Abstract
The origins of third-person perceptions remain uncertain, with most research focusing on psychological mechanisms. We investigate whether media content might also play a role, using a 2 × 2 experiment presenting a single story describing video games as harmful or harmless and using either research and statistics or a specific anecdotal exemplar to illustrate this claim. Results show perceived effects on others are influenced by the use of an exemplar but do not show an effect for explicitly describing games as harmful. The findings suggest that media may influence third-person perceptions and subsequent support for censorship in previously unexplored ways.
Notes
1We included a measure asking participants whether they agreed (on a 7-point scale) that “the article argued that games aren't harmful.” Confirming the effectiveness of that manipulation, those reading the harmful story (M = 2.25) were far more likely to disagree than those reading the harmless story (M = 5.52, t = 16.19, p < .001). Rather than sensitize readers to exemplars, we evaluated whether participants (on a 7-point scale) “found the article interesting to read.” Consistent with research suggesting exemplar-driven stories are more engaging, such a story was rated (M = 4.32) significantly more interesting than the research-based story (M = 3.80, t = 2.61, p < .05).
Note. Numbers displayed in the table are the means of perceived negative effects of video games. In each row, values not sharing a capitalized subscript are significantly different based on paired t tests (p < .05). In each column, values not sharing a lowercase subscript are significantly different based on an independent samples t test (p < .05).
Note. The numbers in Table 3 show the F value for each variable. df = 1, 241 (for self); 1, 237 (for adults, adults-self); 1, 236 (for college students, students-self).
**p < .01. ***p < .001. †p < .1.
Note. Values given are raw means for each condition by level of play, with standard errors in parentheses. The numbers in the same column with the different subscript(s) are significantly different from each other, whereas the numbers in the same column with the same subscript(s) are not significantly different from each other.
Note. The numbers in the table show the unstandardized coefficients (and standard errors) for each variable. Model 1 includes experimental manipulations, level of game play, and additive and subtractive third-person measures. Model 2 includes these variables and interaction terms. N = 241.
*p < .05. **p < .01. †p < .1.