Parasocial Interaction With Liked, Neutral, and Disliked Characters on a Popular TV Series

Abstract

In this study, 174 respondents completed an online questionnaire measuring their responses to a liked, neutral, or disliked character from the ABC drama Lost. Specifically, they reported their perceived similarity, identification while viewing, and parasocial interaction with the character, as well as the extent to which they had tried to change aspects of themselves to be more like the character (“change/influence”). Across the whole sample, perceived similarity was a significant positive predictor of both identification and parasocial interaction, and identification was associated with higher levels of parasocial interaction. Parasocial interaction, but not identification, was a significant positive predictor of reported change/influence. When the three types of characters were examined separately, all four responses were higher for liked and neutral characters than for disliked characters, and parasocial interaction was higher for liked than for neutral characters. Interpretations of the findings, and implications for understanding viewers' involvement with media characters, are discussed.

Notes

¹Several fit indices were utilized to test the model fit, including the ratio of chi-square to degrees of freedom, the comparative fit index (CFI), the root mean squared error of approximation (RMSEA), and the standardized root mean square residual (SRMR). Generally speaking, the recommended fit standard for the ratio of chi-square to degrees of freedom is at or less than 3.0, and CFI is greater than .90. The recommended level of RMSEA is at or less than .08 with its 90% confidence interval excluding .10. Values of SRMR less than .10 are generally considered favorable (Kline, 2005). The goodness of fit statistics of the confirmatory factor analysis suggested that the original measurement model did not fit the data very well, ; RMSEA = .10 with the 90% confidence interval from .09 to .11; CFI = .97; SRMR = .07. Item loadings ranged from .50 to .88. The modification indices suggested adding paths from two identification items (“While viewing Lost, I want the character to succeed in achieving his/her goals”; “When he/she succeeds I feel joy, but when he/she fails, I am sad.”) to parasocial interaction. These two items do seem to overlap conceptually with parasocial interaction, and were removed from the identification scale. The new model fit the data better. All the goodness of fit indices were within acceptable range, ; RMSEA = .08 with the 90% confidence interval from .07 to .09; CFI = .97; SRMR = .06.

Note. The four responses to the characters (perceived similarity, identification, parasocial interaction, and change/influence) could range from 1 to 5. Education could range from 1, no highschool, to 9, doctoral degree.

*p < .05. **p < .01. ***p < .001.

Note. Respondent gender was coded 0 for men and 1 for women. Character type was coded 1 for disliked characters, 2 for neutral characters and 3 for liked characters. Betas in the table are betas at entry.

*p < .05. **p < .01. ***p < .001.

Note. Scores on each scale could range from 1 to 5. All F values (2,157) for univariate ANOVAs were significant at p < .001. For each variable, means for the three characters types that have different subscripts differ at p < .05 by the Tukey's post hoc test.

²We tested this using the current sample. The data set was split by character type, and demographic information was entered in the first step of the hierarchical regression equation, perceived similarity in the second step, identification in the third step, and parasocial interaction in the fourth step. The pattern of results was the same in the three analyses. Perceived similarity was a significant predictor for all three types of characters (liked: β = .60, p < .001; neutral: β = .66, p < .001; disliked: β = .69, p < .001), but neither identification nor parasocial interaction reached significance in any of the analyses.

Additional information

Notes on contributors

Qing Tian

Qing Tian (Ph.D., Georgia State University, 2009) is a recent graduate of the Department of Communication, Georgia State University. Her research interests include mediated relationships, computer-mediated communication, and media uses and effects.

Cynthia A. Hoffner

Cynthia A. Hoffner (Ph.D., University of Wisconsin-Madison, 1988) is Professor in the Department of Communication, Georgia State University. Her research focuses on the role of affect and cognition in how people process and respond to media messages.