Rhetoric can cloud the truth in closed societies, but in a society freely open to researchers—both local and international—the facts are impossible to conceal.
The Republic of China on Taiwan offers an enviable environment for research into the actual functioning of democracy in a society long accustomed to quite different governmental forms. Enviable not only in comparison with the restricted society so vociferously protected across the Taiwan Straits, but also in contrast with other developing nations of the world.
The following article is presented both as an example of the level of professionalism now found among local researchers, who take great pains to survey international literature in their fields (in this case public voting behavior), and as an open invitation for scholars worldwide to undertake research in Taiwan.
The work of scholarship in Taiwan is an international enterprise, and as more foreign scholars from all fields including the social sciences work here, their efforts will provide even better understanding of the challenges and successes experienced by Chinese people during 40 years of economic growth and political tutelage in a creatively designed Republic.
Major Approaches in the Research of Voting Behavior
Political participation is essential in the formation and continuation of a democratic nation and elections may be the most important way to promote political participation, for it is through elections that people select political leaders and shape public policies while the government seeks legitimacy.
Voting, the climax of election activities, produces a collective decision based on individual choices. Election returns decide the personnel and administration as well as showing the degree of public support of any government's policies or actions. In democratic nations, voting behavior indicates the degree to which any government administration enjoys popular support, and it can have profound influence on a country's political development. It is therefore essential to understand the complexities of the election process, and to be able to interpret election statistics in useful ways.
Research on voting behavior has long been emphasized by American political scholars, and they have developed a fairly extensive foundation of research criteria and methodology. Generally speaking, voting behavior research falls into three areas: (1) forecasts of election returns, (2) factors that affect voting, and (3) characteristics of individuals or groups and their correlation with voting behavior. American research in these areas has produced results, theories, and methodologies that have had profound influence on similar research around the world.
Three key approaches to voting behavior have evolved in recent years. First, the "sociological approach" has emphasized the correlation between the voters' behavior and their societal background. The research focuses on investigating the influences on voting behavior by primary groups (such as family and kinship groups), secondary groups (such as political party affiliation and unions), and factors of social influence (including race, status, age, religion, social structure, occupational structure, and urbanization). This body of research work centers, therefore, on description and explanation of voting behavior.
Second, the "socio-psychological approach" has discovered that the major variables influencing the formation of people's voting orientations are party identification, candidate orientation, and issue orientation. Moreover, the approach has identified four other key variables: (1) a sense of political efficacy, (2) a sense of citizen duty, (3) a voter's perception of the party and its candidates, and (4) a voter's overall interest in the campaign.
The major contribution of this research approach is to shift the research angle from sociology to psychology, with emphasis on determining the impact of the voters' individual motives, characteristics, and political attitudes. These psychological variables in voting orientations have now been widely adopted in research on voting behavior.
Third, the "economic approach" has had considerable influence in assessing voting behavior. One of the most outstanding researchers in this field is Professor Anthony Downs of Stanford University. In his seminal 1957 book, An Economic Theory of Democracy, he first established a rational voter model that based the research of voting behavior on the concept of rational choice. Based upon a logical rationalization process with a core concept called "cost-profit," Downs set up a "rational voter model" to explain voting behavior.
This model can be used to investigate areas that the sociological and socio-psychological approaches failed to consider adequately. While the latter two can only explain or delineate voting patterns, Downs' model can be used to forecast voting behavior.
All three approaches have had extensive influence on later research, but in Taiwan the socio-psychological approach seems to have had the most profound overall influence thus far.
Taiwan's Research Experience in Voting Behavior
As of 1986, elections held to select regional representatives and local government officials have included the following:
(1) 10 elections for Taiwan Provincial Assemblymen (including two elections for temporary assemblies);
(2) 10 elections for county magistrates;
(3) 11 elections for city and county councilors;
(4) 10 elections for town and city mayors;
(5) 13 elections for representatives of villages, towns, and cities (including two earlier elections held in 1946 and 1948);
(6) 13 elections for village and borough chiefs (including two elections in 1946 and 1948);
(7) five elections for Taipei City Councilors since the city was upgraded to a municipality in 1967; and
(8) two elections for Kaohsiung City Councilors since its upgrading to a municipality in 1979.
Elections for National Representatives include the following:
(1) One by-election in 1969 to elect legislators and members to the National Assembly;
(2) five supplementary elections of lawmakers representing Taiwan district (since 1972); and
(3) three supplementary elections of members to the National Assembly since 1972.
Thus, there has been a total of 84 regional and national elections. Over the years, the elections have had the broad value of training an electorate generally unaware of democratic forms. Because the government has frequently held regional or national elections for administrators and representatives, people have become increasingly interested in political participation. Taiwan has consequently seen a general growth in the sense of political participation, a process revealed in part by the turnouts for elections. According to official statistics, the turnouts for various categories of elections are as follows (given in percentages):
(1) provincial assemblymen (72.44);
(2) county magistrates (73.15);
(3) city councilors and county assemblymen (76.96);
(4) town and city mayors (68.86);
(5) village and town representatives (66.56);
(6) village and borough chiefs (58.99);
(7) Taipei City Councilors (65.64);
(8) Kaohsiung City Councilors (76.22);
(9) By-elections in 1969 (52.45);
(10) Supplementary elections for Legislators from Taiwan district (61.84); and
(11) Supplementary elections for National Assemblymen from Taiwan district (63.48).
The average of the above turnouts was 66.96 percent. In other words, about two thirds of the eligible voters vote in any election. This figure is not low compared to democratic American or European countries, and it serves as a barometer of the prevalence of political participation.
Although many elections have been held in this country, research on voting behavior began rather slowly, with the first on-the-spot research conducted in 1964 by Professor Tsou Wen-hai. Since then, there have been 25 published research reports by school faculties on the voting behavior of various social groups. The actual number of research pieces is actually somewhat higher because some studies were not published. All 25 academic reports utilized either the sociological or socio-psychological approach.
Besides, there have been four other voting behavior studies using the ecological approach, and one M.A. thesis that used the economic model. Another seven studies, six of them M.A. theses, adopted modern methods of content analysis. Finally, one study conducted in 1985 used the economic approach model in the analysis.
Taiwan's development of high standards in the research of voting behavior can be divided into three periods; because local researchers have more or less systematically adopted the methodologies used by American experts in the field, the periods are arranged below not by theory or research approaches but by the actual date of research publication.
— The Burgeoning Period (roughly the 1960s). Two trailblazing works were published at this time: "Research of Local Elections in the Taiwan Area," written in 1964 by Professor Tsou Wen-hai of National Chengchi University, and "Research of Political Behavior of Taipei Voters—A Case Study of the First Election of Taipei City Councilors in November 1969," by Professor Tsai Chi-ching of Tunghai University.
Tsou's research was based on official statistics and analyzed the turnouts of various elections in Taiwan. He also conducted a survey of 3,890 students drawn from a non-random sample of those attending National Taiwan University, National Chengchi University, and Tamkang University. Tsai's research was based on non-random sampling of residents in Taipei's five boroughs.
Both researchers used descriptive statistics to explain the characteristics of their data, and adopted the sociological approach to analyze the characteristics of groups (such as age, sex, and employment) with reference to different voting behavior. Because the research in this field was just beginning, their samplings, questionnaire designs, and analytic methods were fairly rough. It would be difficult to argue that the studies met the requirements of strict scientific methodology. Nevertheless, the two professors marked the beginning of the field in Taiwan, and contributed greatly to subsequent work in the area.
— The Pivotal Period (the 1970s).
During this decade, nine research reports on voting behavior were published, not including M.A. theses. Six of the nine were done between 1977 and 1980. Their general characteristics can be assessed as follows:
(1) Most of the works are based on a sample with fewer than 500 people; only three contain samples between 1,000 to 8,000 people to achieve a large-scale, nation-wide research base.
(2) One research work uses purposive sampling, while the other eight use multi-stage stratified sampling.
(3) Concerning the method of analysis, all use inferential statistics, most use chi square (X2), and two use ANOVA (analysis of variance) and product-moment correlation. One also employs factor analysis and multiple regression, a more advanced statistical approach.
(4) Most use the sociological approach or socio-psychological approach.
— The Development Period (after 1981). After 1981, Taiwan's research on voting behavior began advancing rapidly. Between 1981 and 1987, there have been 15 research studies conducted on a national base. The samplings, questionnaire designs, analyses, and approaches all follow a much stricter process and reach higher standards than earlier studies. Their key characteristics can be summarized as follows:
(1) Almost all studies sample more than 1,000 people.
(2) Most use refined multi-stage stratified sampling or random sampling techniques.
(3) In the analysis stage, almost all use inferential statistics, including factor analysis, ANOVA, path analysis, discriminant analysis, MCA, and log-linear models.
(4) Most researchers use the socio-psychological approach, but Yuan Sung-hsi and his co-workers tried to combine micro- and macro-approaches. Their work starts from studying the political participation experience of Taiwan's eligible voters and goes directly into examining a standard socio-economic mobilization pattern. The goal is to infer the feasibility of applying the social mobilization theory in Taiwan.
Several interesting generalizations can be made about the findings gained from the voter behavior research conducted in Taiwan over the last 23 years.
(1) Voters generally do not have a high level of political knowledge, including limited understanding of the election laws or the actual powers of legislators.
(2) Campaign activities have great influence on votes. About 70 percent of the voters said they decided about their choices during campaign periods or even after the campaign period.
(3) Voters' orientation preferences break down as follows: candidate orientation (55 percent), issue orientation (30 percent), party identification (15 per cent)—indicating that traditional orientations have gradually disappeared, although they still play an important role in regional elections or elections for members of the National Assembly.
In 1987, Hsia Chen of the Journalism Graduate School of National Chengchi University published "Research on Campaign Giveaways: Topics and Strategies in Regional Election for Representatives and Administrators—A Case Study of the Election in 1985." He collected 330 giveaways from 48 candidates in 20 counties of the island as the data.
Major research conclusions from his study include the following:
(1) The higher publicity in the press, the better chances of being elected.
(2) Private-owned newspapers emphasize reports of campaign activities; state-run newspapers emphasize policy issues.
(3) The more active the candidates are, the more inclined they are to use giveaways. Candidates who believe their speeches are distorted by the press are also inclined to use giveaways.
(4) Giveaways are used to build up candidates' images rather than to advocate a policy.
(5) Campaign topics tend to center on political parties or political values and issues.
(6) In the election for supplementary legislators in 1983, there was no clear focus of political contention.
Future Research of Voting Behavior
Seven general trends emerge from a review of Taiwan's development in the research of voting behavior. These provide convincing indications that Taiwan's future development in this field will continue with even more fruitful results.
First, there has been steady development toward greater sophistication in the techniques of research analysis. Research work has moved from essentially descriptive analysis, to explanatory work, to forecast studies. From 1983 to 1985, five articles have been published which have provided very accurate forecasts.
Second, there has been a trend away from a general lack of sufficient theory, theory application, and theory building. Research in the 1960s, for example, was frequently devoted to one question and one variable. Moreover, the research itself suffered from insufficient basing on theory. By the 1970s, however, researchers began using the theory frame works from sociology or social-psychology and set up individual research paradigms to explain voting behavior. But it was only by the end of the 1970s and early 1980s that indigenous theories evolved to the point of building models for perceptive research into local political participation.
Third, there has been a trend toward using a variety of approaches to generate solid research results. In the 1980s, scholars have been able to successfully apply multiple approaches simultaneously in their research activities. They study election phenomena from multiple angles, trying to depict a complete picture of voting behavior. These approaches include the sociological, socio-psychological, ecological, and economic, plus a combination of several of these methods.
Scholars in the 1970s have used various methods to study voting behavior, including personal interviews, telephone interviews, content analysis, field analysis, and panel study (involving continuous surveys on the same sample to understand changes during the campaign period).
Fourth, there has been great use of advanced statistical analysis. The use of statistics developed from descriptive statistics of percentages, to chi square, ANOV A, product-moment correlation, to factor analysis, multiple regression statistics, path analysis, discriminant analysis, and finally to the recent use of the log-linear model and multiple classification analysis.
Fifth, Taiwan's researchers have attained greater sampling precision. In the beginning stages, researchers used nonrandom sampling or purposive sampling. By the 1970s, random sampling (multiple-stage stratified sampling or systematic sampling) were widely used. And in 1985, Ting Tin-yu included ecological variables in sampling, which made the sample much more representative and the results more accurate.
Sixth, interschool cooperation and greater use of technology continue to expand. Beginning in 1983, various professors from different universities started an interschool cooperative research project in this field, and in 1985 and 1986, two other large research projects attracted professors from different fields. Joint research projects now involve experts in politics, sociology, archaeology, psychology, mass communication, and statistics.
Seventh, more research units and agencies now participate in this sort of research field in addition to university faculties. In 1983, National Chengchi University established an Election Research Unit, headed by Professor Lei Fei-lung, to actively push the research projects. Moreover, today many newspapers like the United Daily News and the China Times have developed the research field for their own reporting of elections. These organizations have been joined by parties, candidates, and the Public Opinion Survey Association—all obviously having great interest in voter behavior. The ranks of all these organizations have been swelled by the arrival of new researchers on the scene holding appropriate degrees from local or foreign universities.
These new degree holders, along with their more senior colleagues, are helping transform the field of voter behavior research and related fields. The newcomers have brought with them refined ideas, theories, methodologies, and analytic techniques that should have considerable impact on local research. As a result, the outlook for qualitatively superior research on voting behavior in Taiwan is clearly optimistic. - (Dr. Yih-yan Chen is a professor of political science at National Chengchi University, Taipei).