Previous studies on voting participation used to model individuals’ utilities and duties as the forces driving their participation. It is assumed that people maximize their utilities when they turn out. In the 2004 referendum, there were many controversies as to how to exercise direct democracy as well as fierce partisan battles. It is thus necessary to analyze the determinants of voting in the referendum from the perspective of social psychology. There are few empirical studies on the 2004 referendum. Because data on party identification, a person’s position on the unification/independence issue, and democratic values were collected from among the general public, we used a path analysis model to isolate the factors influencing participation in the referendum. Our preliminary findings were that the evaluation of the referendum was influential, and that the democratic beliefs were not as effective as expected. Neither were the political efficacy or political trust as effective as expected. By contrast, party identification and a person’s position on the unification/independence issue were found to have a great impact on turnout in the referendum. The empirical evidence indicates that the referendum was marred by political controversies, and thus the ideal of direct democracy was not achieved.