집단극화와 컴퓨터 매개 커뮤니케이션에 대한 재고: 익명성 효과

The objective of this paper is to contemplate a possible alternative to the examination, published as a research article in Information Systems Research (Sia et al. 2002), of the effects of communication cues, anonymity, and social presence on group polarization in computer-mediated communication (CMC) settings. Sia et al. introduced the concept of social presence to explain the effects of communication cues and anonymity. Based on earlier studies on CMC and group polarization, Sia et al. argued that (1) the different types of communication cues yield different levels of social presence; (2) anonymity lowers social presence; and (3) the level of social presence is inversely associated with the extent of group polarization. Thus, they emphasized the mediating role of social presence on the impacts of communication cues and anonymity. However, the mediating role of social presence does not sufficiently explain the results of the two experiments. We believe that the complete identity condition in the distributed CMC setting was not fully accomplished in the experiment, and therefore the results would be difficult to interpret. In the face of this insufficiency, we believe the direct impact of anonymity on group polarization can provide a more admissible and clearer explanation for their results. This paper classifies levels of anonymity to provide possible explanations for Sia et al.’s results and presents the results of a new experiment to confirm our explanation. To justify our anonymity view, we conducted an experiment similar to Sia et al.’s, guaranteeing a complete identity in the dispersed CMC setting. If under conditions of complete identification the removal of visual cues does not lead to stronger group polarization, it is the provision of anonymity and not the absence of visual cues that produces a significant impact on group polarization. Our modified experiment compares an identified face-to-face CMC setting and an identified dispersed CMC setting. Each of the two settings has 10 groups of five subjects each. Subjects are undergraduates from a large university. All experimental procedures and the calculations of choice shift and preference change follow Sia et al.’s study. Unlike Sia et al.’s result, the modified experiment shows no difference in group polarization between the identified dispersed CMC setting and the identified face-to-face CMC setting. Thus, in our experiment the removal of visual cues does not produce a significant impact on group polarization, which cannot be explained by the social presence view. Our experiment may suggest that the provision of anonymity rather than the removal of visual cues affects group polarization, supporting the anonymity view as a more appropriate explanation of Sia et al.’s results. Additional research should be done soon to confirm the direct impact of anonymity on group polarization. In particular, modified experiments for Sia et al.’s second experiment should be conducted. In addition, Sia et al. did not measure social presence directly nor analyze the mediating role of social presence. Therefore, future studies should analyze, by measuring social presence, whether communication cues and anonymity affect group polarization through social presence. The results of such studies would explain more about the leading factor affecting group polarization. Moreover, partial identity situations are uncommon in organizations’ group decisions because real names are used and persons are easily recognized. Consequently, more practical results will be obtained in anonymity-related studies if situations of complete identity or total anonymity are the subject. Finally it should be noted that many other possible factors such as group size, topics, and group history affect group polarization. These factors should be examined in future studies. Although we have disagreed with the above portion of Sia et al.’s study, we want to also commend it for a rigorous experimental methodology and its exposition of the social presence view. The purpose of our study is not to criticize the Sia et al.’s study, but solely to present the anonymity view as an alternative to the social presence view in the study of group polarization in CMC. We also want to emphasize that our comments are not intended to disparage the usefulness of social presence as a mediating variable to explain group polarization; our intention instead is to encourage further investigation to better understand group polarization in CMC settings.