The Presence of the Protagonist: Explaining Narrative Perspective Effects Through Social Presence

ABSTRACT

Based on two experiments, this paper advances the concept of social presence as a novel mechanism through which narrative perspective (first- versus third-person) exerts persuasive effects on attitudes toward outgroup policies and behavioral intentions to help outgroup members. Study 1 (N = 503) shows that the first-person perspective, compared to the third-person perspective, increases social presence of the protagonist, but not identification with the protagonist, when the story depicted an outgroup character. This increase in social presence mediates the effect of narrative perspective on support for outgroup policies. Study 2 (N = 410) further suggests that social presence mediates the effect of narrative perspective regardless of the protagonist’s group membership (in-group versus out-group). Furthermore, this project evaluates the role of social presence in light of other, often-studied processes such as identification and transportation. These findings advance the theorizing and research in narrative research and in media psychology more broadly.

Acknowledgments

This work was partially supported by the Singapore Ministry of Education Academic Research Fund Tier 1 Grant.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Notes

1. Sampling and recruitment were done by Qualtrics, which maintains a large online panel of adults recruited through affiliates, advertisements, and partner samples. Panel members received reward points for completing the survey – points which could later be exchanged for various rewards such as PayPal currency and e-vouchers. Survey was advertised via email to panel members. Only Singaporean adults could participate in the study.

2. The study was originally designed as a 2 (narrative perspective: first vs. third) x 2 (target salience: photo vs. no photo) between-subject design with participants randomly assigned to one of four conditions. However, the manipulation of target salience had no effects on any of the variables and also the covariance between study variables did not change by target salience. We thus exclude this factor from the current analyses. Post-hoc power calculations show .61 power for Study 1.

3. Both were rated equally in terms of story believability (M_1st = 5.12, M_3rd = 5.09, p = .838), accuracy (M_1st = 4.86, M_3rd = 4.82, p = .671), completeness (M_1st = 4.67, M_3rd = 4.58, p = .480) and bias (M_1st = 2.41, M_3rd = 2.53, p = .339).

4. We commissioned Survey Sampling International (SSI) to recruit participants from their online panel. SSI also recruits and maintains panel members through online advertisements and referrals. The survey was posted on their main board and reward points were given as an incentive. Post-hoc power calculations show .52 power for Study 2.

5. Depicting the ingroup protagonist as one from a low-income family may possibly contaminate the manipulation since it may inhibit identification for higher income participants. However, controlling for income did not influence the magnitude, direction, or significance of the presented results. Moreover, regardless of the condition, income was not significantly correlated with the level of identification with the character (ingroup condition r = .09, p = .197; outgroup condition r = .09, p = .203).

6. The ingroup/outgroup narratives not only differ in the protagonist’s group status, but also in content as they reflect unique life experiences of a local student vs a migrant worker. This was inevitable as an ingroup member’s experiences are clearly distinct from those of the outgroup’s, at least in the current study’s context. Fortunately, our pretest confirms that the two stories were rated equally on story accuracy (M_ingroup = 4.83, M_outgroup = 4.98, p = .422), completeness (M_ingroup = 4.51, M_outgroup = 4.43, p = .713) and bias (M_ingroup = 4.26, M_outgroup = 4.10, p = .455). The two stories were also rated equally in terms of character evaluation (honest, M_ingroup = 5.30, M_outgroup = 5.56, p = .141; trustworthy, M_ingroup = 5.29, M_outgroup = 5.42, p = .432; honorable, M_ingroup = 5.53, M_outgroup = 5.31, p = .214; genuine M_ingroup = 5.52, M_outgroup = 5.72, p = .244).

7. High correlations among mediators not only lead to unstable regression estimates and inflated standard errors, it also introduces complexities in interpretation (Cohen, Cohen, West, & Aiken, 2003, p. 98–99, 420; Hayes, 2018).

8. We rely on the joint criteria proposed by Hu and Bentler (1999) to assess model fit: CFI≥0.96 and SRMR≤0.09 or SRMR≤0.09 and RMSEA = <0.06.

9. We used item parcels instead of the raw items to represent the identification construct as the original model with 10 raw items showed less than satisfactory model fit (χ²(51) = 282.30, p < .001, CFI = .946, TLI = .931, RMSEA = .105, 90% CI [.093, .117]; SRMR = .051). Item parceling has been used successfully in a number of studies that utilized SEM (e.g., Peter & Valkenburg, 2008) to help reduce idiosyncratic feature of items in favor of more parsimonious models. We also dropped one item from the transportation scale due to redundancy issues as discovered in the residual analysis.