Empirical support for the PEFiC-model was found using structured questionnaires with various character types and focusing on different aspects of the model. The characters against which the PEFiC-model was tested ranged from animated Web agents, such as Bonzi Buddy (Hoorn, 2003), to Hollywood feature film characters (Konijn & Hoorn, 2004), to pictorial representations of world leaders (Konijn & Hoorn, 2003). Underlying this range was a continuum of realism in depiction. Bonzi Buddy was the most unrealistic, cartoon-like, representation of a living creature, followed by fantasy figures played by real actors, such as Dracula and Superman, in the film studies. The film studies also investigated more realistic characters (e.g., Gandhi), while the study of world leaders explored the most realistic representations, that is, newspaper pictures of, for example, president Bush and Bin Laden.
In all cases, the theoretically assumed involvement-distance trade-off could be identified and explained appreciation of the character in the predicted way. Moreover, for both the film character and the world leader studies, a successful fit between model and empirical data was established at questionnaire item level - the severest test possible. The results also indicated that none of the factors discerned by the PEFiC-model were redundant. Viewers indeed engaged with the characters in terms of ethics (good-bad), aesthetics (beautiful-ugly), epistemics (realistic-unrealistic), similarity (similar-dissimilar), relevance (relevant-irrelevant), valence (positive-negative), involvement, distance, and appreciation. The Bonzi Buddy study showed that users also took the factor norm (subjective-group) into account.
As predicted, engaging with fictional/virtual characters and mediated persons occurred in a parallel way, so that characters were experienced simultaneously as positive and negative. Because it turned out that questionnaire items could load on several factors, it seems that not only complete characters but fractions of a character's single feature can be appraised as positive and other fractions as negative. For the development of embodied agents this means that features should be studied for their contribution to an optimal involvement-distance trade-off, which determines the level of appreciation of the agent.
Hoorn, J. F. (2003). The Role of Social Norm in User-engagement and Appreciation of the Web Interface Agent Bonzi Buddy (Tech. Rep.). Amsterdam: Free University, Information Management and Software Engineering. Available at: www.cs.vu.nl/~jfhoorn/Text%20Design/BonziReport.pdf
Konijn, E. A., & Hoorn, J. F. (May 26, 2003). Pushing the Ethic, Aesthetic, and Epistemic Borders in Meeting Mediated People. International Communication Association, Theme Sessions: Local and Everyday Praxis in Virtual Borderlands, San Diego, California, May 23-27, 2003.
Konijn, E. A., & Hoorn, J. F. (2004). Some like it bad. Testing a model for perceiving and experiencing fictional characters. Media Psychology. In press.