Perceptions of Perfection: The Influence of Social Media on Interpersonal Evaluations

ABSTRACT

Through social network sites such as Facebook, people gain information about acquaintances that they would not gain from everyday life. This information typically highlights the most positive aspects of people’s personalities and lives. The goal of this investigation was to determine whether looking at another user’s Facebook profile influences perceptions of that individual’s socially desirable characteristics (e.g., intelligence, attractiveness). One group of participants viewed an acquaintance’s Facebook profile before providing evaluations, and the other evaluated the person without viewing Facebook. Results revealed that participants who viewed another person’s Facebook profile evaluated that person more favorably than those who completed a control task (Study 1) or wrote about the person from memory (Study 2). Theoretical and practical implications are discussed.

Notes

Participants in both conditions completed the Remote Associates Test before choosing friends to evaluate as part of a larger study on social comparison and social media. All participants were told that the test was a predictor of future success. They also evaluated themselves on the same characteristics (attractiveness, intelligence, likability, popularity, and success) as their friends. These manipulations and measures are not central to the core components of the present article and are not discussed further.

One evaluation score (1.80) was more than 3 standard deviations below the mean. When the outlier was excluded from analyses, the pattern of results remained consistent, with a slightly reduced effect size (d = .39). Because the outlier is a realistic and meaningful data point, it was retained in the final analysis.

It is possible that some participants evaluated targets whom they met on social media and have never met in real life. In Study 1, participants were asked, “When was the last time you saw this person?” Of the 605 targets that participants evaluated in this study, only three were people they had not met in person (0.05%). In Study 2, participants were asked, “How did you meet this person?” Only two participants indicated that they met the person online (1%). Because the vast majority of evaluation targets were individuals whom participants knew offline, this issue most likely did not influence the observed pattern of results. Analyses were conducted excluding online-only targets for both studies, and the pattern and strength of the results did not change.

It is important to note that the anchors of the measurement scales differed across Studies 1 and 2. Specifically, Study 1 used comparative judgments (below average to above average), and Study 2 used absolute judgments (not at all to very). Although using consistent anchors would have been ideal, there is a large literature showing that absolute and comparative judgments are highly conflated (see Chambers & Windschitl, 2004, for a review). The extant literature and the consistent effect sizes across the two studies suggest that the change in scale anchors and judgment types did not have an impact on results.

Although Studies 1 and 2 assessed the same five traits, Cronbach’s alpha was notably lower in Study 2. This may have been caused by the relatively small number of items (five) included in Study 2 (Peterson, 1994). A 2 (experimental condition) × 5 (trait) mixed-model analysis of variance showed a very small Condition × Trait interaction (partial η² = .008), indicating that the influence of trait type did not substantially differ based on experimental condition. Furthermore, a principal components analysis did not support extraction of multiple factors. Therefore, we opted to combine the traits into one evaluation score.