control under the KMT during their early adulthood, so they were either loyal to the KMT or became less likely to formulate partisan preferences toward political parties. However, second- and third-generation Taiwanese experienced party competition and free elections during their late adolescence and/or early adulthood, so they were more likely than first-generation voters to formulate preferences for opposition parties. 58 The partisan preferences of Taiwanese were examined first. As shown in Figure 4-4, partisan loyalty toward the KMT among third-generation Taiwanese was the least stable among the three generations of Taiwanese between 1987 and 1996, but there was little fluctuation in the proportions of KMT partisans among the two older generations. For example, 55 percent of third-generation Taiwanese preferred the KMT in 1987, but only 26.1 percent of them preferred the KMT in 1996. A8 to second-generation Taiwanese, the proportion of KMT partisans was 36.9 percent in 1987, rebounded to 42.3 percent in 1990, decreased to 29.4 percent in 1993, and continued to decline, to 25.7 percent, in 1996. First-generation Taiwanese had the smallest proportion of KMT partisans among the three generations in 1987; the proportion varied from 27.5 percent to m _ 55.0 55 - 50 a 46.9 45 q 39 3 4o 9 36.9 42.3 37.8 35 l 35 o 29.9 30 j - ‘61 25 - 29.4 27.5 25.7 20 f —T I l 1987 1990 1993 1996 + 1961 andafter + 1943 to 1960 + 1942 andbefore Source: Surveys conducted by the Electoral Behavior Research Group at the National Taiwan University (NTU). Note: For the numbers of cases upon which these percentage are based, see Table B-3, Appendix B. For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. Figure 4-4. Percentages of KMT Partisans Among Taiwanese: 1987 to 1996. 59 39.3 percent between 1987 and 1996 with a range of 11.8 percent. However, in 1987, third-generation Taiwanese comprised the most loyal group of KMT partisans, whereas first-generation Taiwanese became the most loyal group in 1996. I also examined the distributions of DPP partisans among the three Taiwanese generations between 1987 and 1996. Because there were fewer than 25 cases for first- generation Taiwanese in 1987, 1990, and 1996, I focused on second- and the third- generation Taiwanese. As shown in Figure 4-5, support for the DPP among the three Taiwanese generations increased between 1987 and 1996. The pr0portion of DPP supporters among third-generation Taiwanese varied from 10.5 percent to 19.7 percent; the range was 9.2 percent between 1987 and 1996. As to second-generation Taiwanese, 22 - 19.7 1987 1990 1993 1996 —O— 1961 and after + 1943 to 1960 + 1942 andbefore Source: Surveys conducted by the Electoral Behavior Research Group at the National Taiwan University (NTU). Note: For the numbers of cases upon which these percentage are based, see Table B- 3, Appendix B. For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. Figure 4-5. Percentages of DPP Partisans Among Taiwanese: 1987 to 1996. 60 the proportion of DPP partisans varied from 10.2 percent to 16.8 percent; the range was 6.6 percent. Therefore, the older generation tended to have relatively less fluctuation in its partisan preference for the DPP than did the younger generation between 1987 and 1997. However, the distribution of partisan preferences among mainlanders was another story. Comparing Figure 4-6 with Figure 4-4, one finds that mainlanders were more likely than Taiwanese to support the KMT, and first-generation mainlanders were the most loyal group of the three mainlander generations in its support of the KMT between 1987 and 1996. As shown in Figure 4-6, almost eight out of ten first-generation mainlander voters preferred the KMT in 1987, and more than one out of two first- generation mainlander voters still preferred the KMT in 1996. The proportion of KMT partisans among second-generation mainlanders varied from 48 percent to 73.3 percent; the range was 25.3 percent. The proportion of KMT partisans among third-generation mainlanders varied from 42.7 percent to 76.7 percent; the range was 34 percent. Therefore, the partisan preference of second-generation mainlander was the most stable, but first-generation mainlanders were most likely to support the KMT. The latter group’s social and political experiences during their early adulthood had a long-term effect on their partisan preference. Their strong and positive attitude toward the KMT also affected their offspring’s partisan preference, so younger-generation mainlanders were also more likely to support the KMT than were younger-generation Taiwanese. 61 95 ’ 88.1 85" 75* 65 r 55‘ 48.0 45 r 42.7 35 n r r 1 1987 1990 1993 1996 + 1961 andafter + 1943 to 1960 + 1942 and before Source: Surveys conducted by the Electoral Behavior Research Group at the National Taiwan University (NTU). Note: For the numbers of cases upon which these percentage are based, see Table B-4, Appendix B. For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. Figure 4-6. Percentages of KMT Partisans Among Mainlanders: 1987 to 1996. Fewer mainlanders than Taiwanese preferred the DPP between 1987 and 1996 (see Figure 4-7). However, there were fewer than 25 cases for each observation, so one cannot make confident inferences from the figure (see Table B-4 in Appendix for details). Overall, less than 5 percent of mainlanders supported the DPP between 1987 and 1996. In addition, there was an increase in DPP partisans among third-generation mainlanders between 1987 and 1996. Among mainlanders, first-generation voters had less fluctuation in their support of the DPP than did the two younger generations of voters. Because the two major ethnic groups-~mainlanders and Taiwanese-- had different experiences during formative socialization in their late adolescence and/or early 62 16 i 14 r 12.4 12 i 10 . 8 - 6.5 8.0 6 r 4.4 4 - 2.7 4-2 3.1 1.8 2 ~ 2.1 0.0 1,7 0.0 0 v r r r H 1987 1990 1993 1996 —o— 1961 and after —I-— 1943 to 1960 + 1942 andbefore Source: Surveys conducted by the Electoral Behavior Research Group at the National Taiwan University (NTU). Note: For the numbers of cases upon which these percentage are based, see Table B-4, Appendix B. For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. Figure 4-7. Percentages of DPP Partisans Among Mainlanders: 1987 to 1996. adulthood, their differences in their partisan preferences were compared. First, it was found that mainlanders were more likely than Taiwanese to support the KMT (see Table 4-1). Overall, 30 percent more mainlanders than Taiwanese preferred the KMT between 1987 and 1996. As Table 4-1 indicates, within the first generation, the two ethnic groups differed greatly in their support for the KMT; that difference exceeded 50 percent in 1987 and 1990. The difference between the two ethnic groups among third-generation voters was smaller than in the other two generations, showing that third-generation voters tended to have a more homogeneous political experience, whereas the older generations of voters maintained their memories of early adulthood. Therefore, there was a generational effect on all three generations of the two ethnic groups. 63 Table 4-1. Percentages of KMT Partisansl Among two Ethnic Groups, by Generation, 1987 to 1996 Generation Ethnic Group 1987 2 1990 1993 1996 Range Mean 1961 and After Taiwanese 55.0 46.9 37.8 26.1 28.9 41.5 Mainlander 76.7 63.6 64.9 42.7 34.0 62.0 Difference 3 -21.7 -16.7 -27.1 -16.6 10.5 -20.5 1943 to 1960 Taiwanese 36.9 42.3 29.4 25.7 16.6 33.6 Mainlander 73.3 71.7 62.5 480* 25.3 63.9 Difference 3 -36.4 -29.4 -33.1 -22.3 14.1 -30.3 1942 and Before Taiwanese 27.5 35.0 29.9 39.3 11.8 32.9 Mainlander 78.1 88.1 73.2 54.9 33.2 73.6 Difference 3 -50.6 -53.1 -43.3 - 15.6 37.5 -40.7 Total Taiwanese 39.9 42.0 32.4 28.3 13.7 35.5 Mainlander 76.3 75.0 67.4 48.3 28.0 65.7 Difference 3 -36.4 -33.0 -35.0 -20.0 -14.3 -30.2 1 For the wording of questions in each survey, see Table B-1, Appendix B and Chapter 3. 2 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 3 Difference = (% of KMT partisans among Taiwanese - % of KMT partisans among mainlanders). *N<25. People of different ethnic origins had different partisan preferences. In Table 4-1, it can be seen that first-generation mainlanders were more loyal to the KMT than were the other five groups between 1987 and 1996. Therefore, their experience, such as the China Civil War in 1949, had a long-lasting effect on their partisan loyalty. On the other hand, first- generation Taiwanese were less likely to prefer the KMT between 1987 and 1990. Some political events, such as the February 28 (228) Incident in 1947, might have had a long- term effect on this generation’s partisan preference. Third-generation mainlanders registered higher level of partisanship than third-generation Taiwanese did, perhaps because young mainlanders learned their partisan preferences from their parents, and most of their parents had higher levels of partisan preference. On the other hand, third- generation Taiwanese entered the electorate during an era of party and electoral competition, so their partisan preference was stronger than that of the two other Taiwanese generations between 1987 and 1993. When the NP was established in 1993, mainlanders’ loyalty to the KMT changed. Between 1993 and 1996, mainlanders in the three generations declined about 20 percent in their support of the KMT. However, of the Taiwanese, only those in third generation became less likely to support the KMT in 1996. Therefore, when the NP emerged it seemed to attract mostly mainlanders and third-generation Taiwanese. This issue is discussed in the following section. 65 The Emergence of the New Party Because the NP was established in 1993, we need to consider how it affected the distribution of partisan preferences among the electorate. The NTU research team employed three questions to explore respondents’ preferences for the KMT, the DPP, and the NP in 1996. I constructed an indicator by employing these three questions to classify voters into four categories: preferring the KMT, preferring the DPP, preferring the NP, and indifference. In addition, I compared this indicator with my previous measurements to examine how the emergence of the NP affected people’s partisan preferences. The results are shown in Table 4-2. At the bottom part of the Table, one can see that voters’ preferences for the DPP were relatively stable. When voters made comparisons between two major parties, i.e., the KMT and the DPP, 15.5 percent of them preferred the DPP. However, when voters had to choose among three major parties, 12 percent of them preferred the DPP. On the other hand, the emergence of the NP made some KMT supporters defect from their party. One out of three KMT partisans defected from the KMT when they were presented with a new alternative in the party system in 1996. As shown at the bottom of Table 4-2, the KMT still received more support than the other two major parties, but the margin was relatively small in 1996. When examining the differences among the three generations in Table 4-2, one finds that DPP partisans were less likely to change their partisan preferences when they had to make a choice among the three major parties. However, support for the KMT declined when the NP entered the party system, and third-generation voters were most sensitive to that change. As shown at the top of Table 4-2, third-generation voters 66 Table. 4-2 Two Measurements of Partisan Preference, by Generation, 1996 Generation Partisan Two Major Three Major Difference: Preference Parties 1 Parties 2 2"" - 1“ 1961 and After Prefer KMT 29.1 13.1 -16.0 Prefer DDP 18.6 13.1 -5.5 Prefer NP -- 14.3 14.3 Indifferent 52.3 59.5 7.2 (N) (533) (511) 1943 to 1960 Prefer KMT 28.4 19.2 -9.2 Prefer DDP 14.5 11.2 -3.3 Prefer NP --- 9.1 9.1 Indifferent 5 7. 1 60.4 3.3 (N) (455) (427) 1942 and Before Prefer KMT 43.8 32.7 -11.1 Prefer DDP 10.8 9.5 * -1.3 Prefer NP --- 10.5* 10.5 Indifferent 45.4 47.3 1.9 (N) (251) (162) All Sample Prefer KMT 31.8 19.1 -12.7 Prefer DDP 15.5 11.7 -3.8 Prefer NP -- 1 1.6 11.6 Indifferent 52.7 57.6 4.9 (N) (1,239) (1,158) 1 Items for 1996 were worded as like/dislike questions for two major parties. 2 Items for 1996 were worded as likejdislike questions for three major parties. *N<25. 67 supported the three major parties equally in 1996. When they were given the third option in the party system, 50 percent of the KMT partisans defected from that party in 1996. Voters in the two older generations also changed their partisan preferences when they had to make a choice among the three major parties. About one out of four KMT partisans in the two older generations defected from the KMT. I explored this issue further by analyzing ethnic differences in these voters’ support of the NP. Table 4-3 indicates partisan preferences among Taiwanese by comparing the results of two different measurements in 1996. At the bottom of the table, one can see that Taiwanese voters’ preferences for the DPP were relatively stable between the two measurements. Therefore, the emergence of the NP was less likely to affect the DPP’s strength among Taiwanese electors. When Taiwanese voters faced three major parties instead of two, about one out of four KMT partisans defected from that party in 1996. However, different generations showed some differences in their preferences toward the three major parties. Third-generation Taiwanese voters were more likely to defect from the KMT than were voters in the two older generations. More than one out of three KMT partisans among third-generation Taiwanese defected from the KMT in 1996. As the age of their generation increased, Taiwanese became less likely to defect from the KMT. Support for the KMT among first-generation Taiwanese was almost unaffected by the emergence of the NP. Less than 1 percent of first-generation Taiwanese preferred the NP in 1996. Therefore, partisan preferences were relatively stable among older Taiwanese, but younger Taiwanese were more likely to be attracted by newly established parties. However, when partisan preferences among mainlanders were examined, the findings were different. As seen at the bottom of Table 4-4, more than two out of three 68 Table 4-3. Two Measurements of Partisan Preference Among Taiwanese, by Generation, 1996 Generation Partisan Two Major Three Major Difference: Preference Parties 1 Parties 2 2"" - 1" 1961 and After Prefer KMT 26.1 13.5 -12.6 Prefer DDP 19.7 15.1 -4.6 Prefer NP -- 10.1 10.1 Indifferent 54.2 61.3 7.1 (N) (436) (416) 1943 to 1960 Prefer KMT 25.7 19.2 -6.5 Prefer DDP 15.6 12.2 -3.4 Prefer NP -- 5.4* 5.4 Indifferent 58.7 63. 1 4.4 (N) (397) (369) 1942 and Before Prefer KMT 39.3 33.3 -6.0 Prefer DDP 13.5* 13.1* -0.4 Prefer NP -- 0.7* 0.7 Indifferent 47.2 52.9 5.7 (N) (178) (153) A11 Taiwanese Prefer KMT 28.3 19.0 -9.3 Prefer DDP 17.0 13.6 -3.4 Prefer NP -- 6.7 6.7 Indifferent 54.7 60.7 6.0 (N) (1012) (939) 1 Items for 1996 were worded as like/dislike questions for two major parties. 2 Items for 1996 were worded as like/dislike questions for three major parties. *N<25. 69 KMT-partisan mainlanders defected from the KMT in 1996. Table 4—4 also indicates that young—generation mainlanders were more likely to change their partisan preferences when they chose among three major parties. Third-generation mainlanders not only deserted the KMT, but they also defected from the DPP when the NP entered electoral politics. When first-generation mainlanders had to choose among three major parties, two out of three KMT partisans deserted the KMT and chose the NP. Overall, for mainlanders, support of the KMT among the three generations was significantly affected by the emergence of the NP. When mainlanders faced three major parties, they liked the NP more than the KMT; fewer mainlanders preferred the DPP. Therefore, the emergence of the NP significantly affected the KMT’s support from mainlanders. However, when third-generation Taiwanese and most mainlanders faced two major parties, i.e., the KMT and the DPP, they still preferred the KMT. Therefore, the fates of the three major parties in the elections are closely related. If the NP becomes stronger, the KMT will receive less support from third-generation Taiwanese and most mainlanders. Then the DPP will have a relative advantage among the three major parties. In the legislative elections, the electoral system is the single-nontransferable vote system (SNTV). Therefore, NP partisans have an incentive to cast their sincere votes” in the legislative elections. On the other hand, if the KMT can attract more voters from NP partisans, it still can secure its majority in the elections. In presidential and mayoral elections, the electoral system is the plurality system with a single seat. Therefore, NP ‘6 A strategic or "sophisticated vote is one based on weighing both preferences and probabilities of success. A sincere vote is based upon preferences alone and leads to voting for the most preferred candidate, regardless of circumstances. When a voter’s sophisticated and sincere choices coincide, the choice is called ’straightforward’" (Abramson et al. 1992: 57). For details, see Abramson et a1. (1992). 70 Table 4-4. Two Measurements of Partisan Preference Among Mainlanders, by Generation, 1996 Generation Partisan Two Major Three Major Difference: Preference Parties] Parties 2 2"" - 1'’ 1961 and After Prefer KMT 42.7 11.4* -31.3 Prefer DDP 12.4* 2.3* -10.1 Prefer NP -- 35.2 35 .2 Indifferent 44.9 5 l . 1 6.2 (N) ( 89) ( 88) 1943 to 1960 Prefer KMT 48.0* 16.0”“ -32.0 Prefer DDP 8.0* 6.0* -2.0 Prefer NP -- 38.0”“ 38.0 Indifferent 440* 40.0* -4.0 (N) ( 50) ( 50) 1942 and Before Prefer KMT 54.9 30.3* -24.6 Prefer DDP 4.2’“ 1.5* -2.7 Prefer NP -- 33.3* 33.3 Indifferent 40.9 34.8* -6. 1 (N) ( 71) ( 66) All Taiwanese Prefer KMT 48.3 18.7 -29.6 Prefer DDP 8.6* 3.0* -5.6 Prefer NP -- 35.0 35.0 Indifferent 43.1 43.3 0.2 (N) (209) (203) 1 Items for 1996 were worded as like/dislike questions for two major parties. 2 Items for 1996 were worded as like/dislike questions for three major parties. *N<25. 71 partisans have an incentive to cast their strategic votes if they believe their favorite candidate cannot win the election but their less favorite candidate, i. e., the DPP nominee, might win the election. In the 1998 election in Taipei, the NP’s nominee received only 2.97 percent of the vote in the Taipei mayoral election, but the NP’s city councilors received 18.7 percent of vote and the NP’s Legislative candidates got 17.3 percent of the vote in the same area. Therefore, it is believed that the NP’s supporters cast their strategic votes to support the KMT’s mayoral nominee and to defeat the incumbent DPP mayor. Conclusion Party identification was relatively stable among older generations between 1987 and 1996 in Taiwan. Generations with different political experience during their early adulthood tended to have different partisan preferences. In addition, ethnic groups in Taiwan also had different partisan preferences. From the above analysis, I found that the two older generations tended to have more consistent partisan preferences than the younger generation. Their loyalties toward the KMT were consistent, and they were less likely to be attracted by the DPP. On the other hand, third-generation voters became less likely to support the KMT over time, but their support for the DPP increased between 1990 and 1996. I also found differences in partisan preferences among the two major ethnic groups in Taiwan. When facing two major parties, first-generation mainlanders were the group most loyal to the KMT, and they had the highest partisan strength among the six generation-ethnicity groups between 1987 and 1996. First-generation Taiwanese were least likely to support the KMT in 1987, but they increased their support over time and 72 became the group most loyal to the KMT among the three Taiwanese generations in 1996. Third-generation mainlanders entered the electorate with a higher level of partisan strength than did the third-generation Taiwanese. This indicates that young mainlanders learned their partisan preferences from their parents, and some historical events had a long-term effect on their parents’ partisan preferences. On the other hand, experiencing political openness and electoral competition, younger-generation Taiwanese formulated their partisan preferences when they entered the electorate. Partisan strength among younger-generation Taiwanese was higher than it was among older Taiwanese. The distribution of DPP partisans was more stable than the distribution of KMT partisans. In addition, Taiwanese were more likely to support the DPP than were mainlanders. Among Taiwanese, third-generation voters became more likely to support the DPP than the two other generations in 1996. On the other hand, less than 5 percent of mainlander voters preferred the DPP between 1987 and 1996. When voters had to choose among three major parties, third-generation voters were more likely to defect from the KMT. However, partisan preferences of two older generations were less likely to be affected by the emergence of the NP. There were ethnic differences in partisan preferences when voters faced three major parties. The two older generations of Taiwanese were less likely to be affected by the emergence of the NP, but one out of two third-generation Taiwanese voters defected from the KMT. On the other hand, two out of three mainlanders from the three generations defected from the KMT when they had an additional alternative, i.e., the NP. But young mainlanders were more likely to be attracted by the NP than were older mainlanders. 73 As Beck (1974: 205) wrote, "The newest members of the electorate provide the dynamic element to electoral politics. They are the ones most likely to break the partisan continuity between past and future and to force comprehensive changes in the policy agenda." Support of the DPP increased between 1987 and 1996, and most supporters were third-generation Taiwanese. In addition, between 1993 and 1996, the newly emerged NP attracted young voters. Therefore, about two out of three mainlanders defected from the KMT in 1996. Therefore, party identification is relatively stable, but it is still mutable. When a political party changes its position on a certain issue, this might affect its partisans' loyalty. The establishment of the DPP in 1986 provided an electoral challenge to the KMT. It attracted many young Taiwanese voters and increased its votes in the legislative elections monotonically. As the first Taiwanese President, Lee Teng- hui, adopted a more ambiguous stance on the issue of unification with mainland China versus Taiwan's independence and recruited more Taiwanese into the power circle, some mainlanders defected from the KMT and turned their support to the NP. These political and social events also changed the social bases of partisan support. In the following chapter, I discuss how other demographic background affected people's partisan preferences. In Chapter 6, I examine how political parties changed their positions on national identity and other issues that might affect their social base. 74 CHAPTER 5 SOCIAL BASES OF PARTISAN SUPPORT The social bases of party support in Taiwan are explored in this chapter. Ixmgitudinal data sets from the National Taiwan University are used in examining how different sociological backgrounds might affect voters’ partisan preferences. In addition, I discuss how different social groups changed their partisan support when the NP entered the electoral competition. Social Cleavage and Partisan Support An effective political party must attract enough votes to survive in the elections. Therefore, parties need to provide attractive platforms to gain sufficient support from the electorate. Lipset (1981) argued that the role of religion, the struggle among different social classes, and the justice of economic distribution are three major issues in Western nations. In modern times, three major issues have emerged in Western nations: first, the place of the church and/or various religions within the nation; second, the admission of the lower strata, particularly the workers, to full political and economic ’citizenship’ through universal suffrage and the right to bargain collectively; and third, the continuing struggle over the distribution of the national income. (Lipset 1981: 71) 75 Different parties might take different positions on these issues to maximize their votes in the elections. However, class conflict might create a major social cleavage in Western electoral politics. Lipset argued that, “On a world scale, the principal generalization which can be made is that parties are primarily based on either the lower classes or the middle and upper classes” (Lipset 1981:230). Dalton (1998) examined how individuals' social characteristics might affect their partisan support. He gave several reasons to explain why social characteristics might affect a voter's choice of party. First, a person’s social position often indicates his or her values and political beliefs. . .Second, social characteristics indicate some of the political cues to which an individual is exposed. . .Third, social groups can be an important reference point in orientating voters to political issues and providing information about politics. (Dalton 1998: 165-166) However, the effect of traditional social cleavage on partisan alignment is declining (Abramson et al. 1998; Dalton 1998). Dalton also argued that new issues or values, such as postmaterialism, might not be useful cues for voters to form their partisan support because they are attitudinal variables and are more volatile than individuals' social characteristics. The new style of citizen politics therefore includes a more fluid and volatile pattern of party alignments. Political coalitions and voting patterns will lack the performance of past class and religious cleavages. Without clear social cues, voting decisions will become a more demanding task for voters, and voting decisions will become more dependent on the individual beliefs and values of each citizen. (Dalton 1998: 194) In Taiwan, some political scientists also have examined how different social characteristics mi ght affect voters' partisan support (Chen 1992; Chu 1992; Ho 1994; Hsieh and Niou 1996; Lin 1998; Lin et al. 1998; Liu 1987; Shyu 1993; Tsai 1997; You 1994b). Lin and his colleagues provided comprehensive models to explain how people's 76 demographic background and political attitudes affect their partisan preferences. Their models included people’s attitudes toward national identity, money politics, and procedural democratic values, as well as their gender, age, ethnic origin, vocation, and subjective social economic status. As discussed in Chapter 4, voters’ ethnicity is an important social cleavage among the electorate. In this chapter, I turn my focus to other social cleavages such as gender, educational level, vocation, and subjective social class perception. I still use political generation as a mediator variable to examine generational differences among different social groups with regard to their partisan preferences. First, I examine gender differences in voters’ partisan preferences. Because females tend to have a lower educational level and less information than males, Lipset (1981: 231) argued that "women tend to support the conservative parties more than men do." In Taiwan, the ruling party, the KMT, emphasizes that it can maintain the social order, so we can expect that females are more likely than males to support the KMT. As to voters’ educational levels, Lipset (1981: 335) argued that "the better educated individuals are, the more likely they are to favor all forms of ’noneconomic liberalism.” Therefore, we can expect that educated voters in Taiwan are more likely to support the opposition parties because both the DPP and the NP support social reforms. PeOple’s vocation might also affect their partisan preferences. It is conventional wisdom that labor workers are more likely to support leftist parties. Finifter (1974) also showed that people’s workplace as a social context might affect their partisan support. In Taiwan, the KMT has been the ruling party since 1949, so we can expect that government employees are more likely than people with other vocations to support the ruling party. 77 In this chapter, I also examine how people’s subjective socioeconomic status might affect their partisan preferences. Since position in a stratification system is always relative and gratification or deprivation is experienced in terms of being better or worse off than other people, it is not surprising that the lowest classes in all countries show various signs of resentment against the existing distribution of rewards by supporting political parties and other organizations which advocate some form of redistribution. (Lipset 1981: 48) Therefore, we can expect that people of the upper class will tend to support the KMT.17 Social Bases of Support for the KMT In this section, I examine how different social backgrounds might affect voters' partisan preferences for the KMT. Because the KMT has been the ruling party since 1949, we can expect that people who belong to the upper or dominant groups will tend to support the KMT. First-Generation Males Were More Likely to Support the KMT As shown at the bottom of Table 5-1, 51 percent of males had a positive attitude toward the KMT in 1987. However, only 30 percent reported such an attitude in 1996. Females' positive attitude toward the KMT was relatively stable. It started at 38 percent in 1987, rebounded to 46.2 percent in 1990, and dropped to 33.5 percent in 1996. Compared with males, females became more likely to support the KMT since 1993. ‘7 Because limitations of space in this chapter, I present a "party preference" index in Appendix C. The "party preference" index measures relative party strength as the difference between the percentage of one group preferring the KMT and the percentage of the same group preferring the DPP. See also Miller (1991). 78 Table 5-1. Percentages of KMT Partisans,l by Gender, 1987 to 1996 Generation Gender 1987 2 1990 1993 1996 Range Mean 1961 and After Male 62.2 47.0 40.8 28.1 34.1 44.5 Female 53.7 52.2 43.9 30.1 23.6 45.0 Difference 3 8.5 -5 .2 -3.1 -2.0 1943 to 1960 Male 44.1 44.4 30.2 22.7 21.7 35.4 Female 36.9 47.3 36.8 34.2 13.1 38.8 Difference 7.2 -2.9 -6.6 -11.5 1942 and Before Male 46.7 54.5 45.0 46.5 9.5 48.2 Female 23.7 33.0 33.3 40.2 16.5 32.6 Difference 3 23.0 21.5 11.7 6.3 Total Male 51.2 48.3 38.1 30.3 20.9 42.0 Female 37.9 46.2 38.5 33.5 12.7 39.0 Difference 13.3 2.1 -0.4 -3.2 National mean 44.6 47.3 38.3 31.8 15.5 40.5 N 1,359 1,001 1,300 1,239 1 For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. 2 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 3 Difference = Male - Female. 79 I also examined gender differences within each generation with regard to their partisan preferences. At the top of Table 5-1, we can see that males in the third generation (born in 1961 and after) had the highest percentage of positive attitudes toward the KMT in 1987 of the six social groups considered, but their support for the KMT declined monotonically in the following elections. Specifically, 62.2 percent of them supported the KMT in 1987, but only 28.1 percent of them had a positive attitude toward the KMT in 1996. The mean was 44.5 percent and the range was 34.1 percent between 1987 and 1996. A majority of females in the third generation supported the KMT in 1987 and 1990, but the proportion declined to 43.9 percent in 1993 and dropped to 30 percent in 1996. The mean was 45 percent and the range was 23.6 percent. Third-generation males were more likely than their female counterparts to support the KMT in 1987. Since 1990, however, females in the third generation have become more likely than males in that generation to support the KMT. In the middle of Table 5-1, we can see that 44 percent of males in the second generation (born between 1943 and 1960) had a positive attitude toward the KMT in 1987 and 1990, but the proportion declined to 30 percent in 1993 and was only 22.7 percent in 1996. The mean was 35 .4 percent and the range was 21.7 percent. As to second-generation females, 37 percent of them preferred the KMT in 1987 and 47.3 percent did so in 1990, but the pr0portion of female KMT partisans dropped to 36.8 percent in 1993 and declined further to 34.2 percent in 1996. The mean was 38.8 percent and the range was 13.1 percent. Males in the second generation were more likely than females to support the KMT in 1987, but females have become more likely to support the KMT since 1990. 80 Compared with other males, those in the first generation (born in 1942 and before) were the most stable in their preference of the KMT; 46.7 percent of males in that generation supported the KMT in 1987, and this figure increased to 54.5 percent in 1990. However, it dropped to 45 percent in 1993 and slightly increased to 46.5 percent in 1996. The mean was 48.2 percent and the range was 9.5 percent. Support of the KMT among first-generation females increased monotonically between 1987 and 1996. Less than one out of four of them had a positive attitude toward the KMT in 1987, but about four out of ten supported the KMT in 1996. The mean was 32.6 percent and the range was 16.5 percent. Between 1987 and 1996, first-generation males were more likely than their female counterparts to support the KMT. The difference between males and females in their support for the KMT was 23 percent in 1987, but it decreased to 6.3 percent in 1996. Table 5-1 also indicates that third-generation males were the group most loyal to the KMT in 1987, but first-generation males have become the party’s most supportive group since 1990. According to Table 5-1, first-generation males were most conservative voters among the six groups examined. However, the KMT received more support from females than males in the two younger generations. The Educated Tended to Support the KMT Between 1987 and 1993 As shown at the bottom of Table 5-2, a majority of voters with college and above (13 years and above) had a positive attitude toward the KMT between 1987 and 1993, but the proportion dropped to 30 percent in 1996. The mean was 52.2 percent and the range was 39.4 percent. Support of the KMT among people with a high school education (10 to 12 years) declined monotonically between 1987 and 1996. Specifically, 58.4 percent of 81 them preferred the KMT in 1987, but the proportion declined to 31.7 percent in 1996. The mean was 47.1 percent and the range was 26.7 percent. As to people with a less than high school education (9 years and less), their support for the KMT was more stable. It started at 32.6 percent in 1987 and remained at 33 percent in 1996. The mean was 47.1 percent and the range was 7.6 percent. Thus, between 1987 and 1993, people with higher educational levels tended to support the KMT. However, in 1996, people with a lower educational level were more likely to support the KMT. As to voters in the third generation, people with college and above were more likely to support the KMT in 1987 and 1993; 67 percent of them preferred the KMT in 1987 and about 50 percent did so in 1990. The proportion rebounded to 53.3 percent in 1993 but declined to only 28.3 percent in 1996. The mean was 49.6 percent and the range was 39 percent. As to people in this group with a high school education, 64.4 percent of them preferred the KMT in 1987 and 55.5 percent did so in 1990. In 1993, the proportion dropped significantly, to 39 percent, and it declined still further, to 30.2 percent, in 1996. The mean was 47.3 percent and the range was 34.2 percent. Support of the KMT by third-generation voters with less than a high school education was more stable than that by people at the other two levels of education. The proportion was 41 percent in 1987, but it dropped to 38.7 percent in 1990, to 30.7 percent in 1993, and to 28 percent in 1996. The mean was 34.6 percent and the range was 13 percent. As to voters in the second generation, people with a higher educational level were more likely to support the KMT between 1987 and 1993. More than seven out of ten voters with college and above preferred the KMT in 1987, but the proportion declined to 57.1 percent in 1990. The proportion of KMT partisans dropped further to 47.4 percent, 82 Table 5-2. Percentages of KMT Partisansl by Level of Education, 1987 to 1996 Generation Level of Education 2 1987 3 1990 1993 1996 Range Mean 1961 and After Less than high school 41.0 38.7 30.7 28.0 13.0 34.6 High school 64.4 55.5 39.0 30.2 34.2 47.3 College and above 67.3 49.5 53.3 28.3 39.0 49.6 Difference 4 26.3 10.8 22.6 0.3 1943 to 1960 Less than high school 29.7 33.7 23.6 28.1 10.1 28.8 High school 48.1 56.2 42.6 29.5 26.7 44.1 College and above 71.1 57 .1 47.4 27.3 43.8 50.7 Difference 4 41.4 23.4 23.8 -0.8 1942 and Before Less than high school 31.2 39.7 34.5 41.6 10.4 36.8 High school 57.6* 625* 51.3* 55.2* 11.2 56.7 College and above 71.4* 74.2* 70.3 45.9* 28.3 65.5 Difference 4 40.2 34.5 35.8 4.3 Total Less than high school 32.6 37.1 29.5 33.0 7.6 33.1 High school 58.4 56.5 41.6 31.7 26.7 47.1 College and above 69.4 55.8 53.5 30.0 39.4 52.2 Difference 4 36.8 18.7 24.0 -3.0 National mean 44.6 47 .3 38.3 31.8 15.5 40.5 N 1,359 1,001 1,300 1,239 1 For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. 2 Levels of education for each category were: less than high school = 0 to 9 years, high school = 10 to 12 years, and college and above = 13 years and above. 3 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 4 Difference = "College and above" - "Less than High School." * N<25. 83 in 1993, and only 27.3 percent of voters in this educational category had a positive attitude toward the KMT in 1996. The mean was 50.7 percent and the range was 43.8 percent. As to second-generation voters with a high school education, 48.1 percent of them supported the KMT in 1987, and the proportion increased to 56.2 percent in 1990. However, the figure declined to 42.6 percent in 1993 and to less than 30 percent in 1996. The mean was 44.1 percent and the range was 26.7 percent. As to second-generation voters with a less than high school education, 29.7 percent of them supported the KMT in 1987 and 33.7 percent in 1990. The proportion declined to 23.6 percent in 1993, but it rebounded to 28.1 percent in 1996. The mean was 28.8 percent and the range was 10.1 percent. Between 1987 and 1993, people in the first generation with a higher educational level were more likely to have a positive attitude toward the KMT. More than seven out of ten voters with college and above had a positive attitude toward the KMT, although the proportion declined to 46 percent in 1996. The mean was 65.5 percent and the range was 28.3 percent between 1987 and 1996. As to first-generation voters with a high school education, 57.6 percent of them preferred the KMT in 1987 and the proportion increased to 62.5 percent in 1990. This figure declined to 51.3 percent in 1993 and rebounded to 55.2 percent in 1996. The mean was 56.7 percent and the range was 11.2 percent. One out of three first-generation voters with less than a high school education supported the KMT in 1987; the proportion increased to 39.7 percent in 1990, declined to 34.5 percent in 1993, but rebounded to 41.6 percent in 1996. The mean was 36.8 percent and the range was 10.4 percent. 84 Government Employees Were Loyal Supporters of the KMT Between 1987 and 1996, government employees were more likely than those in other vocations to support the KMT. They preferred the KMT for several reasons. First, many civil servants followed Chiang Kai-shek to Taiwan in 1949, so they were likely to support the KMT. Second, civil servants enjoy job tenure and other welfare measures whenever they enter the government, so for that reason also government employees are likely to support the KMT. As shown at the bottom of Table 5-3, three out of four government employees had a positive attitude toward the KMT in 1987 and 1990. However, their support for the KMT declined to 66.5 percent in 1993 and to 45 percent in 1996. The mean was 62.1 percent and the range was 32.2 percent. Between 1987 and 1993, support of the KMT by government employees outnumbered that of people in other vocations by 30 percent. However, the difference declined to 15 percent in 1996. As to third-generation voters, eight out of ten state employees had a positive attitude toward the KMT in 1987 and 1990. However, only six out of ten state employees in this group preferred the KMT in 1993, and only one out of three in 1996. The mean was 57 percent and the range was 50.7 percent. Second-generation state employees were less likely to support the KMT than state employees in the other two generations in 1987 and 1990. About two out of three second- generation government employees had a positive attitude toward the KMT in 1987 and 1990. The proportion of KMT partisans was 63 percent in 1993 and it declined to 42 percent in 1996. The mean was 56.8 percent and the range was 26.7 percent. 85 Table 5-3. Percentages of KMT Partisans,l by Vocation, 1987 to 1996 Generation Vocation 1987 2 1990 1993 1996 Range Mean 1961 and After Government employee 81.8 80.0* 60.0* 31.1* 50.7 57.0 Other 55.2 48.7 40.4 28.9 26.3 39.3 Difference 3 26.6 31.3 19.6 2.2 1943 to 1960 Government employee 68.4 657* 62.9 41.7* 26.7 56.8 Other 37.5 44.0 28.7 26.8 17.2 33.2 Difference 3 30.9 21.7 34.2 14.9 1942 and Before Government employee 88.2* 850* 73.5 60.8 27.4 73.1 Other 33.2 43.3 33.1 39.5 10.2 38.6 Difference 3 55.0 41.7 40.4 21.3 Total Government employee 77.3 74.7 66.5 45.1 32.2 62.1 Other 41.8 45 .4 34.0 30.0 15 .4 36.5 Difference 3 35.5 29.3 32.5 15.1 National mean 44.6 47 .3 38.3 31.8 15.5 40.5 N 1,359 1,001 1,300 1,239 k 1 For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. 2 3 Difference = "Government Employee" - "Other." * N<25. 86 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. First-generation government employees were the group most loyal to the KMT. More than eight out of ten civil servants supported the KMT in 1987 and 1990, and about three out of four preferred the KMT in 1993. Their support for the KMT declined in 1996, but six out of ten government employees in this generation still held a positive attitude toward the KMT. The mean was 73.1 percent and the range was 27.4 percent. The Disappearance of Class Voting in 1996 People who considered themselves as middle class and above (middle class, middle-high class, and high class) were more likely than those from lower classes to support the KMT between 1987 and 1993, Specifically, 47.7 percent and 49.7 percent of them had a positive attitude toward the KMT in 1987 and 1990 respectively. The proportion of KMT partisans in the middle class and above declined to 42.4 percent in 1993, and only 31.9 percent of them supported the KMT in 1996. The mean was 42.9 percent and the range was 17.8 percent. As to people considering themselves below middle class (low class to low-middle class), they were less likely to support the ruling party between 1987 and 1993. About 33.3 percent of them preferred the KMT in 1987 and nearly 40 percent of them did so in 1990. The proportion declined to 24 percent in 1993 but rebounded to 31 percent in 1996. The mean was 32 percent and the range was 15.7 percent. As shown at the bottom of Table 5-4, there was no significant difference between the two groups in their support of the KMT. Among third-generation voters, 60 percent of people with subjective middle class (and above) identity preferred the KMT in 1987, and 50 percent of them had a positive attitude toward the KMT in 1990. The proportion declined slightly, to 46 percent, in 87 Table 5-4. Percentages of KMT Partisans,1 by Subjective Socioeconomic Status, 1987 to 1996 Generation Subjective 1987 3 1990 1993 1996 Range Mean Socioeconomic Status 2 1961 and After Below middle class 49.2 49.0 25.7* 27.9* 23.5 38.0 Middle class and above 59.6 50.2 46.0 29.2 30.4 46.3 Difference 4 10.4 1.2 20.3 1.3 1943 to 1960 Below middle class 32.0 34.7 21.4* 21.6* 10.6 27.4 Middle class and above 43.0 49.7 35 .8 30.1 19.6 39.7 Difference 4 11.0 15.0 14.4 8.5 1942 and Before Below middle class 26.7 39.7 24.8 46.1 21.3 34.3 Middle class and above 38.4 49.2 47.8 43.8 10.8 44.8 Difference" 11.7 9.5 23.0 -2.3 Total Below middle class 33.3 39.7 24.0 30.9 15.7 32.0 Middle class and above 47.7 49.7 42.4 31.9 17.8 42.9 Difference 4 14.4 10.0 18.4 1.0 National mean 44.6 47.3 38.3 31.8 15.5 40.5 N 1,359 1,001 1,300 1,239 1 For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. 2 "Below middle class" includes low-middle class and low class. "Middle class and above" includes middle class, middle-high class, and high class. 3 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 4 Difference = "Middle class and above" - " Below middle class." *N<25. 88 1993, but only 29 percent of them supported the KMT in 1996. The mean was 46.3 percent and the range was 30.4 percent. As to people in this generation who considered themselves belonging to low class or low-middle class (below the middle class), almost 50 percent of them preferred the KMT in 1987 and 1990. Their support of the KMT dropped dramatically to 25.7 percent in 1993 and 27.9 percent in 1996. The mean was 38 percent and the range was 23.5 percent. As to second-generation voters, 43 percent of people with middle class identity (and above) had a positive attitude toward the KMT in 1987; the proportion increased to 50 percent in 1990. However, the proportion declined significantly to 35.8 percent in 1993 and only 30.1 percent in 1996. The mean was 39.7 percent and the range was 19.6 percent. Among second-generation voters with lower-class identity, about 32 percent of them supported the KMT in 1987, 34.7 percent in 1990, 21.4 percent in 1993, and 21.6 percent in 1996. The mean was 27.4 percent and the range was 10.6 percent. Concerning first-generation voters with middle-class identity and above, 38.4 percent of them preferred the KMT in 1987, and the proportion rebounded to 49.2 percent in 1990. The proportion of KMT partisans was 47.8 percent in 1993 and 43.8 percent in 1996. The mean was 44.8 percent and the range was 10.8 percent. As to people in this generation with low- and low-middle-class identity, 26.7 percent of them had a positive attitude toward the KMT in 1987, and the proportion rebounded to 39.7 percent in 1990. In 1993, the proportion dropped to 24.8 percent, but it increased to 46.1 percent in 1996. The mean was 34.3 percent and the range was 21.3 percent. It is interesting that people with lower-class identity were more likely to support the KMT than were those with upper-class identity. 89 Summary In summary, as to voters’ gender, it was found that first-generation males were the most loyal group among the three generations to support the KMT since 1990. However, females in the two younger generations became more likely than males to support the KMT since 1993. People’s educational levels also affected their partisan preferences. Educated voters were more likely to prefer the KMT between 1987 and 1993, but the KMT gained support from people with less than a high school education in 1996. This might have been caused by the emergence of the NP--I will return to this point later in this chapter. Whether voters were civil servants was a good predictor of their support of the KMT. Between 1987 and 1996, civil servants were the most loyal supporters of the KMT, but third-generation government employees became less likely to support the KMT. Between 1987 and 1993, people’s subjective class identity affected their partisan preferences. The KMT received more support from people with middle— and upper-class identity. But the class difference in KMT support was small in 1996. Social Bases of Support for the DPP In this section, the social bases of support for the DPP in Taiwan are examined. The DPP promotes social justice and political reforms, so we can expect that liberal voters would be more likely to support the DPP. Because the proportion of DPP partisans was relatively small, one must be careful in making inferences from the tables in this section. 90 Disappearance of the Gender Gap Between DPP Partisans As the bottom of Table 5-5 shows, males were more likely than females to have positive attitudes toward the DPP between 1987 and 1996. In 1987, 10.9 percent of males supported the DPP and 11.5 percent did in 1990. The proportion of DPP partisans increased to 16 percent in 1993 and 1996. Females’ support for the DPP also increased between 1987 and 1996. Only 5 .4 percent of females held a positive attitude toward the DPP in 1987, and the proportion increased to about 9 percent in 1990 and 1993. In 1996, 14.5 percent of females preferred the DPP. The difference between males and females in their support for the DPP decreased from 5.5 percent in 1987 to 1.9 percent in 1996. Among third-generation voters, 10 percent of males supported the DPP in 1987, but the proportion dropped to 6.6 percent in 1990. The proportion of DPP partisans rose to 15.1 percent in 1993 and increased further to 19.7 in 1996. As to third-generation females, 8.4 percent of them supported the DPP in 1987; the proportion increased to 11 percent in 1990, decreased to 9.4 percent in 1993, but rebounded to 17.4 percent in 1996. The gender difference in DPP support was less between third-generation voters than among voters in the other two generations. The proportion of DPP partisans among second-generation males increased between 1987 and 1993. Specifically, 14.4 percent of them preferred the DPP in 1987, 16.8 percent in 1990, and 20.2 percent in 1993. However, the figure dropped to 16.3 percent in 1996. Less than 5 percent of second-generation females supported the DPP in 1987, and only 7 percent did so in 1990. The proportion increased to 9.4 percent in 1993 and 12.6 percent in 1996. The difference between second-generation males and females 91 Table 5-5. Percentages of DPP Partisans,l by Gender, 1987 to 1996 Generation Gender 1987 2 1990 1993 1996 Range Mean 1961 and After Male 10.1* 6.6* 15.1 19.7 9.6 12.9 Female 84* 11.0* 9.4* 17.4 9.0 11.6 Difference 1.7 -4.4 5.7 2.3 1943 to 1960 Male 14.4 16.8 20.2 16.3 5.8 16.9 Female 47* 7.0* 9.4* 12.6 7.9 8.4 Difference 9.7 9.8 10.8 3.7 1942 and Before Male 79* 9.0* 12.8 10.4* 4.9 10.0 Female 31* 8.0"“ 7.7* 11.2* 8.1 7.5 Difference 4.8 l .0 5. 1 08 Total Male 10.9 11.5 16.3 16.4 5.5 13.8 Female 5.4 8.8 9.0 14.5 9.1 9.4 Difference 3 5.5 2.7 7 .3 1.9 National mean 8.2 10.2 12.9 15.5 7.3 11.7 N 1,359 1,001 1,300 1,239 1 For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. 2 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 3 . Drfference = Male - Female. *N<25. 92 was larger than that for the other two generations. Between 1987 and 1993, second- generation males were the most likely of six groups included in Table 5-5 to support the DPP. Less than 8 percent of first-generation males had a positive attitude toward the DPP in 1987, and 9 percent of them did in 1990. The proportion of DPP partisans increased to 12.8 percent in 1993, but it decreased to 10.4 percent in 1996. Less than 4 percent of first-generation females supported the DPP in 1987, and the proportion was only 8 percent in 1990. In 1993, only 7.7 percent of first-generation females preferred the DPP, but the proportion increased to 11.2 percent in 1996. The DPP Received More Support From Educated Voters in 1996 As shown at the bottom of Table 5-6, voters with college and above were more likely to support the DPP than those from the other two educational levels, except in 1993. Specifically, 12 percent of them had positive attitudes toward the DPP between 1987 and 1993, and the proportion increased to about 20 percent in 1996. Support of the DPP among voters with a high school education increased slightly between 1987 and 1996. In 1987, 8.9 percent of them had a positive attitude toward the DPP, but the proportion slightly declined to 8 percent in 1990 and then increased to 11.5 percent in 1993 and 14.4 percent in 1996. Support of the DPP among people with less than a high school education increased gradually between 1987 and 1993. It started at only 7.2 percent in 1987, rose to 10.5 percent in 1990, and increased further to 14 percent in 1993. In 1996, the proportion of DPP partisans among voters with less than a high school education was 13.6 percent. 93 As to third-generation voters, people with college and above were more likely to support the DPP in 1987, 1990, and 1996. Of that group, 14.3 percent had a positive attitude toward the DPP in 1987; the proportion dropped to about 12 percent in 1990 and 1993, but it increased significantly to 23.2 percent in 1996. This group was the most volatile of the nine educational groups included in Table 5-6 in its support of the DPP, but this group had the highest proportion supporting the DPP in 1996. As to third- generation voters with a high school education, 9.6 percent of them preferred the DPP in 1987, the proportion declined to 6.8 percent in 1990, and it increased to 11 percent in 1993. In 1996, the proportion increased to 15.7 percent. Support of the DPP among third— generation people with less than high school increased monotonically between 1987 and 1996. The proportion was 8.7 percent in 1987 and 9.3 percent in 1990, and it increased to 16 percent in 1993 and 1996. As to second-generation voters, people with college education and above were more likely to support the DPP than those from the other two educational levels in 1987, 1990, and 1996. About one out of ten voters with college and above level had a positive attitude toward the DPP in 1987; the proportion increased to 14.3 percent in 1990, 14.7 percent in 1993, and 17 percent in 1996. As to second-generation voters with a high school education, about 7 percent of them supported the DPP in 1987 and 1990, and the proportion increased to 13 percent in 1993 and 1996. As to voters in this generation with less than a high school educational level, about 10 percent of them were KMT partisans in 1987; the proportion increased to 14 percent in 1990 and 16.2 percent in 1993, and then declined to 14 percent in 1996. In the two younger generations, voters with a high school education were less likely to support the DPP than were those at the other two 94 Table 5-6. Percentages of DPP Partisans,1 by Level of Education, 1987 to 1996 Generation Year of Education 2 1987 3 1990 1993 1996 Range Mean 1961 and After Less than high school 8.7* 9.3* 15.9* 16.0* 7.3 12.5 High school 9.6* 6.8* 11.0* 15.7 8.9 10.8 College and above 14.3* 11.7* 11.8* 23.2 11.5 15.3 Difference 4 5.6 2.4 -4.1 7.2 1943 to 1960 Less than high school 9.5 13.9 16.2 14.0 6.7 13.4 High school 7.5* 7.4* 13.2* 13.7* 6.3 10.5 College and above 11.1* 14.3* 14.7* 17.0* 5.9 14.3 Difference 4 1.6 0.4 -1.5 3.0 1942 and Before Less than high school 4.8* 7.8* 11.4 11.9* 7.1 9.0 High school 9.1* 15.6* 7.7* 6.9* 8.7 9.8 College and above 7.1* 6.5* 8.1* 8.1* 1.6 7.5 Difference 4 2.3 -1.3 -3.3 -3.8 Total Less than high school 7.2 10.5 14.0 13.6 6.8 11.3 High school 8.9 8.0 11.5 14.4 6.4 10.7 College and above 12.0 12.0 12.3 19.8 7.8 14.0 Difference 4 4.8 1.5 -1.7 6.2 National mean 8.2 10.2 12.9 15.5 7.3 11.7 N 1,359 1,001 1,300 1,239 1 For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. 2 Levels of education for each category were: less than high school = 0 to 9 years, high school = 10 to 12 years, and college and above = 13 years and above. 3 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 4 Difference = "College and above" - "Less than High School." * N<25. 95 educational levels. As to the first-generation voters, fewer than one out of ten with college and above had a positive attitude toward the DPP between 1987 and 1996. Among voters with a high school education, the proportion of DPP supporters was 9.1 percent in 1987, 15.6 percent in 1990, only 7.7 in 1993, and 6.9 percent in 1996. Less than 4.8 percent of voters with less than a high school education preferred the DPP in 1987; the proportion increased to 7.8 percent in 1990, 11.4 percent in 1993, and 11.9 percent in 1996. The DPP Attracted More Younger Government Employees As discussed above, government employees were the most loyal social group to support the KMT, so one might expect their support for the DPP to be very low. However, as shown in Table 5-7, the DPP has begun to attract younger civil servants since 1993. As shown at the bottom of Table 5-7, less than 6 percent of government employees had a positive attitude toward the DPP in 1987 and less than 7 percent did in 1990. That proportion increased to 9.4 percent in 1993 and to about 14 in 1996. However, Voters’ support of the DPP was relatively low between 1987 and 1996, so the differences between state employees and other vocation in their partisan preferences toward the DPP were relatively small. In 1996, the difference was only 1.8 percent. As to third-generation state employees, 5 percent or less of them had a positive attitude toward the DPP in 1987 and 1990. However, the proportion increased to 15 percent in 1993 and 17.8 percent in 1996. It is interesting that support of the DPP by 96 Table 5-7. Percentages of DPP Partisans,1 by Vocation, 1987 to 1996 Generation Vocation 1987 2 1990 1993 1996 Range Mean 1961 and After Government employee 30* 5.0* 15.0* 17.8* 14.8 10.2 Other 10.2 9.5 12.2 18.6 9.1 12.6 Difference 3 -7.2 -4.5 2.8 -0.8 1943 to 1960 Government employee 79* 8.6* 12.9* 14.6* 6.7 11.0 Other 9.8 12.0 15.4 14.5 2.2 12.9 Difference 3 -1.9 -3.4 -2.5 0.1 1942 and Before Government employee 59* 5.0* 2.9* 9.8* 6.9 5.9 Other 5.4* 8.8* 12.6 11.0* 7.2 9.5 Difference 3 0.5 -3.8 -9.7 -1.2 Total Government employee 5.7 6.7 9.4 13.9 8.2 8.9 Other 8.5 10.3 13.5 15.7 7.2 12.0 Difference 3 -2.8 -3.6 -4.1 - l .8 National Mean 8.2 10.2 12.9 15.5 7.3 11.7 N 1,359 1,001 1,300 1,239 1 For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. 2 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 3 Difference = "Government Employee" - "Other." * N<25. 97 government employees outnumbered that by people in other vocations by 2.8 percent in 1993. As to second-generation state employees, less than one out of ten voters supported the DPP between 1987 and 1993. The proportion increased to 13 percent in 1993 and to 14.6 percent in 1996. As demonstrated in the last section, first-generation government employees were the most loyal group to support the KMT. In Table 5-7 it can be seen that less than 6 percent of first-generation state employees supported the DPP between 1987 and 1993; and the proportion was 9.8 percent in 1996. Therefore, from Table 5-7, one can see that the DPP received more support from younger civil servants than older ones, and that the proportion of DPP partisans among civil servants was close to the national mean in 1996. This might be positive news for the DPP. An Inconsistent Pattern of Class Support for the DPP As shown in Table 5-8, no consistent pattern was found for various social classes’ support of the DPP. As people who consider themselves less than middle class might want to support the opposition parties to bring about a change, one might expect that they would be more likely to support the DPP. However, as indicated at the bottom of Table 5- 8, people who considered themselves middle class and above were more likely than lower social class individuals to support the DPP in 1987 and 1996, although the differences were less than 3 percent. As to people of lower social class (below middle class), the proportions of DPP partisans were 6.9 percent in 1987, 12.8 percent in 1990, 15.6 percent 98 in 1993, and 14.3 percent in 1996. Among voters of middle class and above, 8.6 percent had a positive attitude toward the DPP in 1987, 9.4 percent in 1990, 12 percent in 1993, and 16.4 percent in 1996. As to third-generation voters, people who considered themselves low class or low-middle class (below middle class) had a more positive attitude toward the DPP between 1987 and 1996 than did those who considered themselves middle class and above. About 14.3 percent of them preferred the DPP in 1987, and 15.7 percent in 1990 and 1993. Their support of the DPP increased to 20.9 percent in 1996. Less than 9 percent of people with subjective middle class (and above) identity preferred the DPP in 1987 and 1990, and 11.2 percent of them had a positive attitude toward the DPP in 1993. The proportion increased to 18.5 percent in 1996. The third generation was the only one that demonstrated a consistent pattern of support for the DPP. As to second-generation voters, about 11 percent of people who considered themselves middle class (and above) had a positive attitude toward the DPP in 1987 and 1990, and the proportion increased to about 15 percent in 1993 and 1996. As to people with lower class identity, only 4.1 percent of them supported the DPP in 1987, 15.8 percent in 1990, 17.9 percent in 1993, and 14.4 percent in 1996. Among first-generation voters, 5.6 percent of people with middle class (and above) identity preferred the DPP in 1987 and the proportion increased to 8.5 percent in 1990. The proportion of the DPP partisans in this social class was 8.9 percent in 1993 and 13.7 percent in 1996. As to people with low- and low-middle-class identity, 5.3 percent had a positive attitude toward the DPP in 1987, and the proportion was 6.8 percent in 1990. In 1993, the proportion increased to 13.8 percent, but it dropped to 6.6 percent in 99 Table 5-8. Percentages of DPP Partisans,1 by Subjective Socioeconomic Status, 1987 to 1996 Generation Subjective 1987 3 1990 1993 1996 Range Mean Socioeconomic Status 2 1961 and After Below middle class 143* 15.7* 15.7* 20.9* 6.6 17.4 Middle class and above 8.5 7.9* 11.2 18.5 10.6 12.5 Difference 4 -5.8 -7.8 -4.5 -2.4 1943 to 1960 Below middle class 41* 158* 17.9* 144* 13.8 16.0 Middle class and above 11.1 11.3 14.6 14.9 3.8 13.6 Difference 4 7-0 '4-5 -3-3 0.5 1942 and Before Below middle class 53* 6.8* 13.8* 6.6* 8.5 9.1 Middle class and above 5.6* 8.5* 8.9* 13.7* 8.1 10.4 Difference 4 0.3 1.7 -4.9 7.1 Total Below middle class 6.9 12.8 15.6 14.3 7.4 14.2 Middle class and above 8.6 9.4 12.0 16.4 7.8 12.6 Difference 4 1.7 -3.4 -3.6 2.1 National mean 8.2 10.2 12.9 15.5 7.3 11.7 N 1,359 1,001 1,300 1,239 1 For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. 2 "Below middle class" includes low-middle class and low class. "Middle class and above" includes middle class, middle-high class, and high class. 3 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 4 Difference = "Middle class and above" - " Below middle class." *N<25. 100 1996. Therefore, in 1996, people with higher-class identity were more likely than those with lower-class identity to support the DPP. Summary As demonstrated above, second-generation males had more positive attitudes than those in other generations toward the DPP between 1987 and 1993. However, third- generation males were the most loyal supporters of the DPP in 1996. People with college and above tended to support the DPP between 1987 and 1996, except in 1993. People with a less than high school education had a more positive attitude toward the DPP in 1993. The DPP received more support from educated people in the two younger generations in 1996. Because the KMT has been the ruling party since 1949, one might expect that civil servants would be less likely than people of other vocations to support the DPP. However, it was found that support of the DPP among younger state employees has increased since 1993. The DPP also attracted more people with lower-class identity in the two younger generations in 1990 and 1993. However, there was no consistent pattern of DPP support among different socioeconomic classes. Social Bases of Support for the NP As demonstrated in Chapter 4, DPP partisans were less likely than KMT partisans to change their partisan preferences when they had to make a choice among three major parties. However, support of the KMT declined when the NP entered the party system, 101 and partisan preferences among third-generation voters were most sensitive to the emergence of the NP. In this section, the social bases of the NP are examined. The NP Received More Support From Females Than Males As shown at the bottom of Table 5-9, more than one out of three males defected from the KMT in 1996. However, Males’ support of the DPP was relatively stable even after the NP entered the electoral competition. When voters faced three major parties in 1996, 19.4 percent of males had a positive attitude toward the KMT, 14 percent of them preferred the DPP, and 10.4 percent were NP partisans. However, the NP received more support from females than males. The proportion of KMT partisans among females was 18.7 percent, the proportion of female DPP supporters was 9.2 percent, and 13 percent of females preferred the NP in 1996. When facing three major parties, voters were most likely to support the KMT. However, the NP became the second choice for females. Table 5-9 also indicates that third-generation females were more likely than those in other generations to change their partisan preferences when selecting among three major parties. When third-generation females had to choose among three major parties, more than one out of two voters defected from the KMT and chose the NP. The proportion of KMT partisans among third-generation females was 10.2 percent, the proportion of DPP supporters was 11 percent, and 16.7 percent preferred the NP in 1996. Thus, third-generation females were more likely to support the NP than the other two major parties. Third-generation males’ partisan preferences also were vulnerable when they faced three major parties. When males had to make a choice among these three 102 Table 5-9. Two Measurements of Partisan Preference, by Gender, 1996 Generation Gender Two Major Three Major Parties 1 Parties 2 KMT DPP KMT DPP NP 1961 and After Male 28.1 19.7 15.8 15.0 12.0 Female 30.1 17.4 10.2 11.0 16.7 Difference 3 -2.0 2.3 5.6 4.0 -4.7 1943 to 1960 Male 22.7 16.3 15.8 14.9 7.2* Female 34.2 12.6 22.9 7.3* 11.2* Difference 3 -11.5 3.7 -7.1 7.6 -4.0 1942 and Before Male 46.5 10.4* 33.6 10.4* 12.8* Female 40.2 11.2* 31.6 8.4* 7.4* Difference 3 6.3 -0.8 2.0 2.0 5.4 Total Male 30.3 16.4 19.4 14.0 10.4 Female 33.5 14.5 18.7 9.2 13.0 Difference 3 -3.2 1.9 0.7 4.8 -2.6 National mean 31.8 15.5 19.1 11.7 11.7 N 1,239 1,158 1 Items for 1996 were worded as like/dislike questions for two major parties. 2 Items for 1996 were worded as like/dislike questions for three major parties. 3 Difference = Male - Female. *N<25. 103 parties, 15.8 percent of them preferred the KMT, 15 percent held a positive attitude toward the DPP, and 12 percent supported the NP. As to second-generation voters, females’ support for the DPP also was affected by the appearance of the NP. Specifically, 22.9 percent of them supported the KMT, 7.3 percent preferred the DPP, and 11.2 percent had positive attitude toward the NP. Males’ support for the DPP was relatively stable. That is, 15.8 percent of them were KMT partisans, 14.9 percent supported the DPP, and only 7.2 percent preferred the NP. Among first-generation males, 33.6 percent of them supported the KMT, 10.4 percent of them preferred the DPP, and 12.8 percent were NP partisans. Further, 31.6 percent of first-generation females were KMT partisans, 8.4 percent preferred the DPP, and 7.4 percent had a positive attitude toward the NP. Overall, the NP received the most support from third-generation females and first generation males. However, second-generation males were less likely to support the NP than the other two major parties. Educated People Tended to Support the NP In 1996, people with higher educational levels were more likely than those with less education to support the NP. As shown at the bottom of Table 5-10, 21 percent of people with college and above supported the NP in 1996; 13.7 percent of them supported the KMT and the DPP. As to people with a high school education, 16 percent had a positive attitude toward the KMT, 10.9 percent preferred the DPP, and 12.7 percent were NP partisans. People with less than a high school education were most likely to support the KMT. More than 25 percent of these voters supported the KMT, 11.1 percent 104 preferred the DPP, and only 4.2 percent had a positive attitude toward the NP. Because most legislative candidates of the NP hold a doctoral degree and one major platform of the NP is maintaining socioeconomic justice, one might expect that the NP would be more likely to gain support from better educated people. From Table 5-10, it can be seen that the newly emerged NP attracted better educated people from each generation. As to third-generation voters, people with less than a high school education were most likely to support the KMT in 1996; 18.7 percent of them preferred the KMT, 14.3 percent supported the DPP, and only 4.4 percent had a positive attitude toward the NP. However, people with a high school education or above were more likely to support the NP; 15.2 percent of them preferred the NP, 12.9 percent had a positive attitude toward the KMT, and 10.7 percent were DPP partisans. A8 to voters with college and above, 17.9 percent were NP partisans, 15.3 percent supported the DPP, and only 10.7 percent preferred the KMT. As to second-generation voters, only those with college and above were more likely to support the NP. Only 3.5 percent of people with less than a high school education had a positive attitude toward the NP, and just 9.2 percent of people with a high school education preferred the NP. However, more than one out of five people with college and above had a positive attitude toward the NP. Among first-generation voters with college and above, more than one out of three supported the NP, whereas only 5 percent of voters with less than a high school education had positive attitude toward the NP, and 11 percent with a high school education had positive attitudes toward the NP. 105 Table 5-10. Two Measurements of Partisan Preference, by Level of Education, 1996 Generation Level of Education Two Major Three Major Parties 1 Parties 2 KMT DPP KMT DPP NP 1961 and After Less than high school 28.0 16.0* 18.7* 14.3* 4.4* High school 30.2 15.7 12.9 10.7* 15.2 College and above 28.3 23.2 10.7* 15.3 17.9 Difference 3 0.3 7.2 -8.0 1.0 13.5 1943 to 1960 Less than high school 28.1 14.0 22.6 9.5* 3.5* High school 29.5 13.7* 16.9* 12.0* 92* College and above 27.3 17.0* 15.1* 14.0* 22.1* Difference 3 -0.8 3.0 -7.5 4.5 18.6 1942 and Before Less than high school 41.6 11.9* 33.1 11.3* 5.0* High school 55.2* 6.9* 37.0* 7.4* 11.1* College and above 45.9* 8.1* 27.3* 3.0* 36.4* Difference 3 4.3 -3.8 -5.8 -8.3 31.4 Total Less than high school 33.0 13.6 25.6 11.1 4.2* High school 31.7 14.4 16.0 10.9 12.7 College and above 30.0 19.8 13.7 13.7 21.0 Difference 3 -3.0 6.2 -11.9 2.6 16.8 National mean 31.8 15.5 19.1 11.7 11.7 N 1,239 1,158 1 Items for 1996 were worded as like/dislike questions for two major parties. 2 Items for 1996 were worded as like/dislike questions for three major parties. 3 Difference 2 "College and above" - "Less than High School." *N<25. 106 Therefore, in 1996, peOple with college and above were most likely to support the NP. Third-generation voters with a high school education also were likely to support the NP. The NP Attracted Younger Civil Servants As shown at the bottom of Table 5-11, government employees were more likely to support the KMT than the other two parties; 27.4 percent of them preferred the KMT, but 18.5 percent of them had a positive attitude toward the NP. Only 10.4 percent had a positive attitude toward the DPP. However, as to third-generation government employees, 20.5 percent supported the NP, 13.6 percent preferred the KMT, and 11.4 percent had a positive attitude toward the DPP. The emergence of the NP had a significant effect on KMT among first-generation state employees, and civil servants in the third generation were more likely to support the NP than they were to two other major parties. The NP received less support from state employees in the second generation than the other two parties. Only 8.7 percent of them supported the NP, whereas 26.1 percent preferred the KMT and 13 percent supported the DPP. Government employees in the first generation were the group with the most positive attitude toward the KMT. When they encountered three major parties, 42.2 percent of them preferred the KMT, 26.7 percent supported the NP, and only 6.7 percent had a positive attitude toward the DPP. 107 Table 5-11. Two Measurements of Partisan Preference, by Vocation, 1996 Generation Vocation Two Major Three Major Parties ‘ Parties 2 KMT DPP KMT DPP NP 1961 and After Government employee 31.1* 17.8* 13.6* 11.4* 20.5* Other 28.9 18.6 13.1 13.3 13.7 Difference 3 2.2 -0.8 0.5 -1.9 6.8 1943 to 1960 Government employee 41 .7* 14.6* 26.1* 13.0* 8.7* Other 26.8 14.5 18.4 11.0 9.2 Difference 3 14.9 0.1 7.7 2.0 -0.5 1942 and Before Government Employee 60.8 9.8* 42.2* 6.7* 26.7* Other 39.5 11.0* 30.3 10.3* 6.3* Difference 3 21.3 -1.2 11.9 -3.6 20.4 Total Government employee 45.1 13.9 27 .4 10.4* 18.5 Other 30.0 15.7 18.0 11.9 10.8 Difference3 15.1 -1.8 9.4 -l.5 7.7 National mean 31.8 15.5 19.1 11.7 11.7 (N) 1,239 1,158 1 Items for 1996 were worded as like/dislike questions for two major parties. 2 Items for 1996 were worded as like/dislike questions for three major parties. 3 Difference = "Government Employee" - "Other." * N<25. 108 Therefore, when choosing among three major parties, civil servants in the third generation were most likely to support the NP. The NP also received support from government employees in the first generation, but the KMT was still their first choice among the three major parties. The NP as a Middle-Class Party As shown at the bottom of Table 5-12, people with middle-class (and above) identity were more likely than their lower-class counterparts to change their partisan preferences when they encountered three major parties. Specifically, 17.5 percent of them had a positive attitude toward the KMT, 12.3 percent preferred the DPP, and 13.6 percent supported the NP. As to voters with lower-class identity, 23 percent preferred the KMT, 10.9 percent supported the NP, and only 7.1 percent had a positive attitude toward the DPP. As to third-generation voters with middle-class (and above) identity, the proportion of KMT partisans was 12.1 percent; 12.8 percent of them preferred the DPP. This group was more likely to support the NP than the other two parties; the proportion of NP partisans was 15 .2 percent. However, people with lower-class identity were more likely to support the two major parties other than the NP. That is 15.9 percent of them had a positive attitude toward the KMT, another 15.9 percent preferred the DPP, but only 11 percent supported the NP. As to second-generation voters with middle-class identity (and above), the proportion of KMT partisans was 19 percent; whereas 11.7 percent preferred the DPP. They were slightly less likely to support the NP; the proportion of NP partisans was 11.4 109 Table 5-12. Two Measurements of Partisan Preference, by Subjective Socioeconomic Status, 1996 Generation Subjective Two Major Three Major Socioeconomic Status 1 Parties 2 Parties 3 KMT DPP KMT DPP NP 1961 and After Below middle class 279* 20.9* 15.9* 15.9* 110* Middle class and above 29.2 18.5 12.1 12.8 15.2 Difference 4 1.3 -2.4 3.8 3.1 -4.2 1943 to 1960 Below middle class 216* 14.4* 18.7* 9.9* 33* Middle class and above 30.1 14.9 19.0 11.7 11.4 Difference 4 8.5 0.5 -0.3 -1.8 -8.1 1942 and Before Below middle class 46.1 6.6* 37.9 6. l * 76* Middle class and above 43.8 13.7* 30.8 12.0* 13.5* Difference 4 -2.3 7.1 7.1 -5.9 -5 .9 Total Below middle class 30.9 14.3 23.0 10.9 7.1* Middle class and above 31.9 16.4 17.5 12.3 13.6 Difference 4 1.0 2.1 5.5 -1.4 -6.5 National Mean 31.8 15.5 19.1 11.7 11.7 (N) 1,239 1,158 1 ”Below middle class" includes low-middle class and low class. "Middle class and above" includes middle class, middle-high class, and high class. 2 Items for 1996 were worded as like/dislike questions for two major parties. 3 Items for 1996 were worded as like/dislike questions for three major parties. 4 Difference = "Middle class and above" - "Below middle class." *N<25. 110 percent. People with lower class identity also were more likely to support the KMT than the other two parties listed. Whereas 18.7 percent of them preferred the KMT, just 9.9 percent had a positive attitude toward the DPP, and a mere 3.3 percent supported the NP. As to first-generation voters with middle-class (and above) identity, the proportion of KMT partisans was 30.8 percent; in contrast, 12 percent of them supported the DPP and 13.5 percent the NP. As to people in this group with lower-class identity, 37.9 percent had a positive attitude toward the KMT, 6.1 percent preferred the DPP, and 7.6 percent supported the NP. Thus, it seems that the NP was more likely to attract people with middle-class (and above) identity in the third and first generations. Summary As demonstrated above, when the NP entered the party system in Taiwan, it significantly affected KMT support. Younger females were more likely to support the NP in 1996. The NP also received more support from educated people. Civil servants in the first and third generations were more likely to have a positive attitude toward the NP. In addition, the NP attracted people with middle-class (and above) identity. Conclusion The social bases of support for the KMT gradually changed between 1987 and 1996. Females in the two younger generations became more likely to support the KMT since 1993. Among the three generations, however, first-generation males have been the 111 group most supporting of the KMT since 1990. As to the effect of people’s educational levels on their support of the KMT, better educated people were more likely to prefer the KMT between 1987 and 1993, but the KMT gained support from people with less than a high school education in 1996. Government employees were still loyal supporters of the KMT, but that party received less support from civil servants in the third generation. As to people’s subjective socioeconomic identity, the KMT received more support from people with middle- and upper-class identity between 1987 and 1993, but the class difference in KMT support was small in 1996. As to the social bases of support for the DPP, it was found that the DPP received the most support from second-generation males between 1987 and 1993. In 1996, third- generation males were the most loyal supporters of the DPP among the six gender groups. As to educational level, people with college and above tended to support the DPP in 1987, 1990, and 1996. However, when different generations were examined, it was found that people with less than a high school education were most likely to support the DPP in 1993, but the DPP received more support from better educated people in the two younger generations in 1996. Because the KMT has been the ruling party since 1949, it was expected that the DPP would receive less support from civil servants. However, support of the DPP by younger government employees has increased since 1993. The DPP also attracted more people with lower-class identity in the two younger generations between 1990 and 1993. However, there was no consistent pattern of DPP support by different social classes. The emergence of the NP significantly affected KMT support. Younger females were more likely to support the NP in 1996. Better educated people were also more likely 112 to support the NP. The NP also received support from civil servants in the first and third generations. People with middle-class (and above) identity were also more likely to support the NP. In examining whether gender, educational level, vocation, and subjective socioeconomic status affected the partisan preferences of people in Taiwan, it was found that the differences in people’s partisan preferences based on these social characteristics became smaller in 1996. As to these social characteristics, there is no clear social cleavage in the electorate in Taiwan. This might be good news for the stability of democracy in Taiwan because the country’s political arena might not be an integrated environment. An integrated environment indicates that "the lives of the members are encased within ideologically linked activities" (Lipset 1981: 74). Lipset argued that parties of integration will isolate their social base from cross-pressure, [so it will] clearly undermine stable democracy, which requires shifts from one election to another and the resolving of issues between parties over long periods of time. . . .The necessary rules of democratic politics assume that conversion both ways, into and out of a party, is possible and proper, and parties which hope to gain a majority by democratic methods must ultimately give up their integrationist emphasis. (Lipset 1981: 75) Because the partisan preferences of people with different social backgrounds might be changeable, political parties have an good incentive to present more attractive issues to gain support from the electorate. Therefore, each party will have a chance to win the election. In the next chapter, I examine how important issues, such as national identity, "concern about money politics," and democratic reform, have affected the partisan preferences of people in Taiwan. 113 CHAPTER 6 ISSUE PREFERENCES, NATIONAL IDENTITY, AND PARTISAN PREFERENCE The standard view is that party identification, as a long-term psychological attachment, affects individuals’ evaluations of issues and candidates, but in turn it is largely unaffected by such evaluations, except in dramatic circumstance (Abramson 1983; Abramson et al. 1994; Campbell et a1. 1960). Therefore, party identification is an exogenous variable to explain people’s political attitudes and behaviors. In recent years, however, there have been challenged to the claim that party identification is unaffected by policy preferences. Jackson (1975) and Franklin and Jackson (1983) argued that party identifications are subject to change as voters’ issue preferences change. Similarly, Markus and Converse (1979) demonstrated that individuals’ past votes affect their party identification. Fiorina (1981) argued that partisan identification is a "running tally" of past experiences, and it is adjusted based on the performance of the incumbent party. Therefore, party identification was used as an endogenous variable in the model. Because the measurements employed by the NTU research team might tap people’s partisan preference, I treated partisan preferences as dependent variables in my 114 models in this chapter. I examined how national identity, political values, and other social forces affect people’s partisan preference. National Identity, Political Values, and Partisan Preference Campbell and his colleagues (1960: 121) defined party identification as "the individual’s affective orientation to an important group-object in his environment." According to political socialization research conducted in the United States, most people acquire party identification at a very early age, and they formulate their partisan loyalties before they acquire much information about politics (Abramson 1983; Greenstein 1965; Hess and Tomey 1965; Hyman 1959). However, when applying the concept of party identification outside the United States, one must know the historical and political context of the country in question. It is hard for the electorate to formulate a stable and long-term affective attachment to a political party in a one-party-dominant regime because the electorate is given no choice in elections. In Taiwan, opposite forces, i.e., Tangwai, began to challenge the KMT after the 1977 local elections (Huang 1992: 43). Tangwai organized a quasi-party organization, the Tangwai Candidates Campaign Committee, to endorse candidates and coordinate election activities in the 1983 legislative election (Sheng 1986: 86-89; Tien 1996: 11). In 1987, Tangwai established the DPP and used an official organization to challenge the KMT in the elections. The electorate in Taiwan has had opportunities to choose candidates outside the KMT for two decades. Hence, voters might have formulated a psychological attachment to Tangwai and supported its 115 candidates in the elections. It is also possible that some voters began to identify themselves as KMT if they thought the KMT was better than Tangwai. Therefore, people’s party identification is a relatively new phenomenon in Taiwan. I argue that competitive elections provide great opportunities for voters to formulate a psychological attachment to political parties. In return, voters use this psychological factor in evaluating candidates and issues. I was interested in examining whether voters’ partisan preferences shape, or are shaped by, national identity in Taiwan. However, it is not easy to distinguish the reciprocal causal relationship between two variables in social science. Most political scientists have applied panel studies to examine non-recursive models of the relationship among party identification, other political attitudes, and behaviors. Unfortunately, no panel data was available to test my model in Taiwan. Instead, I employed an ordinary least squares (OLS) estimator to examine the relationship between partisan preference and other attitudinal variables and demographics. According to Barnes and his colleagues (1985), the political factor, i.e., ideology, provides a good explanation of the development of partisanship in Spain. Fleury and Lewis-Beck (1993) argued that ideology is more important than party identification in affecting vote choice. Miller and Shanks (1996: 15) presented a sequence of six causal stages to explain people’s vote choices: In sequence, the explanatory variables we have assigned to each stages are defined in terms of (1) stable social and economic characteristics, (2) partisan identification and policy-related predispositions, (3) preferences concerning current policy issues and perceptions of current conditions, (4) explicit evaluations of the incumbent President, (5) evaluations of the candidates’ personal qualities, and (6) prospective evaluations of the candidates and the parties. 116 Sears and his colleagues (1980: 671) also suggested that "people acquire stable affective preferences through conditioning in their preadult years, with little calculation of the future costs and benefits of these attitudes. The most important of these are presumably some rather general predispositions, such as party identification, liberal or conservative ideology, nationalism or racial prejudice." Therefore, people’s social and economic characteristics precede their partisan preferences, and partisan preferences and policy-related predispositions might affect each other. In this chapter, I examine which factors affect people’s partisan preferences, and I discuss how party affiliation affects people’s national identity in Chapter 7. Some political scientists have treated partisan preference as an endogenous variable and explained how demographics and other attitudinal variables might affect the partisan preferences of people in Taiwan (Chen 1992; Chu 1992; Ho 1994; Hsieh and Niou 1996; Lin 1998; Lin et al. 1998; Liu 1987; Shyu 1993; Tsai 1997; You 1994b). Lin and his colleagues provided comprehensive models to explain how people’s demographic background and political attitudes affect their partisan preferences. Their models included people’s attitudes toward national identity, money politics, procedural democratic values, and their gender, age, ethnic origin, vocation, and subjective social economic status. My model is based on that of Lin and his colleagues. It includes explanatory variables such as the individual ’8 ethnicity, gender, occupation, and education. Some attitudinal variables were also important to my research; hence, I included the respondent’s subjective perception of socioeconomic status, democratic values, concerns about "money politics," and national identity. To examine whether there are differences in partisan preferences among different generations resulting from distinctive formative socialization 117 experiences, I used political generation as a mediator variable to separate the sample into three subgroups and estimated coefficients for each political generation separately. The model used in this study is as follows: Y1 = :60 +761X1+fi2X2 +fi3X3 +764X4 +fi5X5 +fi6X6 +:67X7 +fi8X8 +61 (6'1) where Yl is party identification X 1 is national identity X 2 is democratic values X 3 is "concern about money politics" X 4 is a dummy variable for mainlander X 5 is the educational level X 6 is subjective socioeconomic status X 7 is a dummy variable for government or KMT official employee X 8 is a dummy variable for female. First, I examined the relationship between party identification and other attitudinal variables. Because two major platforms of the DPP are adopting more rapid democratic reforms and emphasizing Taiwanese identity and self—determination (Cheng and Hsu 1996; Hu and Chu 1992; Huang 1992), I hypothesized that people who prefer democratic reforms and/or have Taiwanese identity are more likely to support the DPP. Another key issue, concern about money politics, is also critical to understanding the formation of party loyalty in Taiwan. As Chu and Lin (1996: 87) argued, "during the ’go- go’ years between 1986 and 1990, skyrocketing real estate and stock prices contributed to a serious deterioration in the fair distribution of wealth. This so-called money game substantially increased the wealth of asset owners but impoverished the lower— and lower- middle-income families." In addition, President Lee had to seek support of the business community and local factions to win a power struggle with mainlander elites within the KMT. "As a result, long tainted elections for local officials seeped into elections for 118 national representative bodies" (Chu and Lin 1996: 88). Therefore, Lin and his colleagues (1996) maintained that a new socioeconomic-justice issue, i.e., concern about money politics, is salient in the electorate in Taiwan. I predicted that people who are concerned about "money politics" are less likely to prefer the KMT. I also explored the social bases of party support in Taiwan. 1 hypothesized that people with higher socioeconomic status are more likely to support the opposition parties because both the DPP and the NP advocate socioeconomic justice (Lin et al. 1998). In addition, the KMT is still the ruling party, so government employees are more likely than those in other vocations to support it. As I show in Chapter 4, mainlanders are more likely than Taiwanese to support the KMT because of their political experiences and collective memory. In the following section, I briefly discuss the variables in the above-mentioned equation.18 Partisan Preference and National Identity 1 employed a 7-point scale to measure the direction and intensity of respondents’ preferences toward the DPP (1) and the KMT (7).19 I used the following questions to construct a national identity scale. For 1993, the question was "Which ethnic group do you think you belong to? Chinese or Taiwanese?" There were five possible answers: Taiwanese-strongly agree (1), Taiwanese-just agree (2), both Taiwanese and Chinese (3), Chinese-just agree (4), and Chinese-strongly agree (5). The 1996 question was similar to the 1993 one: "In Taiwan, some people consider themselves Chinese and others consider themselves Taiwanese. Do you consider yourself Taiwanese or Chinese?" There were ‘8 For a detailed discussion of the following variables, see Lin et al. (1998). My discussion is based on that article. ‘9 For the wordings of questions in partisan preference measurements, see Chapter 3. In 1996, there were three major parties, so I compared one with another. 119 two intensity levels (strongly agree and just agree) for the "both Taiwanese and Chinese" answer, so I collapsed them into one category. The higher the value respondents had, the more they identified themselves as Chinese. Whenever one conducts a secondary data analysis, he or she can use only those variables available in the data sets. Therefore, the measurement of national identity used in this research might be too simple to capture the underlying meaning of national identity. However, most research in Taiwan has used the same measurement as this research did, and some studies (Chen 1998; Hawang 1996) regarding the 1996 presidential election in Taiwan also have shown that this variable significantly affected people’s vote choice. Therefore, this variable was employed in this research to measure people’s national identity. In addition, this variable is more straightforward than an index, so it could serve either as a dependent variable or an independent variable in this study. Democratic Values and Concern About "Money Politics" 1 employed the following five questions to construct an index of democratic values: 1. Government has the authority to decide what opinions can be disseminated in society. 2. The existence of too many organizations will adversely affect the stability and harmony of a community. 3. The existence of too many political parties leads to political turmoil. 4. Frequent interference from the legislative branch in the actions of the executive branch paralyzes government. 120 5. In hearing a major case concerning public order, a judge should follow the opinion of the executive branch of government. Respondents answered these questions using a 5-point scale ranging from strongly disagree (1) to strongly agree (6). I recoded each question so that a higher score indicated that a respondent had a higher level of democratic values. The reliability coefficients (Cronbach’s alpha) of the scale were .59 for 1993 and .65 for 1996. I included respondents who answered at least three of the five above-mentioned questions in constructing this scale. As to people’s concern about "money politics," I created an index by calculating the average score on two questions: 1. Currently, most government policies are heavily influenced by big business conglomerates at the expense of the public interests. 2. Most legislators protect the interests of big business conglomerates; only a few speak for the public. Possible answers for both questions ranged from strongly disagree (1) to strongly agree (6). Therefore, the higher one’s score on this index, the more he or she concerned about "money politics." Social Forces I also included other demographic variables in the model. I coded ethnic origin as 1 representing mainlanders and 0 representing Taiwanese. There were six categories for education: 0 = no schooling, 1 = elementary school, 2 = junior high school, 3 = senior high school, 4 = college, and 5 = graduate school. As to the respondent’s subjective 121 socioeconomic status, 1 used 0 to represent low—middle and low class, and 1 to represent middle to high social class. I recoded the respondent’s vocation as 1 representing governmental or KMT employees and 0 representing other. Gender was a dummy variable; 1 =female and 0 = male. Determinants of Partisan Preference Factors that affected people’s partisan preferences in 1993 and 1996 are shown in Table 6-1. For 1993, I used one equation to explain how attitudinal variables and demographics affected people’s partisan preferences for the DPP versus the KMT. There were three major parties in 1996, so I estimated three equations. As shown in the first column in Table 6-1, all coefficients were significant and met my theoretical expectations; there was no severe multicollinearity problem in the 1993 model. People with Chinese identity were more likely to support the KMT. The signs for two other attitudinal variables were negative, meaning that those who were concerned about "money politics" and those with higher democratic values were more likely to support the DPP. As mentioned above, two major platforms of the DPP are to advocate democratic reforms and to promote Taiwanese identity. Therefore, the results indicate that these two issues were salient to the electorate in deciding their DPP versus KMT partisan preferences in 1993. Another issue was socioeconomic justice. In this study, it was found that concerns about money politics was significant in both 1993 and 1996, so this issue provided a new dimension for major parties in electoral politics in Taiwan. However, whether voters used these three issues to formulate their partisan preferences or employed 122 Table 6-1. Determinants of Partisan Preferences Among the Electorate, 1993 and 1996 1993 1996 1996 1996 DPP v. KMT DPP v. KMT DPP v. NP KMT v. NP Coefficient Coefficient Coefficient Coefficient (Std. Error) (Std. Error) (Std. Error) (Std. Error) National .281 .192 .284 .090 identity (.035)*** (.025)*** (.025)*** (.026)*** Concern about -.173 -.081 .018 .103 “money politics” (.032)*** (.024)*** (.024) (.025)*** Democratic values -.299 -.216 -.029 .182 (.042)*** (.028)*** (.028) (.029)*** Mainlander .661 .361 .785 .406 (.119)*** (.078)*** (.074)*** (.07 8)*** Education .086 -.072 .014 .083 (.037)* (.026)** (.026) (.027)** Socioeconomic .235 .046 .1 17 .097 status (.107)* (.070) (.069) (.072) Government/KMT .579 .232 .019 -. 189 official/employee (.1 18)*** (.087)** (.085) (.088) Female .208 .018 .214 .156 (.083)* (.057) (.056)*** (.058)** Intercept 4.892 4.940 4.940 1 .973 (.256)*** (.181)*** (.181)*** (.184)*** N 982 1,028 965 967 Adj. R2 .264 .176 .279 .138 S.E.E. 1.2608 0.8815 0.8375 0.8750 Conditional 18.417 19.072 19.002 19.050 index *: p S .05, **: p _<_ .01, ***: p S .001, 2-tailed test. Entries are unstandardized coefficients (standard errors are in parentheses). 123 their partisan preferences to take stances on these three issues cannot be answered by this equation. It is possible that people used their partisan preferences as a filter to gain information and take stances on issues. As to other social forces, mainlanders, educated people, government employees, and females were more likely to support the KMT in 1993. As Lipset (1981) noted, females are more conservative, so they were more likely to support conservative parties. The KMT also received support from the upper class in 1993. In 1996, there were three major parties, so I compared one with another for each item and estimated three equations. In the second column of Table 6-1, the DPP is compared with the KMT. Two variables, socioeconomic status and female, were not significant, so they were not good variables for explaining differences in partisan preferences between these two major parties in 1996. Further, respondents’ educational level was significant but the sign was negative, indicating that people with a higher level of education were more likely to prefer the DPP in 1996. Table 6-1 also indicates that national identity, concern about money politics, democratic values, vocation, and ethnic origin were important factors in explaining people’s preference for the DPP vs. KMT. The emergence of the NP changed the outlook of the two-party system in Taiwan. As shown in the third column of Table 6-1 show, national identity, ethnic origin, and gender were significant variables in explaining the difference in supporters of two opposition parties. Not surprisingly, people with Chinese identity were more likely to support the NP; mainlanders and females also preferred the NP. This result is not surprising because most elites in the NP are mainlanders and most leaders in the DPP are Taiwanese. It is well known that mainlanders tend to identify themselves as Chinese, 124 whereas Taiwanese tend to identify themselves as Taiwanese. However, supporters of the two parties did not differ significantly in their concern about money politics and democratic values, indicating that the two parties can cooperate with each other on these two issues to compete with the KMT. The last column in Table 6-1 indicates that national identity, concern about money politics, democratic values, ethnic origin, education, and gender were important variables in distinguishing voters’ preferences for the KMT versus the NP. People with Chinese identity preferred the NP, but the magnitude of this coefficient (.090) was very small. In addition, people who were concerned about "money politics" and those with higher scores on the democratic values index were more likely to support the NP. Mainlanders, educated people, and females also were more likely to support the NP. It is interesting that the NP gained females’ support in 1996, just as the KMT did in 1993. As discussed in Chapter 4, it is possible that differences in partisan preferences among the three generations might have been caused by distinctive formative socialization experiences. Therefore, I used political generation as a mediator variable to separate the sample into three subgroups, and examined how political attitudes and social forces affected the partisan preferences of voters from these three generations. As shown in Table 6-2, all coefficients met the theoretical expectation, and there was no multicollinearity for any equation in 1993. The first column of Table 6-2 indicates which factors affected the partisan preferences of third-generation voters’ (born in 1961 and after) in 1993. It was found that national identity, concern about "money politics," democratic values, and ethnic origin were significant in the formation of third-generation voters’ partisan preferences. However, their educational level, subjective socioeconomic 125 status, vocation, and gender were not significant. Therefore, attitudinal variables played a more important role than other personal characteristics in shaping third-generation voters’ partisan preferences. Their ethnic origins also played a crucial role in this regard. A8 Liu (1996: 230) pointed out, the "fact that mainlanders and Taiwanese share similar political contexts (e.g., schools) at some times but maintain separate contexts (e.g., family) at others indicates that ethnic differences may account for some of the political differences between these two groups." Looking at the second column of Table 6-2, one can see that an additional variable, i.e., vocation, was significant. Compared with the first equation, coefficients of concern about money politics and democratic values were larger in the second equation, indicating that these two variables played more important roles in shaping second-generation voters’ (born between 1943 and 1960) partisan preference than they did in influencing third-generation voters. As to voters in first generation (born in 1942 and before), three attitudinal variables were significant as well. In addition, their ethnic origin, socioeconomic status, and vocation helped them form their partisan preferences. Therefore, in 1993, national identity, concern about "money politics," democratic values, and ethnic origins significantly influenced voters in three generations in deciding their preference for the DPP versus KMT. In 1993, the party system in Taiwan experienced a great change; the NP entered the 1995 legislative election, receiving 13.7 percent of the vote and 46 seats. Therefore, I was interested in examining which factors affected people’s preferences for the NP. Because there were three major parties in the 1995 legislative election, I compared the three major parties to one another pairwise and estimated three equations separately. 126 Table 6-2. Determinants of Partisan Preferences (DPP versus KMT), by Generation, 1993 1961 and After 1943 to 1960 1942 and Before Coefficient Coefficient Coefficient (Std. Error) (Std. Error) (Std. Error) National .294 .266 .284 identity (.059)*** (.054)*** (.078)*** Concern about -. 133 -.209 -. 167 “money politics” (.055)*** (.052)*** (.063)** Democratic values -. 190 -.290 -.500 (.070)** (.067)*** (.086)*** Mainlander .523 .584 .652 (.185)** (.222)** (.258)* Education .142 .097 .1 1 1 (.086) (.068) (.078) Socioeconomic .111 .128 .583 status (.184) (.183) (.194)** Government/KMT .221 .602 .793 official/employee (.215) (.201)** (.232)** Female .215 .248 .256 (.130) (.137) (.189) Intercept 4.223 5.015 5.410 (.466)*** (.442)*** (.490)*** N 380 387 215 Adj. R2 .171 .227 .424 S.E.E. 1.2319 1.2960 1.2306 Conditional 20.864 19.197 17.066 index *: p S. .05, **: p S .01, ***: p S .001, 2-tailed test. Entries are unstandardized coefficients (standard errors are in parentheses). 127 Similar patterns were evident across the three generations in 1996 (see Table 6-3). There was no severe multicollinearity in the three models. However, two of the major parties became more similar, coefficients were not significant in the models. For third- generation voters, concern about "money politics" was not significant, but national identity and democratic values were still important in voters’ distinguishing between these two major parties. As shown in Table 6-3, national identity still had a positive effect on the three generations’ preferences for the DPP versus KMT, but compared with its effect in 1993, the effect of national identity decreased in 1996 for the three generations. In addition, ethnic origin only had a positive effect on third-generation voters’ DPP versus KMT preference, indicating that voters in the two older generations formulated their preferences for the KMT and the DPP based solely on their national identity. Those who identified themselves as Taiwanese preferred the DPP. On the other hand, voters who identified themselves as Chinese supported the KMT. As to third- generation voters, their ethnic origins were still important when they had to choose between the KMT and the DPP. It is possible that voters in the younger generation acquired their partisan preferences from their parents, so their ethnic origins still affected their partisan affiliation. In addition, education affected third—generation voters’ partisan preferences; that is, people with higher education were more likely to prefer the DPP than the KMT. Hence, the DPP attracted more young educated voters in 1996. From Tables 6-2 and 6-3 it is evident that voters in the three generations used their political attitudes to formulate their preferences for the DPP versus the KMT. Because the Tangwai/DPP emerged as a quasi-organized force in the elections for more than a decade, voters experienced competitive elections and formulated their 128 Table 6-3. Determinants of Partisan Preferences (DPP versus KMT), by Generation, 1996 1961 and After 1943 to 1960 1942 and Before Coefficient Coefficient Coefficient (Std. Error) (Std. Error) (Std. Error) National .175 .210 .149 identity (.037)** * (.038)*** (.076)* Concern about -.026 -.079 -. 191 “money politics” (.035) (.036)* (.071)** Democratic values -.231 -.200 -.204 (.042)** (.043)*** (.079)* Mainlander .387 .232 .414 (.107)*** (.134) (.240) Education -.1 10 -.053 -.015 (.050)* (.041) (.076) Socioeconomic .056 .1 16 -.062 status (.108) (.104) (.190) Government/KMT .133 .150 .322 official/employee (.137) (.141) (.232) Female -.003 .135 -.086 (.079) (.086) (.190) Intercept 4.906 4.694 5.539 (.279)*** (.278)*** (.527)*** N 483 363 182 Adj. R2 .137 .165 .153 S.E.E. 0.8481 0.7918 1.1184 Conditional 20.712 19.123 18.509 index *: p S .05, **: p S .01, ***: p S .001, 2-tailed test. Entries are unstandardized coefficients (standard errors are in parentheses). psychological attachment to particular political parties. As Converse (1976: 12-13) pointed out, party identification is "a function of the length of time that the individual has felt some generalized preference for a particular party and has repetitively voted for it" (see also Markus and Converse 1979). The KMT opened local elections in 1951, and national parliament elections began in 1970. These elections became a great vehicle for opposition forces to organize their support and resources to challenge the KMT (Chu 1992; Huang 1995). Chu (1992: 50-1) also argued that "The opposition have turned campaign process into an effective medium of ’resocialization’ for fostering the growth in popular demand for democratic legitimacy." Therefore, election campaigns served as a socialization agent to the electorate to shape their partisan preferences. As mentioned in Chapter 1, the emergence of the NP was brought about by some junior KMT legislators who were dissatisfied with the ambiguous stance taken by President Lee on the issue of unification with mainland China versus Taiwan independence (UM-TI issue). Therefore, we can expect that the two opposition parties, i.e., the DPP and the NP, would take opposite stances on the UM-TI issue (Liu 1995).20 Table 6-4 indicates that national identity and ethnic origin had significant effects on people’s preference for the DPP versus NP. Voters’ concern about money politics did not help them perceive significant differences between the two opposition parties in 1996. However, second-generation voters employed their procedural democratic values to distinguish between the two opposition parties, and people with higher scores on this 2° According to Hsieh and his colleagues (1995), voters placed the NP near the middle position on the UM-TI issue, the DPP on the left (TI), and the KMT on the right (UM), using an 11-point scale (from 0 to 10). 130 index were more likely to prefer the DPP than the KMT. Additionally, females among two younger generations were more likely to support the NP than the DPP. Because most leaders in the NP were former KMT legislators, it was of interest to see which factors affected people’s preferences for these two parties. As shown in Table 6-5, national identity was not a good variable to distinguish these two parties. The national identity of voters in the two older generations did not influence the formation of their preferences regarding these two parties. One possible explanation is that these voters saw the two parties as not different significantly in their stances on the national identity issue. However, as to third-generation voters, national identity influenced their preference for the KMT versus the NP, and people with Chinese identity tended to support the NP. It seems that younger voters perceived the KMT as adopting a more moderate stance on the national identity issue. However, most leaders of the NP are mainlanders, so voters still used their ethnic origin in discriminating between these two parties. Mainlanders in the three generations preferred the NP, and Taiwanese preferred the KMT. Therefore, as Tables 6-2 to 6-5 indicate, the emergence of the NP changed the outlook of the party system in Taiwan. Voters perceived the KMT as being similar to the DPP. Therefore, ethnic origin was no longer a good explanatory variable for DPP versus KMT partisan preference in 1996. However, people’s attitudes toward national identity, money politics, and democratic values still played an important role in explaining their preferences for the DPP versus the KMT, but the magnitude of this role declined in 1996. The difference between the DPP and the NP was determined by people’s national identity and ethnic origins. Most leaders of the DPP were Taiwanese, and they proposed Taiwan 131 Table 6-4. Determinants of Partisan Preference (DPP vs. NP), by Generation, 1996 1961 and After 1943 to 1960 1942 and Before Coefficient Coefficient Coefficient (Std. Error) (Std. Error) (Std. Error) National .350 .232 .218 identity (.037)*** (.039)*** (.063)*** Concern about .008 .004 .057 “money politics” (.036) (.037) (.062) Democratic values -.012 -.029 -.054 (.043) (.046)*** (.068) Mainlander .653 .898 1.060 (.106)*** (.135)*** (.204)*** Education -.036 .022 -.014 (.051) (.044) (.069) Socioeconomic .088 .177 . 127 status (.108) (.109) (.166) Government/KMT .208 -.221 .067 official/employee (. 137) (. 145) (.203) Female .213 .227 .174 (.080)** (.089)* (.167) Intercept 2.882 3.044 2.894 (.281)*** (.290)*** (.442)*** N 465 342 158 Adj. R2 .265 .228 .364 S.E.E. 0.8425 0.8022 0.9016 Conditional 20.584 19.156 17.939 index *: p S .05, **: p S .01, ***: p S .001, 2-tailed test. Entries are unstandardized coefficients (standard errors are in parentheses). 132 Table 6-5. Determinants of Partisan Preferences (KMT versus NP), by Generation, 1996 1961 and After 1943 to 1960 1942 and Before Coefficient Coefficient Coefficient (Std. Error) (Std. Error) (Std. Error) National . 165 .027 .065 identity (.037)*** (.042) (.073) Concern about .029 .075 .314 “money politics” (.035) (.039) (.069)*** Democratic values .217 .172 .133 (.042)** (.048)*** (.077) Mainlander .269 .613 .558 (.105)* (.141)*** (.236)** Education .076 .096 -.014 (.050) (.046)* (.078) Socioeconomic .047 .100 .235 status (.106) (.115) (.188) Government/KMT .079 -.324 -.309 official/employee (. 135) (.152)* (.230) Female .207 .069 .064 (.078)** (.094) (.190) Intercept 2.018 2.295 1.300 (.276)*** (.304)*** (.503)* N 466 343 158 Adj. R2 .141 .121 .169 S.E.E. 0.8251 0.8453 1.0172 Conditional 20.683 19.138 18.155 index *: p S .05, **: p S .01, ***: p S .001, 2-tailed test. Entries are unstandardized coefficients (standard errors are in parentheses). 133 independence since the 1991 national representative election,” so it was reasonable for the DPP to attract people with Taiwanese identity. On the other hand, most leaders of the NP were mainlanders, and they preferred unification with mainland China; so people with Chinese identity and mainlanders supported the NP. As Lin and his colleagues (1998: 24) pointed out, "Perceiving the DPP and the NP as antipodes on the Taiwanese- Chinese identity dimension with the KMT in between (perhaps closer to the NP than to the DPP), voters clearly judged the parties on the basis of their own national identity." However, there were some similarities between the DPP and the NP in 1996. On issues of concern about "money politics" and democratic values, voters perceived no difference between the two opposition parties. Therefore, there were incentives for the two parties to cooperate on these issues on the legislative agenda. As to voters’ preference for the KMT versus the NP, mainlanders were more likely to prefer the NP. Therefore, the emergence of the KMT undermined mainlanders’ support of the KMT. Conclusion In this chapter, the determinants of partisan preferences in Taiwan were examined. The openness of local and national elections provided a great opportunity for opposition forces to organize their resources and garner support from the electorate. Electoral campaigns served as a great socialization agent for the electorate in Taiwan to shape their partisan preferences. 2‘ In 1991, the DPP passed Taiwan independence clause in its Party Congress, and it pushed the national identity issue into the political arena. 134 When voters faced two major parties, i.e., the DPP and the KMT, their national identity, concern about "money politics," procedural democratic values, and ethnic origins were important attitudinal variables affecting their preference for one of the parties. However, the emergence of the NP changed the relative advantages of the KMT and the DPP. Because voters perceived the KMT as being similar to the DPP, ethnic origin was no longer a good explanatory variable for DPP versus KMT partisan preferences in 1996. But other attitudinal variables, such as national identity, concern about "money politics," and democratic values, still played an important role on explaining their DPP vs. KMT partisan preferences in 1996. The difference between the DPP and the NP was shaped by people’s national identity and ethnic origins. Therefore, people who identified themselves as Taiwanese and/or whose ethnic origins were Taiwanese were more likely to support the DPP than the NP. On the other hand, mainlanders and/or people with Chinese identity preferred the NP. However, there were some similarities between the DPP and the NP. On issues of concerns about "money politics" and democratic values, voters perceived no difference between the two opposition parties. Therefore, there are incentives for the two parties to cooperate on these issues on the legislative agenda in the future. As to preferences for the KMT versus the NP, people’s ethnic origins played a more important role in shaping these attitudes. Because people’s political attitudes might be volatile, political parties have incentives to provide better platforms to attract more supporters. The complex relationship between mainland China and Taiwan gives political parties another issue dimension with which gather support. In next chapter, another major issue regarding the 135 national identity of people in Taiwan is discussed. Also examined is whether people’s partisan preference has an important influence on their national identity. 136 CHAPTER 7 THE IDENTIFICATION PROBLEM: WHETHER PARTY IDENTIFICATION AFFECTS NATIONAL IDENTITY The importance of party identification lies not only in its effect on voting behaviors, but also in its effect on other political attitudes and behaviors. The other variable, peoples’ national identity, is also considered to be a long-term psychological variable in Taiwan. In this chapter, the distribution of national identity among voters is discussed, and whether party identification affects people’s national identity is examined. The Functions of Party Identification As Abramson (1983: 72) pointed out, Campbell and his colleagues advanced four major claims about the functions performed by party identification: 1. Party identification contributes to opinion formation. 2. Party identification influences voting behavior. 3. Party identification enhances psychological involvement in politics. 4. High levels of party identification among the electorate provide a check against new party movements and contribute to the established party system. Abramson and his colleagues (1998: 174-185) gave four examples to illustrate the association between party identification and other political attitudes. First, an identifier 137 tends to give more positive evaluations to his or her own party’s candidate. Second, there is a strong relationship between an individual’s party identification and his/her evaluation of the president’s performance. Third, a partisan identifier tends to have close stances with his or her party’s nominee. Finally, an identifier is likely to give positive evaluations to his or her own party on a summary measurement of retrospective evaluations of the past six general elections (1976 to 1996). In Taiwan, many researchers have examined how partisan preferences affect political attitudes and vote choices (L. Chen 1992, 1995, 1998; Y. Chen 1994; Chu 1992, 1996; Fu 1994; Hawang 1996, Ho 1994, Ho and Wu 1996; Liu 1987; Shyu 1991, 1995; Tsai 1997). In this chapter, I examine whether partisan preferences contribute to voters’ opinion formation. Because national identity is one of the most important political attitudes in Taiwan, I discuss whether partisan preference affects national identity. National Identity and Electoral Politics in Taiwan Under authoritarian control, leaders of the KMT, most of whom are mainlanders, indoctrinated in people that the Republic of China (ROC) is the only legitimate government of the whole China (including Taiwan and mainland China), and educated its people as Chinese in Taiwan. As mentioned Chapter 6, two major issues of the opposition movement are democratic reforms and Taiwan’s self-determination. There are at least two advantages to the opposition movement’s promoting Taiwanese identity. First, the Opposition can challenge the legitimacy of the KMT’s rule because the ROC lost its seat in the United Nations in 1971 and ended its formal diplomatic relationship with the 138 United States in 1978. In addition, Taiwanese (including Min-nan-jen and Hakkanese) comprise more than 80 percent of the total population of Taiwan, so the DPP might increase its vote shares in elections by promoting Taiwanese identity. Therefore, national identity is one of the core issues in the democratic process in Taiwan. In this section, the definition of national identity and the formation of national identity are discussed. Then, parties’ roles in the formation of national identity in Taiwan are examined. National Identity and Collective Memory According to Smith (1991), there are several fundamental features of national identity. These features include historic territory or homeland; common myths and historical memories; a common, mass public culture; common legal rights and duties for all members; and a common economy with territorial mobility for members. "A nation can therefore be defined as a named human population sharing an historic territory, common myths and historical memories, a mass, public culture, a common economy and common legal rights and duties for all members" (Smith 1991: 14, emphasis in original). Therefore, voters with Chinese identity have different beliefs concerning their homeland, historic memory, and mass culture from those with Taiwanese identity. Further, national identity involves some sense of political community. "A political community implies at least some common institutions and a single code of rights and duties for all the members of the community. It also suggests a definite social space, a fairly well demarcated and bounded territory, with which the members identify and to which they feel they belong" (Smith 1991: 9). Therefore, national identity provides psychological attachment for the members of its political community. 139 Whether the origin of national identity is primordial or situational is an interesting question. If national identity is primordial, political authorities cannot use education or other means to change people’s national identity. On the other hand, if national identity is situational, it means the state can indoctrinate citizens with that national identity through public education. One can examine the distribution of national identity of electorates to see if significant differences exist among different generations and ethnic groups. Then, we can gain a clear picture of whether the state can impose national identity on its people. Our interest is in whether the state can employ the mass education system to propagate national identity in its citizens. The Formation of National Identity and Political Socialization Gellner argued that nationalism creates nations where they do not exist, but it cannot arouse self-consciousness of nations (cited in Anderson 1986: 15). Smith agreed that state authorities can use public education to influence people’s national identity. National identities also fulfil more intimate, internal functions for individuals in communities. The most obvious is the socialization of the members as ’nationals’ and ’citizens.’ Today this is achieved through compulsory, standardized, public mass education systems, through which state authorities hope to inculcate national devotion and a distinctive, homogeneous culture, an activity that most regimes pursue with considerable energy under the influence of nationalist ideals of cultural authenticity and unity. (Smith 1991: 16) Therefore, we can examine whether the KMT can employ the public mass education system to educate Taiwanese as Chinese, and legitimize its political control. However, when the KMT faced its diplomatic setbacks during the 19708 and the progress of democratization during the 19808 and early 19908, it might have become a problem for the KMT to adhere to Chinese identity. As Horowitz (1994: 35) stated, 140 Democracy is about inclusion and exclusion, about access to power, about the privileges that go with inclusion and the penalties that accompany exclusion. In severely divided societies, ethnic identity provides clear lines to determine who will be included and who will be excluded. Since the lines appear unalterable, being in and being out may quickly come to look permanent. In ethnic politics, inclusion may affect the distribution of important material and nonmaterial goods, including the prestige of the various ethnic groups and the identity of the state as belonging more to one group than another. Again and again in divided societies, there is a tendency to conflate inclusion in the government with inclusion in the community and exclusion from government with exclusion from the community. Therefore, the DPP successfully used Taiwanese identity and democratic reforms to attract votes from the electorate, and national identity became a salient issue in the political arena of Taiwan. The emergence of the NP can be considered another challenge to the KMT among the KMT’s Chinese identifiers. Therefore, it is possible that there "is [a]... tendency in multi-ethnic societies for political parties to be organized along ethnic lines" (Brown 1997 [1993]: 87). As mentioned in Chapter 6, people's national identity, along with other attitudinal and demographic variables explains their partisan preferences. I was also interested in examining whether partisan preference can shape national identity. However, readers must be reminded that it is hard to examine whether there is a reciprocal causal relationship between these two variables because the limitation of available data sets. In this chapter, I examine the distribution of national identity among the electorate, and then I employ OLS to examine whether partisan preference influences national identity. At the end of this chapter, I discuss whether OLS estimation is an appropriate choice in my estimation of this model. 141 Distribution of National Identity Among the Electorate In this section, the distribution of national identity among the electorate in Taiwan is examined. Differences among generations in their national identities also are explored. Because first-generation voters (born in 1942 and before) had life experiences in either Taiwan or mainland China before the Nationalist government retreated from China in 1949, this might have affected their national identity. Second-generation voters (born between 1943 and 1960) received their education under the KMT’s authoritarian control, so this political socialization process might have influenced their national identity. As to third-generation voters (born after 1960), they experienced political openness and electoral competition, so this experience might also have affected their national identity. The effect of ethnic difference on national identity also is examined. Because mainlanders or their parents came from China after 1949, it was predicted that they would be more likely to identify themselves as Chinese. As to Taiwanese, first-generation voters were more likely to identify themselves as Taiwanese because they had Taiwanese life experiences before the KMT came to Taiwan. However, as to Taiwanese in the two younger generations, they might be more likely to identify themselves as Chinese because they received their education under the KMT’s control. National Identity Among the Electorate in Taiwan As shown at the bottom of Table 7-1, in 1993, 35.4 percent of voters identified themselves as Chinese, 35.8 percent as "both Chinese and Taiwanese," and 28.8 percent as Taiwanese. This indicates that more people identified themselves as Chinese than as 142 Taiwanese that year. Generational differences in national identity also were examined. More than four out of ten third-generation voters identified themselves as Chinese. Less than one out of four voters in that generation identified themselves as Taiwanese, and 36.1 percent of them identified themselves as "both Chinese and Taiwanese." Therefore, voters in the third generation were less. likely to identify themselves as Taiwanese but were more likely to have Chinese identity in 1993. A8 to second-generation voters, the distribution of national identity was quite even: In 1993, 33.7 percent of them identified themselves as Chinese, 34.6 percent as "both Chinese and Taiwanese," and 31.7 percent as Taiwanese. That same year, the proportion of first-generation respondents with Taiwanese identity was higher than that in the other two generations: 33.4 percent of them identified themselves as Taiwanese, only 29.4 percent as Chinese, and 37.1 percent as "both Chinese and Taiwanese." However, the distribution of national identity among the electorate changed in 1996. As shown at the bottom of Table 7-1, only one out of five voters identified themselves as Chinese, but more than one out of three voters identified themselves as Taiwanese that year. More people identified themselves as "both Chinese and Taiwanese" or Taiwanese in 1996 than in 1993. This change was evident primarily in the two younger generations. As to third-generation voters, almost one out two voters (46.2 percent) identified themselves as "both Chinese and Taiwanese," 31.2 percent of them as Taiwanese, and only 22.6 percent as Chinese. The distribution of Chinese identity decreased by almost 20 percent among third-generation voters between 1993 and 1996. Like third-generation 143 Table 7-1. National Identity1 Among the Electorate, by Generation, 1993 and Generatioln996 National ID 1993 1996 Difference 1961 and After Taiwanese 21.7 3 1.2 9.2 Both 36.1 46.2 10.1 Chinese 42.3 22.6 -19.3 (N) (452) (454) 1943 to 1960 Taiwanese 31.7 42.2 10.5 Both 34.6 42.4 7.8 Chinese 33.7 15.4 -18.2 (N) (489) (488) 1942 and Before Taiwanese 33 .4 38.2 4.8 Both 37.1 32.7 -4.4 Chinese 29.4 29. 1 -0.3 (N) (377) (306) Total Sample Taiwanese 28.8 36.8 8.0 Both 35.8 41.7 5.9 Chinese 35.4 21.5 -13.9 (N) (1,318) (1,342) 1 Entries are column percentages. For question wordings, see Chapter 6. * N<25. 144 voters, voters in second generation became more likely to identify themselves as Taiwanese and less likely to identify themselves as Chinese in 1996. Specifically, 42.2 percent of second-generation voters identified themselves as Taiwanese, 42.4 percent as "both Chinese and Taiwanese," and only 15.4 percent as Chinese in 1996. Between 1993 and 1996, the proportion of Taiwanese identity among second-generation voters increased by 10.5 percent, and the proportion of "both Chinese and Taiwanese" identity increased by 7.8 percent. Conversely, the proportion of Chinese identity among second- generation voters decreased by 18.2 percent between 1993 and 1996. Compared with the two younger generations, the distribution of national identity among first-generation voters was relatively stable between 1993 and 1996. About three out of ten first-generation voters had Chinese identity in 1993 and 1996. The proportion of those voters with Taiwanese identity increased from 33.4 to 38.2 percent during that period, but the proportion of "both Chinese and Taiwanese" identity decreased from 37.1 percent to 32.7 percent. The question wordings were identical in both surveys, and I found a similar trend by examining data collected by the Election Study Center at National Chengchi University after the legislative elections in 1992 and 1995.22 So, the general trend is that voters became more likely to identify themselves as Taiwanese or "both Taiwanese or Chinese" in 1996. However, a difference might also exist between the two ethnic groups (Taiwanese and mainlanders) in term of their national identity. The distribution of 22 From Figure 2.1 of Tsai’s (1997: 31) thesis, I calculated the distributions of national identity as follows: "Chinese" was 30.8 percent, "Both Chinese and Taiwanese" was 48.1 percent, and "Taiwanese" was 21.1 percent in 1992. In 1995, "Chinese" wa818.7 percent, "Both Chinese and Taiwanese" was 49.7 percent, and "Taiwanese" was 31.6 percent. 145 national identity among Taiwanese was examined first. The results are reported in the following section. National Identity Among Taiwanese Figures at the bottom of Table 7-2 indicate that, in 1993,more than one out of four Taiwanese identified themselves as Chinese, 37.7 percent had "both Chinese and Taiwanese" identity, and 33.7 percent had Taiwanese identity. Comparing third- generation Taiwanese to those in the two older generations, it was found that third- generation Taiwanese were more likely to have Chinese identity in 1993 and 1996. In 1993, 35.3 percent of them had Chinese identity, 38.8 percent had "both Chinese and Taiwanese" identity, and only 25.9 percent had Taiwanese identity. Second-generation Taiwanese were more likely to have Taiwanese identity than the other two national identities; 34.8 percent of them had Taiwanese identity, 34.1 percent had "both Chinese and Taiwanese" identity, and 31.1 percent had Chinese identity in 1993. A8 to first- generation Taiwanese, they were more likely to identify themselves as Taiwanese in 1993. More than four out of ten first-generation Taiwanese had Taiwanese identity, 41.7 percent had "both Chinese and Taiwanese" identity, and only 15.9 percent had Chinese identity. The proportion of Taiwanese identity among third-generation Taiwanese increased from 25.9 percent to 35.4 percent between 1993 and 1996 (see Table 7-2). More than one-third of this group had "both Chinese and Taiwanese" identity in 1993, and the proportion increased to 45.0 percent in 1996. However, the proportion with Chinese identity decreased from 35.3 percent to 19.6 percent between 1993 and 1996-- a 146 decline of 15.7 percent. Table 7-2 also indicates that 43.8 percent of second-generation Taiwanese had Taiwanese identity in 1996. The proportion of Chinese identity among second-generation Taiwanese declined by 16 percent (from 31.1 percent to 15.2 percent) between 1993 and 1996. Therefore, Taiwanese in the two younger generations became less likely to have Chinese identity in 1996 than in 1993. However, the distribution of Chinese identity was relative stable among first—generation Taiwanese. As shown in Table 7-2, about 16 percent of them had Chinese identity in both 1993 and 1996. In addition, compared to the two younger generations, first-generation Taiwanese were more likely to have Taiwanese identity. Their proportion of Taiwanese identity increased from 42.4 percent to almost 50 percent between 1993 and 1996. From Table 7-2 it can be seen that Taiwanese identity among Taiwanese is mutable. The general trend is that young Taiwanese are more likely than older ones to abandon their Chinese identity. Some of them switched their national identity to a more ambiguous stance, i.e., "both Chinese and Taiwanese" identity, but more of them changed to Taiwanese identity. As to first-generation Taiwanese, they started with the lowest level of Chinese identity among the three generations, but maintained a relatively stable but low level of Chinese identity. This finding indicates that the life experiences of first- generation Taiwanese had a long-term effect on their national identity. However, political forces also affected young Taiwanese voters’ national identity. It is interesting to discuss why the distribution of national identity among Taiwanese changed between 1993 and 1996. I will examine the distribution of national identity among mainlanders first, and then return to discuss this point later. 147 Table 7-2. National Identity1 Among Taiwanese, by Generation, 1993 and 1996 Generation National ID 1993 1996 Difference 1961 and After Taiwanese 25.9 35.4 9.5 Both 38.8 45.0 6.2 Chinese 35.3 19.6 -15.7 (N) (371) (449) 1943 to 1960 Taiwanese 34.8 43.8 9.0 Both 34.1 41.0 6.9 Chinese 31.1 15.2 -15.9 (N) (431) (427) 1942 and Before Taiwanese 42.4 49.6 7.2 Both 41.7 33.9 -7.8 Chinese 15.9 16.5 0.6 (N) (283) (230) All Taiwanese Taiwanese 33.7 41 .6 7.9 Both 37.7 41.1 3.4 Chinese 28.6 17.3 -11.3 (N) (1,085) (1,106) ‘ Entries are column percentages. For question wordings, see Chapter 6. *N<25. 148 National Identity Among Mainlanders As shown at the bottom of Table 7-3, in 1993, almost three out of four mainlanders had Chinese identity, less than 2 percent had Taiwanese identity, and 23.7 percent had "both Chinese and Taiwanese" identity. Generational differences in their national identities were examined. However, we must bear in mind that some cells in Table 7-3 have fewer than 25 cases, so we need to be careful when making inferences from these few cases. As shown at the top of Table 7-3, in 1993, almost eight out of ten third-generation mainlanders had Chinese identity, 21.6 percent had "both Chinese and Taiwanese" identity, and only 1.4 percent had Taiwanese identity. As to second-generation mainlanders, 60 percent of them had Chinese identity, 35.4 percent had "both Chinese and Taiwanese" identity, and less than 5 percent had Taiwanese identity. Among mainlanders of all three generations, voters in the second generation were more likely to have "both Chinese and Taiwanese identity" than those in other two generations in both 1993 and 1996. Mainlanders in the first generation, who followed Chiang Kai-shek to Taiwan, had the highest proportion of Chinese identity. More than 80 percent of them had Chinese identity, 18.1 percent had "both Chinese and Taiwanese" identity, but none of them had Taiwanese identity in 1993. However, great changes took place in 1996. The proportion of mainlanders with Chinese identity declined by 31 percent between 1993 and 1996 (see Table 7-3). In addition, the proportion of mainlanders with "both Chinese and Taiwanese" identity increased by 22.4 percent, and the proportion with Taiwanese identity increased by 8.6 percent in that period. Mainlanders became less likely to identify themselves as Chinese, but a few of them still were still less likely to identified themselves as Taiwanese in 1996. 149 Table 7-3. National Identity1 Among Mainlanders, by Generation, 1993 and 1996 Generation National ID 1993 1996 Difference 1961 and After Taiwanese 1.4* 7.7* 6.3 Both 21.6* 53.8 32.2 Chinese 77.0 38.5 -38.5 (N) (74) (91) 1943 to 1960 Taiwanese 4.2* 25.0* 20.8 Both 35.4* 55.8 20.4 Chinese 60.4 19.2* -41.2 (N) (48) (52) 1942 and Before Taiwanese 0.0* 2.7* 2.7 Both 18.1* 29.7 11.6 Chinese 81.9 67.6 -14.3 (N) (72) (74) All Mainlanders Taiwanese l .5 * 10.1* 8.6 Both 23.7 46.1 22.4 Chinese 74.7 43.8 -30.9 (N) (194) (217) ‘ Entries are column percentages. For question wordings, see Chapter 6. *N<25. 150 Table 7-3 also indicates that, in 1996, more than one out of two third-generation mainlanders had "both Chinese and Taiwanese" identity, 38.5 percent had Chinese identity, and just 7.7 percent had Taiwanese identity. The proportion of third-generation mainlanders with Taiwanese identity increased by 6.3 percent between 1993 and 1996, but they were inclined to take a more ambiguous stance on national identity, i.e., "both Chinese and Taiwanese," in 1996. The proportion of third-generation mainlanders describing their national identity as "both Chinese and Taiwanese" increased by 32.2 percent between 1993 and 1996. However, the pr0portion claiming Chinese identity decreased dramatically, by 38.5 percent, between 1993 and 1996. Second-generation mainlanders had the lowest level of Chinese identity among the three generations of mainlanders in both 1993 and 1996. Especially, noteworthy is that the proportion with Chinese identity decreased dramatically in that period, from 60.4 percent to 19.2 percent. In addition, one out of four second-generation mainlanders had Taiwanese identity in 1996, although a majority of them (55.8%) still identified themselves as "both Chinese and Taiwanese" in 1996. When each generation was examined separately, first-generation mainlanders were found to have the most stable national identity among the three generations of mainlanders. The proportion of first-generation mainlanders with Chinese identity was still higher than that of mainlanders in the two younger generations. In 1996, 67.6 percent of them had Chinese identity, about 30 percent identified themselves as "both Chinese and Taiwanese," and less than 3 percent had Taiwanese identity. Between 1993 and 1996, the proportion of first-generation mainlanders with Chinese identity decreased by 14.3 percent, but the proportion with Taiwanese identity increased by only 2.7 percent. 151 From Tables 7-2 and 7-3, it can be seen that national identity has both subjective and objective components among voters in Taiwan. First- generation Taiwanese were once ruled by the Japanese, and they had life experience outside the KMT’s rule. This experience might make it hard for them to have Chinese identity. On the other hand, first- generation mainlanders had their own Chinese life experience when they were young in China, so this also makes it hard for them to have Taiwanese identity. However, the proportion of Chinese identity among first-generation Taiwanese was higher than the proportion of Taiwanese identity among mainlanders, illustrating that political education imposed by the KMT had an effect. Voters in the younger generations might have subjectively chosen their national identity. From the changes between 1993 and 1996, it was seen that more young Taiwanese switched their Chinese identity to Taiwanese identity or to "both Chinese and Taiwanese" identity. In addition, younger mainlanders also changed their national identity. However, one thing to keep in mind is that ethnic origin still plays an important role in people’s national identity. Mainlanders still were more likely to have Chinese identity than Taiwanese, and they were more likely to switch their Chinese identity into a more ambiguous stance, i.e., "both Chinese and Taiwanese." Therefore, national identity seems to be amendable among voters in the two younger generations. As Zaller (1992) pointed out, opinion is the marriage of predisposition and information. People’s ethnic origins indicate the differences in socialization processes of people’s families. Democratization gives the public a chance to experience open discussion on the national identity issue. Therefore, the national identity issue evolves from one-sided information, i. e., propaganda of Chinese identity under the KMT’s 152 authoritarian control, to an open-discussion environment. In addition, as shown in previous chapters, two new political parties used different national identities to attract voters. The DPP has been labeled the "Taiwanese" party, and the NP the "mainlander’s" party. However, the ruling party, the KMT, takes a more ambiguous position on this issue. Because political elites heatedly disagree on the national identity issue, this has also affected people’s positions on the national identity issue.23 Another factor that might affect the national identity of people in Taiwan must be considered. Because national identity includes a psychological component, the interaction between Taiwan and mainland China must be examined. In 1995, mainland China held two missile tests in an area about 80 to 85 miles north of Taiwan, and China also staged a military exercise near Taiwan in October 1995 (Achen, Hsu, and Kuo 1997). These actions irritated people in Taiwan. Therefore, the proportion of Chinese identity decreased dramatically between 1993 and 1996; young Taiwanese were more likely to identify themselves as Taiwanese, but young mainlanders were more likely to take a more ambiguous stance on their national identity, i.e., "both Chinese and Taiwanese." Therefore, mainlanders would rather switch their Chinese identity to an ambiguous stance on national identity, i.e., "both Chinese and Taiwanese." This might indicate that the political climate in Taiwan is promoting Taiwanese identity, and mainlanders, as a minority ethnic and political group in Taiwan, might just make a "politically correct" choice by giving up their Chinese identity. However, they would rather take a middle-of- the-road stance by having "both Chinese and Taiwanese" identity than having Taiwanese identity. As to Taiwanese, they have become the majority group ethnically and ’3 For a discussion of the polarization effect, see Zaller (1992: Chapter 6 and Chapter 9). 153 politically, and Lee Teng-hui, the first Taiwanese president, also promoted Taiwanese identity, so they gave up their Chinese identity and chose Taiwanese identity instead. In this section, the bivariate relationships between national identity and two social forces--political generation and ethnic origin--were examined. In the next section, a more comprehensive model is presented to explain the determinants of national identity. For the purpose of this research, I examined whether people use their partisan preferences a8 cues to take a position on their national identity. How Partisan Preference Affects National Identity In this section, the functions of party identification are discussed. One important function of party identification is to formulate other political attitudes. In this study I examined whether party identification can affect national identity. I included respondents’ partisan preference, ethnic origin, educational level, and gender as explanatory variables in the model. The dependent variable was national identity. I hypothesized that people with a positive attitude toward the KMT or/and the NP are more likely to have Chinese identity. On the other hand, people with a positive attitude toward the DPP are more likely to have Taiwanese identity. As to ethnic origin, mainlanders are more likely to have Chinese identity. People’s educational level is an interesting variable here. I hypothesized that people with a higher educational level are more likely to have Chinese identity. In addition, because second-generation voters received their formal education under the KMT’s authoritarian control, it was of interest to see whether this situation significantly affected their national identity. I also controlled for respondents’ gender. 154 I employed the same recoding schema as described in the previous chapter. However, I included three like/dislike questions regarding the three major parties in 1996 as measurements of partisan preference. Respondents answered questions using a 5-point scale, ranging from dislike it very much (1) to like it very much (5); the higher the value, the more one liked a given party. Table 7-4 indicates whether people’s partisan preferences, ethnic origin, and educational level affected their national identity in 1993 and 1996. In 1993, people with a KMT preference, those with higher educational levels, and mainlanders were more likely to have Chinese identity. In 1996, people who liked the KMT and/or the NP were more likely to have Chinese identity. Those who liked the DPP were more likely to have Taiwanese identity. In addition, mainlanders and people with higher educational levels were more likely to have Chinese identity. Compared with the coefficients for 1993, the effect of ethnic origin and educational level decreased in 1996. As to partisan preferences, it was not surprising to find that people’s attitude toward the NP had a greater effect on their national identity than did their attitude toward the KMT. Each generation was also examined separately in 1993 and 1996. The results in 1993 indicated that there was no severe multicollinearity any of the equations (see Table 7-5). One thing we need to keep in mind is that second-generation voters received their education after the KMT retreated from mainland China, and it is interesting to see how the state as an agent of political socialization affected second-generation voters’ national identity. As shown in Table 7-5, people’s preference for the DPP versus the KMT and their ethnic origins had a significant effect on their national in 1993. Further, second- generation voters with higher educational levels were more likely to have Chinese 155 Table 7-4. Determinants of National Identity, 1993 and 1996 1993 1996 Coefficient Coefficient (Std. Error) (Std. Error) Partisan Preference (DPP vs. KMT) .216 (.024)*** Like/Dislike KMT 175 (.039)*** Like/Dislike DPP -.292 (.039)*** Like/Dislike NP .310 (.038)*** Mainlander .985 .444 (.092)*** (.090)*** Education .149 .081 (.025)*** (.026)** Female -.098 -.085 (.064) (.063) Intercept 1 .764 2.008 (.120)*** (.232)*** N 1206 1 1 15 Adj. R2 .236 .196 S.E.E. 1.0936 1.0328 Conditional 9.324 19.634 Index ' *: p S .05, **: p S .01, ***: p S .001; 2-tailed test. Entries are unstandardized coefficients (standard errors are in parentheses). 156 identity. Therefore, people’s partisan preferences help them formulate their national identity. As to ethnic origin, its effect on first- and third-generation voters was greater than on second-generation voters. Because mainlanders in the first generation might be parents of mainlanders in the third generation, it is possible that the family is an important agent in transmitting national identity to its offspring.24 Table 7-5 also indicates that the ruling party can impose national identity on its citizens through public education, so the educational level of second-generation voters significantly affected their national identity. As to third-generation voters, educational level played no role in forming their national identity because they live in a more open society and can get more two-sided information regarding the debate on the national identity issue. For first- generation voters, educational level played no role in shaping their national identity in 1993. In 1996, another major party was added to the equation. I used three like/dislike questions to explore voters’ attitudes toward three major parties. At the bottom of Table 7-6, it can be seen that there was no sever multicollinearity for any of the equations. Because the DPP (pro-Taiwanese) and the NP (pro-Chinese) take opposite sides on the national identity issue, people’s attitudes toward these two parties significantly affected their national identity. The NP is a new party in the political system in Taiwan, so its effect on third-generation voters was larger than it was on voters in the other two older generations. The same situation applied to the DPP. Its effect on the youngest generation 2“ Some mainlanders in the third generation might also be offspring of second- generation mainlanders. However, in those samples in 1996, most mainlanders in the first generation came to Taiwan when they were adolescent. Therefore, first-generation mainlanders’ average marriage age is older than that of first-generation Taiwanese, so some third-generation mainlanders might be their offspring. 157 Table 7-5. National Identity, by Generation, 1993 1961 and After 1943 to 1960 1942 and before Coefficient Coefficient Coefficient (Std. Error) (Std. Error) (Std. Error) Partisan Preference .225 .236 .157 (DPP Versus KMT) (.041)*** (.040)*** (.042)*** Mainlander .878 .671 1.524 (.144)*** (.187)** (.161)*** Education .043 .163 .085 (.065) (.049)** (.050) Female -.240 .063 -.142 (.104)* (.111) (.119) Intercept 2.197 1.608 1.935 (.265) (.201)** (.203)*** N 419 461 326 Adj. R2 .185 .160 .390 S.E.E. 1.0635 1.1529 1.0178 Conditional 12.140 8.850 9.1 10 Index *: p S .05, **: p S .01, ***: p S .001; 2-tailed test. Entries are unstandardized coefficients (standard errors are in parentheses). 158 was larger than on the two older generations. Further, people’s attitude toward the KMT had a significant effect only on second-generation voters. That is, second-generation voters employed their attitudes toward the three major parties to determine their national identity. Do voters in Taiwan use their partisan preference to determine their stances on the national identity issue? I will return to this point later in the chapter. Table 7-6 also indicates that ethnic origin did not have a significant effect on second-generation voters’ national identity, but educational level did have a significant effect. This finding suggests that political control can impose national identity on citizens. However, mainlanders in the third generation were still influenced by their parents, i.e., first-generation mainlanders, in forming their national identity. Therefore, ethnic origin had a significant effect on voters in the first and third generations. From Tables 7-4 to 7-6, it can be seen that partisan preferences played a significant role in shaping national identity of people in Taiwan between 1993 and 1996. However, when each generation was examined separately in 1996, it was found that only second— generation voters employed their attitudes toward the three major parties to determine their national identity. The KMT has been the ruling party in Taiwan since 1949, but first- and third-generation voters’ attitudes toward the KMT did not affect their national identity in 1996. On the contrary, they employed their attitudes toward two relatively new parties, the DPP and the NP, to determine their national identity. The two new parties take clear but opposite positions on the national identity issue; thus it is possible that first- and third-generation voters used their attitudes toward the two opposition parties to determine their national identity. In addition, the emergence of the NP not only changed the party system in Taiwan, but it might also have changed people’s 159 Table 7-6. National Identity, by Generation, 1996 1961 and After 1943 to 1960 1942 and Before Coefficient Coefficient Coefficient (Std. Error) (Std. Error) (Std. Error) Partisan Preference Like/Dislike KMT .075 .294 .126 (.058) (.070)*** (.083) Like/Dislike DPP -.316 -.286 -.206 (.054)*** (.068)*** (.093)* Like/Dislike NP .350 .261 .266 (.051)*** (.064)*** (.098)** Mainlander .304 -.096 1.052 (.125)* (.179) (.212)*** Education .090 . 142 .048 (.053) (.047)** (.065) Female -.059 .003 -.205 (.088) (.107) (.160) Intercept 2.251 1.522 2.172 (.346)*** (.380)*** (.576)*** N 496 406 213 Adj. R2 .200 .123 .327 S.E.E. 0.9640 1.0467 1.0836 Conditional 21.01 1 19.162 20.193 Index *: p S .05, **: p S .01, ***: p S .001; 2-tailed test. Entries are unstandardized coefficients (standard errors are in parentheses). 160 perceptions of political parties. Because the NP supported unification with mainland China and the KMT began to take a more ambiguous stance on this issue, people might have perceived that the KMT was less likely to support unification with mainland China. Therefore, they employed their preferences for two relatively new parties, 1. e., the DPP and the NP, to determine their national identity. Because of the limitation of available data, I can present another interpretation of the function of party identification. It is also possible that first- and third-generation voters did not have a long-term psychological attachment to any political party. As demonstrated in Chapter 4, third-generation voters’ partisan preferences were relatively unstable compared with those of voters on the two older generations. Therefore, whether they employed their unstable partisan preferences to determine their national identity needed to be reconsidered. First-generation voters had the most stable partisan preference of the three generations between 1986 and 1996, but they did not employ their partisan preference for the KMT, i.e., the ruling party since 1949, to decide their national identity. Therefore, first-generation voters might have cast their votes for the KMT in the elections, but they did not formulate their party identification or employ it as a cue to understand the complex political world. In addition, it is unreasonable to declare that first-generation voters used their partisan preferences for the NP and the DPP, two new parties for them, to channel their political attitudes. Due to the limitation of available data, one reasonable conclusion is that second-generation voters might use their partisan preferences to formulate their national identity. In this section, I also consider how the state can employ the public education system to instill national identity in its citizens. Second-generation voters went to 161 elementary school after 1949, so they received their formal education under the KMT’s authoritarian control. Therefore, their educational levels significantly affect their national identity, even though their ethnic origins played no role in their national identity in 1996. As to first-generation voters, their ethnic origins played an important role in their national identity because they had either Chinese or Taiwanese life experience during their early adulthood. In addition, there might be a reciprocal relationship between partisan preference and national identity. A better way to clarify this point would be to employ a simultaneous model to examine the relationship between these two variables. However, because I could not find good instrumental variables to use for endogenous independent variables from the available data, it was not possible to do so. The potential problem was discussed in the next section. Simultaneity, Consistent Estimation, and Efficiency It would be better to use simultaneous equations to clarify whether there is a reciprocal relationship between partisan preference and national identity. However, it was not possible to do so from the available data in this study. In this chapter, I use OLS to estimate how people’s partisan preferences affected their national identity. However, the estimations might be inconsistent when the explanatory variables in the model are not real exogenous. This means that when the explanatory variables are correlated with the stochastic disturbance, one will get inconsistent estimations. I employed partisan preference as an explanatory variable to explain people’s national identity. It is possible 162 that people’s national identity also affected their partisan preference. Therefore, the coefficients of partisan preferences would be inconsistent if there was a reciprocal causal relationship between national identity and partisan preferences. I employed OLS because the data sets available were cross-sectional, and I could not find proper instrumental variables to estimate two-stage least squares. However, even if I could have found some instrumental variables, I would have had to make sure the model was well-specified. The best instrumental variable is correlated with endogenous variables but uncorrelated with the disturbance term (Johnston and DiNardo 1997:155). A8 Bartels (1991: 777) wrote, [It] is obvious that a "quasi-instrumental variable" estimator—-one based on an instrumental variable that is only approximately uncorrelated with the disturbance-- will not produce consistent estimates of the underlying parameters of interest. ...Moreover, even when a genuine (perfectly exogenous) instrument is available, it may be so inefficient that a quasi-instrumental variable estimator is preferable in practice (e.g., by a squared error criterion) because data are in short supply. In addition, as Bound and his colleagues (1995) suggested, when the correlation between the instruments and the endogenous explanatory variable is weak, it can lead to large inconsistencies in the instrumental variable estimates, even if there is only a weak relationship between the instruments and the error in the structural equation. They also argued that, in a finite sample, instrumental variable estimates are biased in the same direction as OLS estimates. Therefore, Johnston and DiNardo (1997:316-317) suggested, In practice OLS is still widely used in the estimation of structural equations in spite of its knowledged inconsistency. A possible rationalization lies in the contrast between small-sample and large-sample properties. Consistency is a large-sample, or asymptotic, property. Consistent estimators are not necessarily unbiased in finite sample: in fact, they usually display finite sample bias. Moreover, the sampling exceed that of OLS estimators. Thus, in finite samples OLS may show a smaller mean squared error than consistent estimators.25 25 See also Kennedy (1993: 136-137). 163 In addition, as shown in Appendix D, the Hausman test for model specification indicated that there was no simultaneity problem in my models. It is always a temptation for social scientists to contemplate the causal relationship between two variables. However, because the limitation of available variables in data, it is not easy to construct latent variables for national identity and party identification. In addition, one must make sure that models are identified, so structural equation models can be estimated. Therefore, structural equation models were not applied in this research, but I employed OLS in the analysis discussed this chapter. Conclusion In this chapter, I examined the distribution of national identity among the electorate in Taiwan in 1993 and 1996. In addition, I tested whether partisan preference can affect national identity. The distribution of national identity among first-generation voters was very stable between 1993 and 1996. Thirty percent of first-generation voters had Chinese identity in 1993 and 1996, and about one-third of them had Taiwanese identity. However, second- and third-generation voters became less likely to identity themselves as Chinese in 1996. The proportion of voters with Chinese identity declined by about 20 percent between 1993 and 1996. Only 15.4 percent of second-generation voters and about 22.6 percent of third-generation voters had Chinese identity in 1996. On the other hand, the proportion of voters in the two younger generations with Taiwanese identity increased by 1.0 percent 164 between 1993 and 1996: 31.2 percent of third-generation voters and 42.2 percent of second-generation voters had Taiwanese identity in 1996. Taiwanese voters became more likely to have Taiwanese identity in 1996; the proportion of them with Taiwanese identity increased from 33.7 percent to 41.6 percent between 1993 and 1996 (Table 7-2). On the other hand, the proportion of Taiwanese voters with Chinese identity decreased from 28.6 percent to 17.3 percent between 1993 and 1996. The distribution of national identity among first-generation Taiwanese was the most stable one among these three Taiwanese generations: about 16 percent of first- generation Taiwanese had Chinese identity in both 1993 and 1996, and the proportion with Taiwanese identity increased from 42.4 percent to 49.6 percent between 1993 and 1996. However, Taiwanese in the two younger generations became more likely to have Taiwanese identity or "both Chinese and Taiwanese" identity between 1993 and 1996. As to mainlanders, they became less likely to have Chinese identity between 1993 and 1996 (Table 7-3). The distribution of national identity among first-generation mainlanders was also the most stable one among the three generations of mainlanders between 1993 and 1996. The proportion of first-generation mainlanders with Taiwanese identity was less then 3 percent in 1993 and 1996. However, the proportion of first- generation mainlanders with Chinese identity decreased by 14.3 percent in that same period. The proportion of mainlanders in the two younger generations with Chinese identity decreased by about 40 percent between 1993 and 1996. Third-generation mainlanders became more likely to have "both Chinese and Taiwanese" identity, but the proportion of them with Taiwanese identity increased by only 6.3 percent between 1993 165 and 1996. However, the proportion of second-generation mainlanders with Taiwanese identity increased by 20.8 percent in that period. Zaller (1992) argued that opinion is the marriage of predisposition and information. The origin of national identity in Taiwan includes individuals’ objective ethnic origins, debates among political elites, effects of mass media, and some critical events. During 1995 and 1996, mainland China initiated several missile tests near Taiwan and staged military exercises. One consequence of these actions was that people in Taiwan become less likely to have a positive attitude toward China, and the proportion of them with Chinese identity decreased dramatically between 1993 and 1996. In this chapter, I also described the function of party identification in shaping national identity. However, when each generation was examined separately in 1996, it was found that only second-generation voters employed their attitudes toward the three major parties in determining their national identity. Voters in the first and the third generations did not use their attitudes toward the KMT to decide their national identity in 1996, but they employed their attitudes toward two relatively new parties, the DPP and the NP, to determine their national identity. Although we know that the two new parties, the DPP (pro-Taiwanese) and the NP (pro-Chinese), take clear but opposite positions on the national identity issue, it is possible that first- and third-generation voters did not use partisan preferences to formulate their national identity. As to third-generation voters, the distribution of their partisan preferences was relatively unstable between 1986 and 1996. Therefore, it is unreasonable to assume that they employed their partisan preferences in determining their national identity. First-generation voters did not employ their attitude toward the KMT, the ruling party in Taiwan for more than 40 years, to channel their 166 national identity, but relied on two new parties. These findings make me suspicious about the measurement of partisan preferences capturing only first-generation voters’ short-term attitudes toward political parties. First-generation voters did not experience open electoral competition between the major parties during their early adulthood. It is possible that they just cast their votes habitually to support the ruling party, but did not have a long- term psychological attachment toward a particular political party. Therefore, they did not use party identification to formulate their political attitudes. Second-generation voters in Taiwan might have a long-term psychological attachment to certain major parties, and they might be the only generation in Taiwan to use party identification to determine their political attitudes. 167 CHAPTER 8 CONCLUSION The research question posed was: 18 party identification a useful concept for us to understand the political participation of people in Taiwan? I focused on the stability of partisan preferences among voters, the social bases of partisan support, and the relationship between partisan preferences and other political attitudes. Due to the limitation of available data in Taiwan, I employed four cross—sectional data sets to answer the research question. In this chapter, the research findings are summarized, and suggestions are made for future research. Research Findings and Discussion In this section, four major findings of this research are discussed. First, the stability of partisan preferences in the electorate is discussed. Second, the social bases of partisan support are considered. Third, the determinants of people’s partisan preference are presented. Finally, whether voters employ their partisan preferences to determine their national identity is discussed. 168 The Stability of Partisan Preferences in the Electorate In this dissertation, political generation was employed as a mediator variable because voters in different generations might have had different political experiences during their early adulthood. As demonstrated in this dissertation, these experiences significantly affect their partisan preferences. Of the three generations of voters in Taiwan, the first generation had the most stable partisan preferences between 1987 and 1996. Their positive attitudes toward the KMT varied from 34.9 percent to 46.4 percent between 1987 and 1996 in that ten-year period, and the range was 11.5 percent. The distribution of DPP partisans among first- generation voters varied from 5.5 percent to 10.8 percent, and the range was 5.3 percent. As to Taiwanese, partisan preference for the KMT among first-generation Taiwanese was the most stable. The proportions varied from 27.5 percent to 39.3 percent, and the range was 11.8 percent. However, second-generation Taiwanese had the most stable partisan preference for the DPP. The distribution varied from 10.2 percent to 16.8 percent, and the range was 6.6 percent. As to mainlanders, second-generation mainlanders had the most stable partisan preference for the KMT. The proportions varied from 48 percent to 73.3 percent, and the range was 25.3 percent. As to mainlanders’ partisan preference for the DPP, first-generation mainlanders had the most stable distribution. The proportions varied from 0 percent to 4.2 percent, and the range was 4.2 percent. Therefore, the two older generations in Taiwan tended to have stable partisan preferences between 1987 and 1996. 169 The Social Bases of Partisan Support In this study, I examined how people’s ethnicity, gender, education, vocation, and socioeconomic status might affect their partisan preferences. In addition, I investigated how the emergence of the NP affected the social bases of support for two major parties. People’s ethnicity was a good indicator to predict their partisan preferences. When voters faced two major parties, i.e. the KMT and the DPP, first-generation mainlanders were the group most supportive of the KMT between 1987 and 1996. First-generation Taiwanese were least likely to support the KMT in 1987, but they increased their support over time and became the group most supportive of the KMT among the three Taiwanese generations in 1996. As to the DPP, Taiwanese were more likely to support the DPP than mainlanders. Among Taiwanese, third-generation voters were more likely to support the DPP than were the other two generations in 1996. On the other hand, less than 5 percent of mainlander voters preferred the DPP between 1987 and 1996. When voters had to make a choice among the three major parties, those in the third generation were most likely to defect from the KMT. There were ethnic differences in their partisan preferences when they faced the three major parties. Compared with younger Taiwanese, older-generation ones were less likely to be affected by the emergence of the NP. On the other hand, a majority of mainlanders in the three generations defected from the KMT when they had another alternative, i.e., the NP. But young mainlanders were more likely to be attracted by the NP than were older mainlanders. As to voters’ gender, it was found that females in the two younger generations became more likely to support the KMT after 1993. Among the three generations, 170 however, first-generation males were the most loyal group to support the KMT since 1990. As to the social bases of support for the DPP, it was found that the DPP received more support from second-generation males between 1987 and 1993. In 1996, third- generation males became the most loyal supporters of the DPP among the six gender groups. The emergence of the NP significantly affected the KMT’s supporters. The NP was more likely to attract younger females in 1996. The KMT received less support from educated voters than did the DPP. Between 1987 and 1993, educated people were more likely to prefer the KMT, but the KMT gained more support from people with less than a high school education in 1996. As to the social bases of support for the DPP, it was found that people with college and above tended to support the DPP in 1987, 1990, and 1996. However, when different generations were examined, it was found that people with less than a high school education were more likely to support the DPP in 1993, but the DPP received more support from educated people in the two younger generations in 1996. A8 to the NP, educated people also were more likely to support that party. As to people’s vocation, government employees were still loyal supporters of the KMT, but the KMT received less support from civil servants in the third generation. On the other hand, support of the DPP by younger government employees increased since 1993. The NP received support from government employees in the first and third generations. People’s subjective socioeconomic identity also affected their partisan preferences. As to people’s subjective socioeconomic identity, the KMT received more support from people with middle- and upper-class identity between 1987 and 1993, but 171 the class difference in KMT support was small in 1996. As to the social bases of support for the DPP, it attracted more people with lower-class identity in the two younger generations between 1990 and 1993. However, there was no consistent pattern of DPP support from different social classes. In 1996, people with middle-class and above identity were also more likely to support the NP. How Political Preferences Affected People’s Partisan Preferences People’s political attitudes played an important role in determining their partisan preferences in 1993. Their national identity, concerns about "money politics," and democratic values significantly affected their partisan preferences. People with Taiwanese identity, more concern about "money politics," and higher scores on democratic values were more likely to support the DPP. However, the emergence of the NP changed the relative strength of the KMT and the DPP. For voters across the three generations, their national identity, concern about "money politics," and democratic values still had a significant effect on their preference for the DPP versus the KMT, but the magnitudes of these coefficients decreased in 1996. The difference in support between the DPP and the NP came from people’s national identity. That is, people with Taiwanese identity were more likely to support the DPP. As to preference for the KMT versus the NP, people’s democratic values played a more important role, and the NP received more support from people with higher scores on democratic values. 172 Partisan Preference and National Identity In this dissertation, I demonstrated that the distribution of national identity among first-generation voters was very stable between 1993 and 1996. Thirty percent of first- generation voters had Chinese identity in 1993 and 1996, and about 35 percent of them had Taiwanese identity. However, voters in second and third generations became less likely to identify themselves as Chinese in 1996. Among Taiwanese, these voters became more likely to have Taiwanese identity in 1996. The proportion of Taiwanese with Taiwanese identity increased by 11.3 percent (from 33.7 percent to 41.6 percent) between 1993 and 1996. On the other hand, the proportion of Taiwanese with Chinese identity decreased by 7.9 percent (from 28.6 percent to 17.3 percent) between 1993 and 1996. As to mainlanders, they became less likely to have Chinese identity, and the proportion with Chinese identity decreased by 30.9 percent (from 74.7 percent to 43.8 percent) between 1993 and 1996. Mainlanders became more likely to have "both Chinese and Taiwanese" identity in 1996. In 1995 and 1996, mainland China initiated several missile tests near Taiwan and staged military exercises. One consequence of these actions was that people in Taiwan become less likely to have a positive attitude toward China. As a result, the proportion of people with Chinese identity decreased dramatically between 1993 and 1996. In this dissertation, I also examined the function of party identification in shaping national identity. Voters in the three generations employed their preference for the DPP versus the KMT to determine their national identity. However, when each generation was examined separately in 1996, it was found that only second-generation voters employed their attitudes toward the three major parties to determine their national identity. As to 173 voters in the first and the third generations, they did not employ their attitude toward the KMT, the ruling party in Taiwan since 1949, to decide their national identity in 1996. However, they used their attitudes toward two relatively new parties, the DPP and the KMT, to determine their national identity. Therefore, I argue that voters in the second generation might formulate their party identification to understand the complex politics in Taiwan. Because the limitation of available data, I am not sure whether voters in the first and the third generations formulated a long-term commitment toward a particular party and employed it as a cue to understand politics. Suggestions for Future Research One major limitation of most empirical studies is that available data sets cannot completely answer the research questions. However, employing limited data sets, I presented that party identification is a useful concept in understanding the political participation of people in Taiwan. During the last two decades, Taiwan experienced a great transformation both politically and socially. People’s party identification provides a guide to understand the electoral competition among major parties in Taiwan. The dissertation also generated some directions for future research. I began by examining whether some concepts developed in the United State could be applied to Taiwan. A better way to examine the construct validity of some concepts would be to employ empirical data sets to test them. Therefore, in future research, if difference measurements of partisan preference are applied to the same survey, the researcher can examine which measurements capture people’s long-term commitment to political parties. 174 However, a better research design would employ panel studies to examine the stability of people’s partisan preferences on different measurements. Second, most surveys in Taiwan have used representative samples to make inferences about the population. However, as demonstrated in this study, there were too few mainlanders in the sample to make confident inferences. There are at least two ways to solve this problem. One is to oversample mainlanders so that more mainlanders are included in the data set, allowing to make more accurate inferences. Second is to interview mainlanders only, and to examine their attitudes toward national identity, partisan preferences, and other political issues. As demonstrated in this study, mainlanders were the group most loyal to the KMT. After President Lee Teng-hui promoted Taiwanese identity, recruited mostly Taiwanese into the power circle of the KMT, and the NP emerged, mainlanders defected from the KMT to support the NP. It would be interesting to examine whether political change in Taiwan has made mainlanders’ political efficacy lower than it once was. Future studies would need to include more mainlanders to make valid inferences. Finally, it might be interesting to examine whether there is an interviewer’s ethnicity effect on people’s responses to national-identity and other ethnicity-related questions. Because past survey data did not include interviewers’ ethnicity, there is no way to know whether or not the change in people’s national identity was an artifact. In the future, interviewers’ ethnic background should be recorded, and researchers should examine whether it significantly affects people’s ethnic attitudes and national identity responses. 175 APPENDICES 176 APPENDIX A Table A-1. Great Events in Taiwan Year Event 1945 The Chinese government took over Taiwan after Japan surrendered 1946 The Constitution was ratified 1947 Parliament (Legislative Yuan, National Assembly, and Control Yuan) members were elected; February 28 Incident ignited Taiwanese rebellion against the Nationalist govemment 1949 The Nationalisthovemment fled to Taiwan; martial law was imposed 1950 Land reform program was instituted in Taiwan 1951 First popular elections for mayors, magistrates, and provincial assemblymen were instituted 1958 Mainland China conducted a massive artillery bombardment of Quemoy from August 23 to October 4 1960 Lei Cheng was jailed; effort to establish a new party failed 1971 Taiwan was expelled from the United Nations 1972 First parliament election since 1947 was held 1977 Chungli Incident, the first violent demonstration 1978 US-PRC relations were normalized 1979 Kaohsitmg Incident, most Tang-wai leaders were jailed 1980 The Election Law was passed 1983 Tangwai organized a quasi-party organization, the Tangwai Candidates Campaign Committee, to endorse candidates and coordinate election activities in the 1983 legislative election 1986 The DPP was established 1987 Martial law was lifted 1989 The first election since lifting of martial law was held 1991 All representatives of the National Assembly were elected by the voters in Taiwan 1992 All legislators of the Legislative Yuan were elected by the voters in Taiwan 1993 The NP was established Source: Hsieh and Niou 1996; Liu 1990: 36; Sheng 1986: 86-89; Tien 1989, 1996. 177 APPENDIX B Table B-1. Distribution of Partisan Preference‘: 1987 to 1996 Partisan Preference 1987 1990 1993 1996 Range Mean Prefer KMT 44.6 47.3 38.3 31.8 15.5 40.5 Prefer DPP 8.2 10.2 12.9 15.5 7.3 11.7 Indifferent 47.2 42.5 48.8 52.7 10.2 47.8 N 1,359 1,001 1,300 1,239 1 Question wording for the 1987 survey was Party ID_A measurement, 1990 and 1996 was like/dislike questions for two major parties, and 1993 was PID_C measurements. See Chapter 3 for discussion of different measurements. 178 Table B-2. Distribution of Partisan Preferences‘, by Generation, 1987 to 1996 Generation Partisan 1987 2 1990 1993 1996 Range Mean Preference 1961 and After Prefer KMT 58.2 49.8 42.4 29.1 29.1 44.9 Prefer DDP 9.3 9.0 12.2 18.6 9.6 12.3 Indifferent 32.5 41.2 45.4 52.3 19.8 42.9 (N) (452) (333) (441) (533) 1943 to 1960 Prefer KMT 40.5 45.8 33.3 28.4 17.4 37.0 Prefer DDP 9.6 12.3 15.0 14.5 5.4 12.9 Indifferent 49.9 41.9 51.7 57.1 15.2 50.2 (N) (469) (400) (492) (455) 1942 and Before Prefer KMT 34.9 46.4 40.1 43.8 11.5 41.3 Prefer DDP 5.5* 8.6* 10.6 10.8 5 .3 8.9 Indifferent 59.6 45 .0 49.3 45.4 14.6 49.8 (N) (438) (267) (367) (251) I For wording of questions on each survey, see Table B-1 and Chapter 3. 2 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. *N<25. 179 Table B-3. Partisan Distribution’ Among Taiwanese, by Generation, 1987 to 1996 Generation Partisan 1987 2 1990 1993 1996 Range Mean Preference 1961 and After Prefer KMT 55.0 46.9 37.8 26.1 28.9 41.5 Prefer DDP 10.8 10.5 14.4 19.7 9.2 13.9 Indifferent 34.2 42.6 47.8 54.2 20.0 44.7 (N) (389) (277) (360) (436) 1943 to 1960 Prefer KMT 36.9 42.3 29.4 25.7 16.6 33.6 Prefer DDP 10.2 12.8 16.8 15.6 6.6 13.9 Indifferent 52.9 44.9 53.8 58.7 13.8 52.6 (N) (423) (352) (435) (397) 1942 and Before Prefer KMT 27.5 35 .0 29.9 39.3 11.8 32.9 Prefer DDP 5.9* 10.7* 13.9 13.5* 8.0 11.0 Indifferent 66.6 54.3 56.2 47.2 19.4 56.1 (N) (371) (206) (274) (178) 1 For wording of questions on each survey, see Table B-1 and Chapter 3. 2 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. *N<25. 180 Table B-4. Partisan Distribution‘ Among Mainlanders, by Generation, 1987 to 1996 Generation Partisan 1987 2 1990 1993 1996 Range Mean Preference 1961 and After Prefer KMT 76.7 63.6 64.9 42.7 34.0 62.0 Prefer DDP 0.0* 1.8* 2.7* 12.4* 12.4 4.2 Indifferent 23.3* 34.6* 32.4* 44.9 21.6 33.8 (N) ( 60) ( 55) ( 74) ( 89) 1943-1960 Prefer KMT 73.3 71.7 62.5 48.0* 25.3 63.9 Prefer DDP 4.4* 6.5* 2.1* 8.0"‘ 5.9 5.3 Indifferent 22.3 * 21 .8* 35 .4* 440* 22.2 30.9 (N) ( 45) ( 46) ( 48) ( 50) 1942 and Before Prefer KMT 78.1 88.1 73.2 54.9 33.2 73.6 Prefer DDP 3.1* 1.7* 0.0* 4.2* 4.2 2.3 Indifferent 18.8* 10.2* 26.8* 40.9 30.7 24.2 (N) (64) (59) ( 71) (71) I For wording of questions on each survey, see Table B-1 and Chapter 3. 2 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. *N<25. 181 Table B-5. Distribution of the Electorate‘ by Political Generations Generations 1987 2 1996 3 1961 and After 16.0 38.8 1943 to 1960 49.4 39.0 1942 and Before 34.6 22.2 Source: 1987 and 1996 Population and Housing Survey Report, R.O.C. ‘ The electorate is defined as citizens 20 years old and above. 2 In 1987, the generations were defined as 1963 and after, 1943 to 1962, and 1942 and before. 3 In 1996, the generations were defined as 1962 and after, 1943 to 1961, and 1942 and before. 182 APPENDIX C The "party preference" index measures relative party strength as the difference between the percentage of one group preferring the KMT and the percentage of the same group preferring the DPP. Table C-l Party Preference Index,‘ by Gender, 1987 to 1996 Generation Gender 1987 2 1990 1993 1996 Range Mean 1961 and After Male 52.1 40.4 25.7 8.4 43.7 31.7 Female 45.3 41.2 34.5 12.7 32.6 33.4 Difference 3 6.8 -0.8 -8.8 -4.3 1943 to 1960 Male 29.7 27.6 10.0 6.4 23.3 18.4 Female 32.2 40.3 27.4 21.6 18.7 30.4 Difference 3 -2.5 -12.7 -17.4 -15.2 1942 and Before Male 38.8 45.5 32.2 36.1 13.3 38.2 Female 20.6 25.0 25.6 29.0 4.0 25.1 Difference 3 18.2 20.5 6.6 7.1 Total Male 40.3 36.8 21.8 13 .9 26.4 28.2 Female 32.5 37.4 29.5 19.0 18.4 29.6 Difference 3 7.8 -0.6 -7.7 -5.1 National Mean 36.4 37.1 25.4 16.3 20.8 28.8 N 1,359 1,001 1,300 1,239 1 For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. Each entry is the proportion of KMT supporter minus the proportion of DPP supporters. A positive sign indicates a KMT plurality. 2 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 3 Difference = Male - Female. 183 Table C-2. Party Preference Index,‘ by Level of Education, 1987 to 1996 Generation Level of Education 2 1987 2 1990 1993 1996 Range Mean 1961 and After Less than high school 32.3 29.4 14.8 12.0 20.3 22.1 High school 54.8 48.7 28.0 14.5 40.3 36.5 College and above 53.0 37.8 41.5 5.1 47.9 34.4 Difference 4 20.7 8.4 26.7 -6.9 1943 to 1960 Less than high school 20.2 19.8 7.4 14.1 12.8 15.4 High school 40.6 48.8 29.4 15.8 33.0 33.7 College and above 60.0 42.8 32.7 10.3 49.7 36.5 Difference 4 39.8 23.0 25.3 -3.8 1942 and Before Less than high school 26.4 31.9 23.1 29.7 8.8 27.8 High school 48.5 46.9 43.6 48.3 4.9 46.8 College and above 64.3 67.7 62.2 37.8 29.9 58.0 Difference 4 37.9 35.8 39.1 8.1 Total Less than high school 25.4 26.6 15.5 19.4 11.1 21.7 High school 49.5 48.5 30.1 17.3 32.2 36.4 College and above 57.4 43.8 41.2 10.2 47.2 38.2 Difference 4 32.0 17.2 25.7 -9.2 National mean 36.4 37.1 25.4 16.3 20.8 28.8 N 1,359 1,001 1,300 1,239 ‘ For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. Each entry is the proportion of KMT supporter minus the proportion of DPP supporters. A positive sign indicates a KMT plurality. 4 Levels of education for each category were: less than high school = 0 to 9 years, high school = 10 to 12 years, and college and above = 13 years and above. 4 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 4 Difference = ”College and above" - "Less than High School." * N<25. 184 Table C-3. Party Preference Index,‘ by Vocation, 1987 to 1996 Generation Vocation 1987 2 1990 1993 1996 Range Mean 1961 and After Government Employee 78.8 75.0 45.0 13.3 65.5 53.0 Other 45.0 39.2 28.2 10.3 34.7 30.7 Difference 3 33.8 35.8 16.8 3.0 1943 to 1960 Government Employee 60.5 57.1 50.0 27.1 33.4 48.7 Other 27.7 32.0 13.3 12.3 19.7 21.3 Difference 3 32.8 25.1 36.7 14.8 1942 and Before Government Employee 82.3 80.0 70.6 51.0 31.3 71.0 Other 27.8 34.5 20.5 28.5 14.0 27.8 Difference 3 54.5 45.5 50.1 22.5 Total Government Employee 71.6 68.0 57.1 31.2 40.4 57.0 Other 33.3 35.1 20.5 14.3 20.8 25.8 Difference 3 38.3 32.9 36.6 16.9 National Mean 36.4 37.1 25.4 16.3 20.8 28.8 N 1,359 1,001 1,300 1,239 ‘ For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. Each entry is the proportion of KMT supporter minus the proportion of DPP supporters. A positive sign indicates a KMT plurality. 4 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 4 Difference = "Government Employee" - "Other." * N<25. 185 Table 04. Party Preference Index,‘ by Subjective Socioeconomic Status, 1987 to 1996 Generation Subjective 1987 3 1990 1993 1996 Range Mean Socioeconomic Status 1961 and After Below middle class 34.9 33.3 10.0 7.0 27.9 21.3 Middle class and above 51.1 42.3 34.8 10.7 40.4 34.7 Difference 4 16.2 9.0 24.8 3.7 1943 to 1960 Below middle class 27.9 18.9 3.5 7.2 20.7 14.4 Middle class and above 31.9 38.4 21.2 15.2 23.2 26.7 Difference 4 4.0 19.5 17.7 8.0 1942 and Before Below middle class 21.4 32.9 11.0 39.5 28.5 26.2 Middle class and above 32.8 40.7 38.9 30.1 10.6 35.6 Difference 4 1 1.4 7.8 27 .9 -9.4 Total Below middle class 26.4 26.9 8.4 16.6 18.5 19.6 Middle class and above 39.1 40.3 30.4 15.5 23.6 31.3 Difference 4 12.7 13.4 22.0 -1.1 National Mean 36.4 37.1 25.4 16.3 20.1 28.8 (N) 1,359 1,001 1,300 1,239 ‘ For the wording of question on each survey, see Table B-1, Appendix B, and Chapter 3. Each entry is the proportion of KMT supporter minus the proportion of DPP supporters. A positive sign indicates a KMT plurality. 2 "Below middle class" includes low-middle class and low class. "Middle class and above" includes middle class, middle-high class, and high class. 3 For 1987 the actual cohort categories were 1958 and after, 1943 to 1957, and 1942 and before. 4 Difference = "Middle class and above" - " Below middle class." *N<25. 186 APPENDIX D HAUSMAN TEST FOR MODEL SPECIFICATION I use OLS to estimate coefficients for the national identity model in chapter 7. I used the Hausman specification error test to detect whether there was a simultaneity problem in my model. For 1993, my model is as follows: Y1 = ,60 +6le + Ax, + Ax, + Ax, + ,65X4 + fiéxs + ,6,X6 + ,53X7 +6l ........ (D1) Y2=a0+a,1/,+612X3+613X4+614X7+82 .......................................................... (DZ) where Yl is KMT versus DPP partisan preference Y2 is national identity X 1 is democratic values X 2 is concern about "money politics" X 3 is a dummy variable for mainlander X 4 is the educational level X 5 is subjective socioeconomic status X 6 is a dummy variable for government or KMT official employee X 7 is a dummy variable for female. 187 If there is no simultaneity problem, i.e., Y, and Y2 are mutually independent, Y2 and 6, should be uncorrelated. From equation (D1) and equation (D2), we get the following reduced-form equations: Y, = )0 + ),X, + bx, +)3X3 +74X4 + 75X5 +16X6 +),X7 +u, ............ (D3) Y2 = 6,, +6,X, +62X2 +cS3X3 + 64X, +(55X5 +66X6 +(§,X7 +u2 ............ (D4) The Hausman test uses the following steps. First, I regressed Y, on explanatory variables in equation (D3) to get 12,. Then, I regressed Y2 on explanatory variables in equation (DZ) and 1?, , and performed a t test on the coefficient of 12, . The null hypothesis for the Hausman test is that there is no simultaneity. If it is not significant, I fail to reject the null hypothesis and there is no simultaneity problem in equation (D2). If the coefficient for ii, is significant, I reject the null hypothesis and there is a simultaneity problem. The results are reported in the following table: Variables Y, X 3 X 4 X 7 ti, Coefficients .255 .943 .145 -.073 -.032 Std. Error .075 .133 .031 .075 .080 t Values 3.412 7.098 4.630 -.973 -.396 Adj. R2 =.238 SEE = 1.1245 df=981 Conditional Index = 25.333 From the above table, we find that the coefficients for the explanatory variables are close to the OLS estimation reported in Table 7-4. However, the coefficients for 1?, are not 188 significant. Therefore, I fail to reject the null hypothesis, and there is no simultaneity problem in this model. As to the 1996 models, there are three variables for partisan preferences, so I used explanatory variables in equation (D3) to get predicted values for partisan preferences and predicted error values. In this situation, I use an F test to see if the coefficients for the three error terms were jointly equal to zero. The F ratio was 1.223 with 3 and 961 degree of freedom. The critical value at 5% was 2.60. Therefore, I failed to reject the null hypothesis and there was no simultaneity problem in my models. 189 BIBLIOGRAPHY 1. Publications in English Abramson, Paul R. 1972. "Intergenerational Social Mobility and Partisan Choice." American Political Science Review 66(4): 1291-1294. Abramson, Paul R. 1974. "Generational Change in American Electoral Behavior." American Political Science Review 68(1): 93-105. Abramson, Paul R. 1979a. "Developing Party Identification: A Further Examination of Life-cycle, Generational, and Period Effects." American Journal of Political Science 23(1): 78-96. Abramson, Paul R. 1979b. "Generational Change and the Decline of Party Identification in America: 1952-1974." American Political Science Review 70(2): 469-478. Abramson, Paul R. 1983. Political Attitudes in America. San Francisco: W. H. Freeman and Company. Abramson, Paul R. 1989. "Generations and Political Change in the United States." Research in Political Sociology 4: 235-180. Abramson, Paul R., and John H. Aldrich. 1982. "The Decline of Electoral Participation in America." American Political Science Review 76: 502-21. Abramson, Paul R., John H. Aldrich, Phil Paolino, and David W. Rohde. 1992. "’Sophisticated’ Voting in the 1988 Presidential Primaries." American Political Science Review 86: 55-69. 190 Abramson, Paul R., John H. Aldrich, and David W. Rohde. 1994. Change and Continuity in the 1992 Elections. Washington DC: CQ Press. Abramson, Paul R., John H. Aldrich, and David W. Rohde. 1998. Change and Continuity in the 1996 Elections. Washington DC: CQ Press. Abramson, Paul R., and Charles W. Ostrom, Jr. 1991. "Macropartisanship: An Empirical Reassessment," American Political Science Review 85: 81-92. Abramson, Paul R., and Charles W. Ostrom. 1994. "Question Wording and Partisanship: Change and Continuity in Party Loyalties During the 1992 Election Campaign" Public Opinion Quarterly 58: 21-48. Achen, Christopher H. 1975. "Mass Political Attitudes and the Survey Response." American Political Science Review 69: 1218-1231. Achen, Christopher H., Yung-ming Hsu, and Su-feng Kuo. 1997. "The Impact of the Straits Crisis on Taiwan’s Presidential Election: China’s Coercion Backfires." Paper presented at the Conference Group on Taiwan Studies at the 1997 annual meeting of the American Political Science Association, Washington DC, August 28-31, 1997. Aldrich, John H. 1995. Why Parties? The Origin and Transformation of Political Parties in America. Chicago: The University of Chicago Press. Aldrich, John H., John L. Sullivan, and Eugene Borgida. 1989. "Foreign Affairs and Issue Voting: Do Presidential Candidates ’Waltz Before A Blind Audience?” American Political Science Review 83:123-141. Anderson, Benedict. 1986. Imagined Communities. London: Verso. Barnes, Samuel H., Peter McDonough, and Antonio L*pez Pina. 1985. "The Development of Partisanship in New Democracies: The Case of Spain." American Journal of Political Science 29(4): 695-720. 191 Bartels, Larry M. 1991. "Instrumental and ’Quasi-Instrumental’ Variables." American Journal of Political Science 35(3): 777-800. Beck, Paul Allen. 1974. "A Socialization Theory of Partisan Realignment." In The Politics of Future Citizens: New Dimensions in the Political Socialization of Children, ed. Richard G. Niemi. San Francisco: Jossey-Bass. Beck, Paul Allen, and M. Kent Jennings. 1975. "Parents as "Middlepersons" in Political Socialization." Journal of Politics 37(1): 83-107. Beck, Paul Allen, and M. Kent Jennings. 1979. "Political Period and Political Participation." American Political Science Review 73(3): 737-50. Berelson, Bernard R., Paul Lazarsfeld, and William N. McPhee. 1954. Voting. Chicago: University of Chicago Press. Bobo, Lawrence, and Franklin D. Gilliam, Jr., 1990. "Race, Sociopolitical Participation, and Black Empowerment." American Political Science Review 84: 377-93. Bound, John, David A. Jaeger, and Regina M. Baker. 1995. "Problems With Instrumental ' Variables Estimation When the Correlation Between the Instruments and the Endogenous Explanatory Variable Is Weak." Journal of American Statistics Association 90(430): 443-50. Brady, Henry E., Sidney Verba, and Kay Lehman Schlozman. 1995. "Beyond SES: A Resource Model of Political Participation." American Political Science Review 89: 271-94. Brody, Richard A. and Benjamin 1. Page. 1972. "The Assessment of Policy Voting," American Political Science Review 66: 450-58. Brody, Richard A. 1978. "The Puzzle of Political Participation in America." In The New American Political System, ed. Anthony King. Washington, D. C.: American Enterprise Institute. 192 Brown, Michael E. 1993. Ethnic Conflict and International Security. Princeton University Press. In The Ethnicity Reader: Nationalism, Multiculturalism, and Migration, ed. Montserrat Guibemau and John Rex. 1997. Cambridge: Polity Press. Budge, Ian, Ivor Crewe, and Dennis Farlie, eds. 1976. Party Identification and Beyond. London: Wiley. Budge, Ian, and Dennis Farlie. 1976. "A Comparative Analysis of Factors Correlated with Turnout and Voting Choice." In Party Identification and Beyond, ed. Ian Budge, Ivor Crewe, and Dennis Farlie. London: Wiley. Bumham, Walter Dean. 1970. Critical Elections and the Mainsprings of American Politics. New York: Norton. Butler, D., and D. Stokes. 1974. Political Change in Britain. 2“d ed. London: Macmillan. Cain, Bruce E. and John Ferejohn. 1981. "Party Identification in the United States and Great Britain." Comparative Political Studies 14(1): 31-47. Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1960. The American Voter. New York: John Wiley & Sons. Campbell, Angus, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1966. Elections and the Political Order. New York: John Wiley & Sons. Campbell, Angus, and Henry Valen. 1966. "Party Identification in Norway and the United States." In Elections and the Political Order, ed. Angus Campbell, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. New York: John Wiley & Sons. Campbell, Donald T., and Julian C. Stanley. 1966. Experimental and Quasi-Experimental Designs for Research. Chicago: Rand-McNally. Carmines, Edward G., and James A. Stimson. 1980. "The Two Faces of Issue Voting," American Political Science Review 74: 78-91. 193 Carmines, Edward G., and Richard A. Zeller. 1979. Reliability and Validity Assessment. Sage University Paper Series on Quantitative Applications in the Social Sciences, 07-17. Newbury Park, CA: Sage. Cassel, Carol A., and Robert Luskin. 1988. "Simple Explanations of Turnout Decline." American Political Science Review 82: 1321-30. Chang, Yu-tzung. 1997. "Ethnic Conflict and Democratic Consolidation in Taiwan: Dissolving the Logic of Nation-State and Democratic Policies." Issues & Studies 33(4): 77-93. Chen, Lu-huei. 1998. "Presidential Voting of 1996 in Taiwan: An Analysis." Journal of Electoral Studies 5(2): 161-84. Cheng, Tun-jen, and Yung-ming Hsu. 1996. "Issue Structure, the DPP’s Factionalism, and party Realignment." In Taiwan’s Electoral Politics and Democratic Transition: Riding the Third Wave, ed. Hung-Mao Tien. Armonk, New York: ME. Sharpe. Chu, Yun-han. 1992. Crafting Democracy in Taiwan. Taipei: Institute for National Policy Research. Chu, Yun-han. 1998. "Taiwan’s Unique Challenges." In Democracy in East Asia, ed. Larry Diamond and Marc F. Plattner. Baltimore: The Johns Hopkins University Press. Chu, Yun-han, and Tse-min Lin. 1996. "The Process of Democratic Consolidation in Taiwan: Social Cleavage, Electoral Competition, and the Emerging Party System." In Taiwan’s Electoral Politics and Democratic Transition: Riding the Third Wave, ed. Hung-Mao Tien. Armonk, New York: ME. Sharpe. Claggett, William. 1981. "Partisan Acquisition Versus Partisan Intensity: Life-Cycle, Generation, and Period Effects, 1952-1976." American Joumal of Political Science 25(2): 193-214. 194 Clarke, Harold D., and Marianne C. Stewart. 1984. "Dealignment of Degree: Partisan Change in Britain, 1974-83." Journal of Politics 46: 689-718. Clarke, Harold D., and Marianne C. Stewart. 1987. "Partisan Inconsistency and Partisan Change in Federal States: The Case of Canada." American Journal of Political Science 31(2): 383-407. Converse, Philip E. 1964. "The Nature of Belief Systems in Mass Public." In Ideology and Discontent, ed. David Apter. New York: Free Press. Converse, Philip E. 1969. "Of Time and Partisan Stability." Comparative Political Studies 2: 139-71. Converse, Philip E. 1976. The Dynamics of Party Support: Cohort-Analyzing Party Identification. Beverly Hill: Sage. Converse, Philip E. 1979. "Rejoinder to Abramson." American Journal of Political Science 23: 97-100. Converse, Philip E., and Georges Dupeux. 1962. "Politicization of the Electorate in France and the United State." Public (Opinion Quarterly 26: 1-26. In Angus Campbell, Philip E. Converse, Warren E. Miller, and Donald E. Stokes. 1966. Elections and the Political Order. New York: John Wiley & Sons. Converse, Philip E., and Gregory B. Markus. 1979. “‘Plus 0a Change...’ The New CPS Election Study Panel.” American Political Science Review 73:2-49. Converse, Philip E., and Roy Pierce. 1985. "Measuring Partisanship." Political Methodology 11: 143-66. Cook, Thomas D. and Donald T. Campbell. 1979. Quasi-Experimentation: Design and Analysis Issues for Field Settings. Chicago: Rand-MCN ally. Crittenden, John. 1962. "Aging and Party Affiliation." Public Opinion Quarterly 26:648- 57. 195 Crewe, Ivor. 1976. "Party Identification Theory and Political Change in Britain." In Party Identification and Beyond, ed. Ian Budge Ivor Crewe, and Dennis Farlie. London: Wiley. Cutler, Neal E. 1969-70. "Generation, Maturation, and Party Affiliation: A Cohort Analysis." Public Opinion Quarterly 332583-88. Dalton, Russell J. 1998. Citizen Politics. 2'”d ed. Chatham, New Jersey: Chatham House. Dalton, Russell J ., and Martin P. Wattenberg. 1993. “The Not So Simple Act of Voting.” in Political Science: The State of the Discipline H, ed. Ada Finifter. Washington, DC: American Political Science Association. Diamond, Larry, and Marc F. Plattner, eds. 1994. Nationalism, Ethnic Conflict, and Democracy. Baltimore: The Johns Hopkins University Press. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper & Row. Erikson, Robert S., Michael B. MacKuen, and James Stimson. 1998. “What Moves Macropartisanship? A Response to Green, Palmquist, and Schickler.” American Political Science Review 92:901-12. Finifter, Ada W. 1974. "The Friendship Group as a Protective Environment for Political Deviants." American Political Science Review 68: 607-25. Fiorina, Morris P. 1981. Retrospective Voting in American National Politics. New Haven: Yale University Press. Fleury, Christopher J ., and Michael S. Lewis-Beck. 1993. "Anchoring the French Voter: Ideology Versus Party." Journal of Politics 55:1100-09. Franklin, Charles H., 1985. "The Dynamics of Political Identification." American Political Science Review 77:957-73. 196 Franklin, Charles H., and John E. Jackson. 1983. "The Dynamics of Political Identification." American Political Science Review 772957-73. Glenn Norval D., 1972. "Sources of the Shift to Political Independence: Some Evidence From a Cohort Analysis." Social Science Quarterly 53: 494-519. Glenn Norval D., 1977. Cohort Analysis. Beverly Hills, CA: Sage. Glenn Norval D., and Ted Hefner. 1972. "Further Evidence on Aging and Party Identification." Public Opinion Quarterly 36:31-47. Green, Donald Philip, and Bradley Palmquist. 1990. "Of Artifacts and Partisan Instability." American Journal of Political Science 34: 872-902. Green, Donald Philip, Bradley Palmquist, and Eric Schickler. 1998. "Macropartisanship: ‘ A Replication and Critique." American Political Science Review 92: 883-899. Greenstein, Fred 1. 1965. Children and Politics. New Haven, Conn.: Yale University Press. Hess, Robert D., and Judith V. Tomey. 1965. The Development of Basic Attitudes and Values Toward Government and Citizenship During the Elementary School Years. Chicago: University of Chicago. Higley, John, Tong-yi Huang, and Tse-min Lin. 1998. "Elite Settlement and Democratic Consolidation in Taiwan." Journal of Democracy 9(2): 148-63. Holmberg, S*ren. 1994. "Party Identification Compared Across the Atlantic." In Elections at Home and Abroad, ed. M. Kent Jennings and Thomas E. Mann. Ann Arbor: University of Michigan Press. Horowitz, Donald L. 1994. "Democracy in Divided Societies." In Nationalism, Ethnic Conflict, and Democracy, ed. Larry Diamond and Marc F. Plattner. Baltimore and London: The Johns Hopkins University Press. 197 Hsieh, John Fuh-sheng. 1996. "The SNTV System and Its Political Implication." In Taiwan’s Electoral Politics and Democratic Transition: Riding the Third Wave, ed. Hung-Mao Tien. Armonk, New York: ME. Sharpe. Hsieh, John Fuh-sheng, and Emerson M. S. Niou. 1996. "Salient Issues in Taiwan’s Electoral Politics" Electoral Studies 15(2): 219-35. Hu, Fu, and Yun-han Chu. 1992. "Electoral Competition and Political Democratization in Taiwan." In Political Change in Taiwan, ed. Tun-jen Cheng and Stephan Haggard. Boulder and London: Lynne Rienner Publishers. Huang, Chi, and Samuel S. G. Wu. 1995. "Inherited Rivalry: A Chronology." In Inherited Rivalry: Conflict Across the Taiwan Straits, ed. Tun-jen Cheng, Chi Huang, and Samuel S. G. Wu. Boulder: Lynne Rienner Publishers. Huang, The-fu. 1997. "Party Systems in Taiwan and South Korea." In Consolidating the Third Wave Democracies: Themes and Perspective, ed. Larry Diamond, Marc F. Plattner, Yun-han Chu, and Hung-mao Tien. Baltimore and London: The Johns Hopkins University Press. Huang, Teh-fu. 1996. "Elections and the Evolution of the Kuomintang." In Taiwan’s Electoral Politics and Democratic Transition: Riding the Third Wave, ed. Hung- Mao Tien. Armonk, New York: ME. Sharpe. Hyman, Herbert H. 1959. Political Socialization: A Study in the Psychology of Political Behavior. Glencoe, Ill. : Free Press. Inglehart, Ronald. 1977. The Silent Revolution in Europe: Changing Values and Political Styles Among Western Publics. Princeton, NJ: Princeton University Press. Inglehart, Ronald. 1984. "The Changing Structure of Political Cleavages in Western Society." In Electoral Change in Advanced Industrial Democracies: Realignment or Dealignment? ed. R. J. Dalton, S.C. Flanagan, and P. A. Beck. Princeton, NJ: Princeton University Press. 198 Inglehart, Ronald, and Avram Hochstein. 1987. "Alignment and Dealignment of the Electorate in France and the United States." Comparative Political Studies 5: 343- 72. Jackson, John E. 1975. "Issues, Party Choices, and Presidential Votes." American Journal of Political Science 19(2): 161-85. Jennings, M. Kent. 1979. "Another Look at the Life Cycle and Political Participation." American Journal of Political Science 23(4): 755-71. Jennings, M. Kent, and Thomas E. Mann, eds. 1994. Elections at Home and Abroad. Ann Arbor: University of Michigan Press. Jennings, M. Kent, and Gregory B. Markus. 1984. "Partisan Orientations Over the Long Hall: Results From the Three-Wave Political Socialization Panel Study." American Political Science Review 78: 1000-18. Jennings, M. Kent, and Richard G. Niemi. 1975. "Continuity and Change in Political Orientation: A Longitudinal Study of Two Generations." American Political Science Review 69(4): 1316-35. Jennings, M. Kent, and Richard G. Niemi. 1978. "The Persistence of Political Orientations: An Over-Time Analysis of Two Generations." British Journal of Political Science 8: 333-63. Jennings, M. Kent, and Richard G. Niemi. 1981. Generations and Politics. Princeton: Princeton University Press. Johnston, Richard. 1992. "Party Identification Measures in the Anglo-American Democracies: A National Survey Experiment." American Journal of Political Science 36(2): 542-559. Kaase, Max. 1976. "Party Identification and Voting Behavior in the West German Election of 1969." In Party Identification and Beyond, ed. Ian Budge, Ivor Crewe, and Dennis Farlie. London: Wiley. 199 Kaase, Max. and Hans-Dieter Klingemann. 1994. "The Cumbersome Way to Partisan Orientations in a ’New’ Democracy: The Case of the Former GDR." In Elections at Home and Abroad, ed. M. Kent Jennings and Thomas E. Mann. Ann Arbor: University of Michigan Press. Kau, Michael. 1988. "Taiwan and Beijing’s Campaigns for Unification." In Taiwan in a Time of Transition, ed. Harvey Feldman, Michael Y. M. Kau, and Ilpyong J. Kim. New York: Paragon House Publishers. Kish, Leslie. 1965. Survey Sampling. New York: John Wiley and Sons. Keith, Bruce E., David B. Magleby, Candice J. Nelson, Elizabeth Orr, Mark C. Westlye, and Raymond E. Wolfinger. 1986. "The Partisan Affinities of Independent Leaners." British Journal of Political Science 16:155-85. Keith, Bruce E., David B. Magleby, Candice J. Nelson, Elizabeth Orr, Mark C. Westlye, and Raymond E. Wolfinger. 1992. The Myth of the Independent Voter. Berkeley, Los Angeles, and London: University of California Press. Key, V. 0., Jr. 1964. Politics, Parties, and Pressure Groups. 5th ed. New York: Thomas Y. Crowell. Key, V. 0., Jr. 1959. "Secular Realignment and the Party System." Journal of Politics 21: 198-210. Key, V. 0., Jr. 1955. "A Theory of Critical Elections." Journal of Politics 17: 3-18. Knoke, David, and Michael Hout. 1974."Socia1 and Demographic Factors in American Party Affiliation, 1952-72." American Sociological Review 39(5): 700-13. Kritzer, Herbert M. 1983. "The Identification Problem in the Cohort Analysis." Political Methodology 9:35-50. Lazarsfeld, Paul F., Bernard Berelson, and Hazel Gaudet. 1944. The People’s Choice: How the Voter Makes Up his Mind in a Presidential Campaign. New York: Columbia University Press. 200 LeDuc, Lawrence, Harold D. Clarke, Jane Jenson, and Jon H. Pammett. 1984. "Partisan Instability in Canada: Evidence from a New Panel Study." American Political Science Review 782470-84. Lin, Chia-lung. 1998. "Paths to Democracy: Taiwan in Comparative Perspective." Ph. D. dissertation, Yale University. Lin, Tse-min, Yun-han Chu, and Melvin J. Hinich. 1996. "Conflict Displacement and Regime Transition in Taiwan: A Spatial Analysis." World Politics 48: 453-81. Lin, Tse-min, Yun-han Chu, Tong-yi Huang, and Baohui Zhang. 1998. "Elections and Elite Convergence: The Consolidation of Democracy in Taiwan." Working paper, Department of Government, University of Texas at Austin. Lipset, Seymour Martin. 1981. Political Man: The Social Bases of Politics. Baltimore: Johns Hopkins University Press. Lipset, Seymour Martin, and Stein Rokkan, eds. 1967. Party Systems and Voter Alignments. New York: Free Press. Liu, I-chou. 1990. "The Electoral Effect of Social Context Control on Voters: The Case of Taipei, Taiwan." Ph. D. dissertation, University of Michigan. Liu, I-chou. 1995a. "Generational Divergence in Party Image Among Taiwan Electorate." Issues & Studies 31(2):87—114. Liu, I-chou. 1996. "The Behavior of Taiwanese Voters in 1992: Consolidation of Partisan Ties." In Taiwan ’s Electoral Politics and Democratic Transition: Riding the Third Wave, ed. Hung-Mao Tien. Armonk, New York: ME. Sharpe. Lui, Fei-lung. 1992. "The Electoral System and Voting Behavior in Taiwan." in Political Change in Taiwan, ed. Tun-jen Cheng and Stephan haggard. Boulder and London: Lynne Rienner Publishers. 201 Luskin, Robert C., John P. McIver, and Edward G. Carrnines. 1989. "Issues and the Transmission of Partisanship." American Journal of Political Science 33:440-58. MacKuen, Michael B., Robert S. Erikson, and James Stimson. 1989. “Macropartisanship.” American Political Science Review 83: 1 125—42. MacKuen, Michael 8., Robert S. Erikson, and James Stimson; Paul R. Abramson and Charles W. Ostrom, Jr. 1992. "Question Wording and Macropartisanship." American Political Science Review 86: 475-86. Mannheim, Karl. [1928] 1952. "The Problem of Generations." In Essays on the Sociology of Knowledge, ed. Paul Kecskemeti. New York: Oxford University Press. Markus, Gregory B. 1979. "The political Environment and the Dynamics of Public Attitudes: A Panel Study." American Journal of Political Science 23(May):338- 59. Markus, Gregory B. 1983. "Dynamic Modeling of Cohort Change: the Case of Political Partisanship." American Journal of Political Science 27:717-39. Markus, B. Gregory, and Philip E. Converse, 1979. "A Dynamic Simultaneous Equation Model of Electoral Choice," American Political Science Review 73: 1055-70. Mason, Karen Oppenheim, William M. Mason, H. H. Winsborough, and W. Kenneth Poole. 1973. "Some Methodological Issues in Cohort Analysis of Archival Data." American Sociological Review 38:243-58. Miller, Arthur H., and Martin P. Wattenberg. 1983. "Measuring Party Identification: Independent or No Partisan Preference?" American Journal of Political Science 27(1): 106-21. Miller, Warren. 1976. "The Cross-National Use of Party Identification as a Stimulus to Political Inquiry." In Party Identification and Beyond, ed. Ian Budge, Ivor Crewe, and Dennis Farlie. London: Wiley. 202 Miller, Warren E. 1991. "Party Identification, Realignment, and Party Voting: Back to the Basics." American Political Science Review 85(2): 557-68. Miller, Warren E. and J Merrill Shanks. 1996, The New American Voter. Cambridge: Harvard University Press. Nie, Norman H., Sidney Verba, and John R. Petrocik. 1979. The Changing American Voter. Enlarged ed. Harvard University Press. Niemi, Richard G. and M. Kent Jennings, 1991. "Issues and Inheritance in the Formation of Party Identification," American journal of Political Science 35(4): 971-88. Niemi, G. Richard, and Herbert F. Weisberg, eds. 1993a. Classics in Voting Behavior. Washington, D. C.: Congressional Quarterly Inc. Niemi, G. Richard, and Herbert F. Weisberg. 1993b. Controversies in Voting Behavior. 3rd ed. Washington, D. C.: Congressional Quarterly Inc. Norpoth, Helmut. 1978. "Party Identification in West Germany." Comparative Political Studies 11( 1): 36—61. Norpoth, Helmut, and Jerrold G. Rusk. 1982. "Partisan Dealignment in the American Electorate: Itemizing the Deductions since 1964." American Political Science Review 76: 522-37. Page, Benjamin 1. and Richard A. Brody. 1972. "Policy Voting and the Electoral Process: The Vietnam War Issue," American Political Science Review 66: 979-95. Page, Benjarrrin I. and Calvin C. Jones. 1979. "Reciprocal Effects of Policy Preferences, Party Loyalties, and the Vote." American Political Science Review 73(4): 1071- 89. Petrocik, John R. 1974. "An Analysis of Intransitivities in the Index of Party Identification," Political Methodology 1: 31-47. 203 Richardson, Bradley M. 1991. "European Party Loyalties Revisited. " American Political Science Review 85(3): 751-75. Riley, Matilda White. 1973. "Aging and Cohort Succession: Interpretations and Misinterpretations." Public Opinion Quarterly 37: 35-49. Rosenstone, Steven J. and John Mark Hansen. 1993. Mobilization, Participation, and Democracy in America. New York: Macmillan. Schickler, Eric, and Donald Philip Green. 1997. "The Stability of Party Identification in Western Democracies: Results From Eight Panel Surveys." Comparative Political Studies 30(4): 450-83. Sears, David 0., Carl P. Hensler, and Leslie K. Speer. 1979. "Whites’ Opposition to ’Busing’: Self-interest or Symbolic Politics." American Political Science Review 73(2): 369-84. Sears, David 0., Richard L. Lau, Tom R. Tyler, and Harris M. Allen, Jr. 1980. "Self- Interest vs. Symbolic Politics in Policy Attitudes and Presidential Voting." American Political Science Review 74(3): 670-84. Shively, W. Philips. 1972. "Party Identification, Party Choice, and Voting Stability: The Weimar Case." American Political Science Review 66: 1203-25. Shively, W. Phillips. 1979a. "The Development of Party Identification Among Adults: An Exploration of a Functional Model." American Political Science Review 73(4): 1039-54. Shively, W. Phillips. 1979b."Rejoinder to Abramson." Political Methodology 6:457-61. Shively, W. Phillips. 1979c."The Relationship between Age and Party Identification: A Cohort Analysis." Political Methodology 6:437-46. Shively, W. Philips. 1980. "The Nature of Party Identification: A Review of Recent Development." In The Electorate Reconsidered, ed. John C. Pierce and John L. Sullivan. Beverly Hills, Calif.: Sage Publication. 204 Shively, W. Phillips. 1992. “From Differential Abstention to Conversion: A Change in Electoral Change, 1864-1988.” American Journal of Political Science 36:309-30. Smith, Anthony D. 1991. National Identity. London: Penguin Books. Stewart, Marianne C., and Harold D. Clarke. 1998. "The Dynamics of Party Identification in Federal Systems: The Canadian Case." American Journal of Political Science 42(1): 97-116. Sullivan, John L., James Piereson, and George E. Marcus. 1982. Political Tolerance and American Democracy. Chicago: University of Chicago Press. Sundquist, James L. 1983. Dynamics of the Party System. Washington, DC: Brookings Institution. Thomassen, Jacques. 1976. "Party Identification as a Cross-National Concept: Its Meaning in the Netherlands." In Party Identification and Beyond, ed. Ian Budge, Ivor Crewe, and Dennis Farlie. London: Wiley. Tien, Hung-mao. 1989. The Great Transition: Political and Social Change in the Republic of China. Stanford: Hoover Institution Press. Tien, Hung-mao 1996. "Elections and Taiwan's Democratic Development" in Taiwan’s Electoral Politics and Democratic Transition: Riding the Third Wave, ed. Hung- mao Tien. Armonk, New York: ME. Sharpe. Tien, Hung-mao. 1997. "Taiwan's Transformation." In Consolidating the Third Wave Democracies: Regional Challengers, ed. Larry Diamond, Marc F. Plattner, Yun- han Chu, and Hung-mao Tien. Baltimore: The Johns Hopkins University Press. Tien, Hung-mao, and Tun-jen Cheng. 1997. "Crafting Democratic Institutions in Taiwan." The China Journal 37: 1-26. Uslaner, Eric M. 1989. "Multiple Party Identifiers in Canada: Participation and Affect." Journal of Politics 51(4): 993-1003. 205 Weisberg, Herbert F. 1980, "A Multidimensional Conceptualization of Party Identification," Political Behavior 2: 33-60. Whiteley, Paul F. 1988. "The Causal Relationship Between Issues, Candidate Evaluations, Party Identification, and Vote Choice--the View From ’Rolling Thunder’. " Journal of Politics 50(4): 961-84. Wolfinger, Raymond E., and Steve J. Rosenstone. 1980. Who Votes? New Haven: Yale University Press. Wu, Jaushieh Joseph. 1995. Taiwan’s Democratization: Forces Behind the New Momentum. Oxford: Oxford University Press. Zaller, John. 1992. The Nature and Origins of Mass Opinion. New York: Cambridge University Press. 206 2. Publications in Chinese Chang, Mau-kuei. and Hsin-huang Michael Hsiao. 1987. "The Chinese Complex and Taiwanese Complex of College Students: An Analysis of Self-identification and Attitudes toward Inter-marriage." China Tribune 25(1): 34-53. Chang, Mau-kuei. and Hsin-yi Wu. 1997. "Education and Its Effects on Tong-Du Inclinations." Taiwanese Political Science Review 2: 107-89. Chen, Yih-yan. 1994. "Predicating Voter’s Choice in legislator Election 1992: A Cluster Analysis." Journal of Electoral Studies 1(1): 1-37. Chen, Lu-huei. 1992. "Party Identification, Party Evaluation, and the Voter’s Choice in the Second National Assembly Election." In A Study of Taiwanese Voting Behavior in the Second National Assembly Election, ed. Yih-yan Chen et al., A report of research project sponsored by the National Science Council of Executive Yuan. Chen, Lu-huei. 1994. "The Electoral Effect of Party Effort and Social Context on Voters: The Case of Kuomintang Veteran Association (Huang-Fu-Hsing)" Journal of Electoral Study 1(2): 53-96. Chen, Lu-huei. 1995. "The Unification-Independence Issue, and the Voter’s Political Attitudes and Political Participation." In An Interdisciplinary Study of Voting Behavior in the Election for Taiwan Governor, ed The-fu Huang et al. A report of research project sponsored by the National Science Council of Executive Yuan. Chen, Wen-chun. 1995. "Independence vs. Reunification Issue and Voting Behavior in Taiwan: An Analysis of Gubernatorial and Mayoral Elections in 1994." Journal of Electoral Studies 2(2): 99-136. Chu, Yun-han. 1996. "The Partisan Factor in Electoral Choice: A Case Analysis of the 1991 National Assembly Election." Journal of Electoral Studies 3(1): 17-48. Chu, Yun-han. and Yu-tsung Chang. 1995. "Betraying Kuomintang: The Dealignment and Realignment of the Party Identification of the Electorate in Taipei City." Paper presented at the Second Annual Conference of Taiwanese Political Science Association. Taipei: December, 23-24, 1995. 207 Fu, Hung-der. 1994. "Political Culture and Voting Behavior: The Legislative Election of 1989 and the National Assembly Election of 1991." Journal of Electoral Studies (1)2: 27-52. Fu, Hung-der. 1996. "The Determinants of Voting Choice: Structural, Psychological, and Rational Factors." Journal of Electoral Studies 3(2): 157-86. Hawang, Shiow-duan. 1996. "The Importance of Candidate Images and Capabilities in Presidential Elections." Journal of Electoral Studies 3( 1): 103-36. Ho, Szu-yin. 1994. "Taiwan Voters’ Party Preference Change: 1989-1992." Journal of Electoral Studies 1(1): 39-54. Ho, Szu-yin, and J aushieh Joseph Wu. 1996. "Measuring Party Identification in Taiwan’s Party System." Journal of Electoral Studies 3(1): 1-16. Hsieh, John Fuh-sheng, Emerson M. S. Niou., and Huei-ping Lin. 1995. "Issue Voting in the 1994 Gubernatorial and Mayoral Elections: An Application of Rational Choice Approach." Journal of Electoral Studies 2(1): 77-92. Hu, Fu. 1993. "Elections and Political Transformation." Paper presented at the Conference on Electoral Behavior, Constitutional Order and Political Change. Held by Workshop 306 at National Taiwan University, Taipei, December 17, 1993. Hu, Fu. Yu-tsung Chang, and Sheng Ouyang. 1994. "Taiwan’s Political Cleavages ad Its Impact on the Electorate’s National Identity and Voting Choice." Paper presented at the Conference on Democratization, Party Politics, and Elections. Held by Workshop 306 at National Taiwan University, Taipei, July 8, 1994. Hu, Fu, Teh-yu Chen, Ming-ton Chen, and Chia-lung Lin. 1990. The Electorate’s Voting Behavior: An Analysis of the Additional Legislators Elections in 1986. Taipei: Central Election Commission. Hu, Fu, Teh-yu Chen, Ming-ton Chen, and Ying-lung You. 1986. The Electorate’s Voting Behavior. Taipei: Central Election Commission. 208 Hu, Fu, Teh-yu Chen, Yun-han Chu, Huo-yan Shyu, Ying-lung You, and Ming-ton Chen. 1993. The Electorate’s Voting Behavior: An Analysis of the Additional Legislators Elections in 1989. Taipei: Central Election Commission. Huang, Teh-fu. 1992. The Democratic Progressive Party and Political Democratization in Taiwan. Taipei: Taiwan Elite. Huang, Teh-fu. 1994. "Party Competition and Political Democratization: New Challenges to Taiwan’s Party System." Journal of Electoral Studies 1(2): 199-220. Huang, Teh-fu. and Chang Yu-tzung. 1994. "Toward A Three-Party Competition System? Democratic Consolidation and the Change of Party System in Taiwan Area." Chinese Political Science Review 23: 197-225. Lan, Ching-jang. 1996. A Panel Study of Taiwanese Voter’s Party Identification and Change. M. A. thesis, Department of Political Science, National Cheng-chi University. Lin, Chia-lung. 1988. The Social Bases of the Kuomintang and Democratic Progressive Party: A Comparative Analysis of Taiwan Electorate’s Party Support, 1983-1986. M. A. thesis, unpublished, Department of Political Science, National Taiwan University. Lin, Chia-lung. 1989. "The Opposition Movement Under an Authoritarian-Clientelist Regime: Political Explanations on the Social Base of the Democratic Progressive Party in Taiwan." Taiwan Research Quarterly 2(1): 117-43. Lin, Chia-lung. 1994. "Electoral Interactions of the Immigrant Regime, Opposition Movements and Local Factions: The Change of KMT’s Nomination Policy and Taiwan’s Transition to Democracy." Paper presented at the Conference on Toward a Taiwanese Political Science, Held by the Taiwanese Political Science Association, Taipei: December 17-18, 1994. Liu, I-chou. 1987. "Voters’ Party Preferences." In The Voting Behavior in a Transitional Society, Fei-ling Lui et al.. A Report of Research Project Sponsored by the National Science Council of Executive Yuan. 209 Liu, I-chou. 1991. "A Study on the Election Facilitation Effects of the KMT’s Responsibility Zones." In Collection from Symposium on Party Politics and Constitutional Democracy, ed. Democracy Foundation. Taipei: Democracy Foundation. Liu, I-chou. 1995b. "The Conception of ’China’ among the People of Taiwan." Paper presented at the Second Annual Conference of Taiwanese Political Science Association. Taipei: December, 23-24, 1995. Liu, I-chou. 1995c. "The New Party System of Taiwan." Issue and Study 34(10): 1-10. Liu, I-chou. 1996. "Forecasting Elections: Tests of Some ’Simple’ Models." Journal of Electoral Studies 3(2):107-29. Shen, Shiao-chi. 1995. "Ethnic Politics and Taiwan Democratization." Paper presented at the Second Annual Conference of Taiwanese Political Science Association. Taipei: December, 23-24, 1995. Sheng, Shing-yuan. 1986. KMT versus Opposition: An Analysis of the Electoral Competition in 1983 Taiwanese Legislative Election. Taipei: Kuen Kuang. Shyu, Huo-yan. 1991. "Party Identification and the Vote Choice: An Analysis of Electorate’s Party Image, Partisan Preferences and Voting Behavior." Bulletin of Social Sciences and Philosophy, Academia Sinica 4(1): 1-57. Shyu, Huo-yan. 1993. "Electoral Competitions and the Change in Cleavage Structure: The Persistence of Electoral Bases of the KMT and the DPP." Bulletin of Social Sciences and Philosophy, Academia Sinica 6(1): 37-74. Shyu, Huo-yan. 1995. "’Lee Teng-hui Complex’ and Voting Behaviors: A Psycho- Political Analysis of Vote-choices in the 1994 Taiwan’s Gubernatorial and Taipei Mayoral Elections." Journal of Electoral Studies 2(2): 1-36. Shyu, Huo-yan. 1996. "National Identity and Partisan Vote-choices in Taiwan: Evidence From Survey Data Between 1991 and 1993." Taiwanese Political Science Review 1: 85-127. 210 Tsai, Meng-hsi. 1997. Changes in Ethnic Identity, Unification and Independent Position, and Party Preference in Taiwan From 1991 to 1996. M. A. thesis, Department of Political Science. National Chengchi University. Wang, Fu-chang. 1994. "Ethnic Assimilation and Mobilization: An Analysis of Party Support in Taiwan." Bulletin of the Institute of Ethnology, Academia Sinica 77: 1- 34. Wang, Fu-chang. 1995. "The Conflict Between Democratic Politics and Ethnic Politics in Taiwan." Paper presented at the conference of The Building and Consolidation of Democracy in Taiwan, held by Taiwan Research Foundation. Taoyuan: Feb. 28, 1995. Wang, Fu-chang. 1996. "Consensus Mobilization of the Political Opposition in Taiwan: Comparing Two Waves of Challenges, 1979-1989." Taiwanese Political Science Review 1: 129-209. Wu, Nai-teh. 1995. "Party Identification or Partisanship? In Search of the Party Identification of the Taiwan Electorate." Paper presented at the second annual conference of the Taiwan Political Science Association, Taipei: December 23-24, 1995. Wu, Nai-the. 1996. "Liberalism, Ethnic Identity and Taiwanese Nationalism. " Taiwanese Political Science Review 1:5-39. Wu, Rwei-ren. 1994. "Three Motherlands: The Formation and the Competition of National Identities in Taiwan’s Early Post-War Period." Paper presented at the second annual conference of the Taiwan Political Science Association, Taipei: December 17-18, 1994. You, Ying-lung. 1993. "Issue-Oriented Voters in Taiwan: An Analysis of the Voters of the Second National Assembly Election." Paper presented at the conference on Electoral Behavior, Constitutional Order and Political Change. Held by Workshop 306 at National Taiwan University, Taipei, December 17, 1993. You, Ying-lung. 1994a. "A Political Psychological Analysis of Taiwan Electorate’s National Identities." Paper presented at the conference on Democratization, Party 211 Politics, and Elections. Held by Workshop 306 at National Taiwan University, Taipei, July 8, 1994. You, Ying-lung. 1994b. "Party Image, Ideology, and Secular Realignment in Taiwan." Bulletin of the Institute of Ethnology, Academia Sinica 78: 61-99. You, Ying-lung. 1996a. "Ethnic Identity and Political Cognition: An Analysis of Taiwanese Voters. Taiwanese Political Science Review 1:41-84 You, Ying-lung. 1996b. Popular Opinion and Taiwan ’s Political Change: An Analysis of Taiwan’s Popular Opinion and Electoral Politics in the 1990s. Taipei: Yuen Tan Press. 212