Reviews, expectations, and the experience of stories

ABSTRACT

Theory suggests that information encountered prior to a story affects the actual experience of the story due to elicited expectations. In two experiments (N = 100; N = 167) short movies that were introduced with a positive review yielded higher transportation scores than the same movies introduced with a more negative (or neutral) review. Mediation analyses indicate that the reviews had an influence on recipients’ expectations, which in turn predicted the experience of the movie. Using the more fine-grained narrative engagement scale, we found evidence for a consistent effect on narrative presence, whereas the influence on emotional engagement, narrative understanding, and attentional focus varied between experiments. Moderation analyses (moderated mediation) showed that recipient’s opinion seeking and need for cognitive closure were unrelated to the influence of reviews on expectations and the link between expectations and narrative experience. Our findings add to the theory of story processing and they are of practical relevance for everyone who intends to influence recipients’ experience of narrative worlds.

Acknowledgments

We wish to thank Patrick Bacherle for his theoretical input and advice.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Please refer to Gunter (2018) for findings on moderating factors, such as film genre or critic popularity.

2. Although Shedlosky-Shoemaker et al. (2011) measured the participants’ expectations prior to the film and used the variable for a manipulation check, they did not analyze whether the differences in enjoyment and transportation were indeed due to different expectations or whether other variables drive the effect.

3. Throughout the manuscript higher expectations indicate more positive expectations.

4. More specifically, participants with the three highest need for affect scores were assigned to three different groups (with the highest score assigned to group 1, the second-highest score to group 2, and the third-highest score to group 3). Then the next three highest scores were assigned to the three groups (in an opposite order; the highest score to group 3, the second-highest score to group 2, and the third-highest score to group 1), and so on (with the two orders alternating). In doing so, we ensured that need for affect is equal across all groups. Subsequently, the three groups were randomly assigned to the three conditions.

5. Additional items that were asked, but not analyzed further, addressed participants’ mood and sleepiness as well as their enjoyment and evaluation of the film.

6. We conducted t-tests with Welch correction whenever a Levene’s test indicated inhomogeneity of variances. Welch-corrected t-Tests are indicated with a subscript W (t_W).

7. We also considered following the safeguard power analysis (Perugini, Gallucci, & Costantini, 2014) as an alternative way to determine the sample size for Experiment 2. This approach suggests to use the lower boundary (60%, two-tailed confidence interval) of the original study’s effect size to determine the sample size of the replication study. In our case, a power analysis using the lower boundary of the original effect size (i.e., d = .91 with, 60% CI [0.68, 1.11]) identified an a priori sample size of 96 participants. Because one aim of Experiment 2 was to replicate the original findings with a sufficiently large sample size, we followed the guidelines by Simonsohn (2015) instead.

8. Additionally, participants were asked for their enjoyment. However, we did not further analyze enjoyment.