The Impact of Media Narratives on the Audience’s Self: Immediate and Delayed Effects

Abstract

Studies commonly measure the effects of media narratives on the self immediately following exposure. We explored whether these effects on viewers’ self-esteem, self-efficacy and motivation are confined to the immediate post-exposure period and examined how they are related to the audience’s identification with the protagonist. In Study 1, participants watched movie clips that included inspiring narratives depicting protagonists who succeeded in accomplishing their goals. Implicit self-esteem was significantly higher immediately after watching than after a short break. Unexpectedly, identification was positively correlated only with delayed self-esteem. In Study 2 we changed the control condition and added explicit measures. The reduction in implicit self-esteem between the two measurement times was replicated in the no-treatment control group. Moreover, self-efficacy and motivation were higher than in the control group and were stable over time, and identification was positively related only to delayed implicit self-esteem and motivation. We conclude that the effect of inspiring narratives on the self extends beyond the immediate time of exposure, and identification with the characters contributes to this persistent effect. The findings are explained by theories of contrast and assimilation, identification and inspiring narratives.

Past research demonstrated that people tend to assimilate themselves with the characteristics of media characters they identify with. Commonly, this assimilation is demonstrated immediately following the exposure to movies or TV series (e.g., Sestir & Green, 2010). The current research explored whether the effects of media engagement on the self extend beyond the time of exposure, when the focus of the viewer’s attention shifts from the movie world to their own real world. Moreover, we aimed to examine the role of engagement with media characters on the effect on the self, both immediately following the viewing and after a delay.

1. Engagement with Narratives and Their Characters

One growing area of research in media psychology deals with the psychological processes involved in consuming narrative texts, whether in the form of newspaper articles (e.g., Thier et al., 2021), books (Wimmer et al., 2021), social media (e.g., Rieger & Klimmt, 2019), commercials (e.g., Thomas & Grigsby, 2024), TV programs (Tchernev et al., 2023) or films (e.g., Igartua, 2010). When people are fully engaged in a story to the degree that they lose touch with their own real world, they are said to be transported into the narrative (Green & Brock, 2000). While transportation refers to general engagement with the narrative, identification refers to engagement with the narrative’s characters. When media consumers identify with media characters, they feel as if they are merged with them (Cohen, 2001). The interest in these media engagement processes, and in identification in particular, stems mainly from their proven impact on the acceptance of attitudes that accord with the narrative and/or are conveyed by its protagonists (e.g., Tal-Or & Cohen, 2016). While most research on the effect of these processes has been conducted in the context of narrative persuasion, some studies have demonstrated their effect on the self (e.g., Johnson et al., 2015), including effects on empathy level and personality (e.g., Mar et al., 2008).

2. The Effect of Engagement with Narratives on the Self-Concept

Most studies about the effect of the narrative on the self assume that exposure to narratives makes the media consumer’s self-concept similar to that of the protagonist, through a process of assimilation (Richter et al., 2014), especially when they identify with the media character (Green & Appel, 2024). Accordingly, a number of studies have demonstrated that following exposure to a narrative text in which the media consumers engage with the narrative’s protagonists, the media consumers perceive themselves as having personality traits resembling those of the characters. In one of these studies (Sestir & Green, 2010), individuals who identified with a protagonist ascribed traits to themselves that were characteristic of the protagonist more quickly than those who identified less with the protagonist. Both those who identified with the media characters and those who were transported into the narrative were more likely to change their reported traits to those of the character. Similarly, Richter et al. (2014) noted that when the story included stereotypical information about women, the participants who were transported into it perceived themselves as more feminine. While these studies focused on assimilation with a particular character, Gabriel and Young (2011) argued that engagement with a narrative might lead the media recipients to feel as if they belong to the collective depicted in it. They showed that reading a story about wizards or vampires led their participants to adopt the respective collective’s characteristics.

3. The Underlying Mechanisms of the Effect of Engagement with Narratives on the Self

Researchers have long acknowledged that exposure to a reference object can result in perceiving other objects either as more similar to the anchor—an assimilation effect—or as more dissimilar to that object—a contrast effect. Various explanations have been offered for the occurrence of these effects. One refers to the mindset of the individual when making the comparison. A mindset that focuses on similarities rather than differences between the objects leads to assimilation (e.g., Bless & Burger, 2016). Another explanation refers to the cognitive effort exerted when making the comparison. Unlike in assimilation, for the contrast effect to occur, a person has to exert effort in processing (Martin et al., 1990).

These explanations might account for the differing media effects on the self. Exposure to media often leads to contrast effects, as is well documented in studies of the detrimental effect of the presentation of the thin-ideal body image, especially in ads, on women’s body dissatisfaction (e.g., Grabe et al., 2008). While exposure to media in general can lead to contrast effects, engagement with media narratives usually leads to assimilation (e.g., Richter et al., 2014; Sestir & Green, 2010). Both the mindset of seeking similarities with the protagonist and the lack of cognitive effort exerted may explain this outcome. Because lay theories regard similarities between people as crucial to a satisfying relationship, people are motivated to find similarities between themselves and people they like (Morry, 2005), so they might be in a mindset of finding similarities between themselves and favorable protagonists. This is especially true when identifying with these characters, a process in which the media recipient merges with the media character (Cohen, 2001). Similarly, when engaged in the narrative, one’s attention is directed to the narrative world, experiencing it with the characters (e.g., Slater & Rouner, 2002). Thus, there is less chance of exerting cognitive effort in contrasting oneself with the characters, so assimilation is more likely to occur.

But what happens in the time period following exposure to the narrative? Does the assimilation effect last? Do media recipients return to perceiving themselves as they did before being exposed to the narrative? Or, perhaps, is the assimilation effect followed by a contrast effect?

4. Contrast Following Assimilation Effects

Although a possible change from assimilation to contrast has not been studied yet with regard to the response to media characters, there is some evidence in the psychological literature that an initial assimilation effect turns into a contrast effect, especially when people are directed to think about their own lives. In a study by Dermer et al. (1979, see Study 2), the participants were asked to imagine very good or very bad scenarios happening to them. For example, some were instructed to imagine that they were severely injured in a car accident, and some were asked to imagine that they won a luxurious vacation. Initially, an examination of their feelings revealed that those who imagined good scenarios felt better than those who imagined negative ones. While this can be interpreted as spreading activation of the positive emotions (e.g., Berkowitz, 1990), it can also be seen as an assimilation effect. Later on, the participants were asked to report on their satisfaction with various aspects of their own real lives such as health and physical appearance, and here, a contrast effect was obtained such that those who imagined negative scenarios were more satisfied with their lives than those who imagined positive scenarios (Dermer et al., 1979).

McMullen (1997) explains that people can process mental simulation in an experiential or evaluative mode. In the experiential mode, they focus on the fantasy world as if it were real, resulting in an assimilation effect. In the evaluative mode, on the other hand, they compare the fantasy world to their own world, and thus, a contrast effect is obtained. Importantly, according to McMullen (1997), the underlying mechanism for these two modes is the focus of attention. While in the experiential mode, the focus is only on the fantasy world, in the evaluative mode, the focus is on both the fantasy and real worlds. This shift from the fantasy world to the real world occurs when people are reminded of their real world, or factual events. The same process might occur when people are engaged with narratives and then return to and are reminded of their own reality. While being engaged with the narrative, the recipient’s attention is focused mainly on their world; however, a short time after they finish viewing the narrative, attention shifts to include also the real world—a process that might lead to a contrast effect.

The contrast effect might be more pronounced the more one identifies with the protagonist when reading or watching the narrative. Given that in identification, media consumers feel as if they are the protagonists, they might feel that they are comparing the present situation with their situation in the past. Thus, in this situation, the contrast between the media consumer and media character might be seen as a form of temporal comparison in which people compare their present self with their past self. Just as in temporal comparisons, here too we feel bad when the past is better than the future (e.g., Wilson & Ross, 2001). This situation might be more devastating than simple upward comparison with a successful other, because people feel that they were this successful other or that they were living this beautiful life just a short while ago. Since they feel that they had achieved something and now it is gone, people might also be devastated as a result of losing something they had (Tversky & Kahneman, 1981). Thus, we hypothesize that identifying with a protagonist enhances the tendency to assimilate the traits of the character into the self while watching the narrative and shortly after. However, following a short period of time in which the individual is able to think and their attention shifts to the real world, a contrast effect occurs and the media recipients feel further away from the media character.

Note that this suggested effect on the self, involving contrast following assimilation, differs from the assumed effects of narratives on attitudes and beliefs, which often tend to be long-lasting (e.g., Appel & Richter, 2007). When people leave the world of the narrative and reenter their own real world, they are unlikely to be faced with information that totally negates the assertions made in the narrative. However, when people who identify with a beautiful, successful protagonist leave the story world and come back to reality, they are faced with their own selves and realities that are often far from those of the protagonist. Thus, we suggest that with regard to the effects of the narrative on the self, the immediate assimilation effect might diminish and be replaced by a contrast effect. A contrast that follows assimilation with a successful media character might lead to the reverse effects described above as the consequences of the assimilation with the media characters. Examples include a negative perception of one’s characteristics and lower self-esteem.

5. The Current Research—Aim and Hypotheses

Most studies that examined the effect of engagement with narratives on the self examined the effect immediately following exposure to the narrative (e.g., Richter et al., 2014; Sestir & Green, 2010). The major aim of the current research was to demonstrate a difference between the immediate and delayed effects of narrative engagement on the self. The delayed effects are assumed to occur shortly after being exposed to the narrative, when the media consumers shift their attention from the narrative world to their own real world (McMullen, 1997). Thus, in the current research, following exposure to the narrative, the participants were given a 15 min break that allowed them to shift their attention to their own real world.

In this study, viewers watched a movie clip in which a protagonist overcame difficulties and succeeded in realizing their dreams. Psychological research demonstrates that people’s state self-esteem fluctuates in line with the feedback that they receive from other people (Leary et al., 2003) and with the successes and failures they experience (see van Dellen et al., 2011). Since in the process of identification, people feel momentarily that they are merged with the media character, the character’s success should make their self-esteem rise immediately following their viewing of the successful protagonist, as they might feel as if the success is their own. However, as their attention shifts to their own world and their own self, we expect their self-esteem to decrease, as they compare their own achievements with those of the media character, who is now a distinct entity from themselves. In other words, while identification is expected to lead to assimilation immediately following the exposure to the narrative, it is expected to lead to a contrast effect after a short delay. Thus, we hypothesized the following:

Hypothesis 1 (H1).

Self-esteem levels would be higher immediately following the watching of the movie clips than after the break.

Hypothesis 2 (H2).

Self-esteem levels measured immediately following the watching would be higher than those of the control group, and the delayed measures would be lower than those of the control group.

Hypothesis 3 (H3).

The level of identification with the protagonist would be positively correlated with the viewers’ self-esteem at time 1 and negatively correlated with it at time 2.

Measuring self-esteem using a self-report questionnaire at two points close in time might be problematic. It has been argued that participants might remember their reports at time 1 and reproduce them at time 2, especially with explicit tests taken twice within a short time interval (Knapp & Brown, 1995). Therefore, given that we intended to measure self-esteem twice in a short time frame, immediately after watching a movie clip and after a break, we used implicit measures of self-esteem for repeated measures, before and after the break, similarly to other studies analyzing IAT score changes (Zhang et al., 2021). Explicit measures, however, were measured as between-subjects variables, so that half of the participants reported them immediately following the viewing and half after a break. Thus, the study employed a 2 (between-subjects variable: timing of explicit self-esteem measurement) × 2 (within-subjects variable: timing of implicit self-esteem measurement) design.

6. Study 1

6.1. Method

We designed an online experiment in which the participants were assigned to watch one of four movie clips of people who succeeded in overcoming obstacles.1 The usage of four clips instead of one aimed to improve the validity of the study. We measured self-esteem both explicitly and implicitly. Explicit self-esteem was a between-subjects variable and was measured either immediately following the viewing or after a 15 min break. Implicit self-esteem was a within-subjects variable that was measured twice, immediately following the viewing and after the break.

6.2. Participants

Sample size planning was based on an a priori power analysis of repeated measures and a within–between interaction using G*Power (version 3.1.9.2; Faul et al., 2009). It indicated that a sample of 215 participants could reveal a small-size effect (f = 0.12, based on van Dellen et al., 2011) with a power of 0.80 for a design including five groups and two measurements.

We were able to recruit 389 participants, 50% female, through Panel4All, an Israeli online crowdsourcing platform (https://www.panel4all.co.il, accessed on 24 November 2024). The participants were native Hebrew speakers, aged 20 to 40 years old. Participants were dropped from the study if they did not agree to the consent form (n = 5). We also excluded from the analysis participants who did not return from the break within 15–25 min after it began (n = 7). The remaining participants were rewarded in accordance with Panel4All’s policy. After the exclusion of the above-mentioned participants, the final sample included 377 participants: 183 women and 194 men, ages 20–532 (M = 30.65, SD = 5.758).

6.3. Procedure

The participants connected to the study via a personal computer or a laptop. They clicked on the link to an online questionnaire that was prepared using Qualtrics software (Qualtrics©, http://www.qualtrics.com/uk/, 2019). After signing an informed consent form by clicking through an item stating their understanding of the explanations and agreement to participate, the participants were randomly assigned to one of five conditions: four experimental conditions that differed in the order in which the measures were presented and the movie clip participants watched, and one control group in which the participants did not watch a movie clip (see Supplementary Materials—S1—Table S1a).

In each of the experimental conditions, the participants were first asked to watch a 15 min movie clip and received a short summary of the film’s narrative. They were assigned to watch one of four clips. Following the movie clip, the participants were asked to complete an implicit association test (IAT) of self-esteem, as well as identification and transportation scales. Half of the participants were also asked to respond to statements from an explicit self-esteem scale.3 Then, the participants were asked to take a 15 min break. They were informed that they would not be able to continue the study if they did not return from the break within 25 min. After the break, the participants again took the IAT. The half of participants who did not respond to the statements on the explicit self-esteem scale before the break did so at this stage, so half of the participants answered the explicit self-esteem scale immediately following watching and half did so after a break. We also measured the desire of all participants to continue watching the movie and their enjoyment of watching it. Then, we asked the participants to answer two questions related to their understanding of the movie clip and a few demographic questions. Lastly, we asked the participants whether they suspected anything during the experiment and what they thought the aim of the study was. After they completed the questionnaire, the participants received a written debriefing.

6.4. Materials

We prepared four movie clips, each lasting about 13 min, in which the protagonist was physically attractive and succeeded in realizing their dreams. Each movie clip presented a different narrative, but all had the common theme of going through difficulties until ultimately achieving success. Two of the narratives included a male protagonist and two a female one. The clips were taken from the following movies: The Pursuit of Happiness (Muccino, 2006), Peaceful Warrior (Salva, 2006), Joy (Russell, 2015) and She’s the Man (Fickman, 2006). The movies selected were international releases, but not widely known locally in the age group of the participants. Only 15% of participants reported having seen the movies before, indicating that prior familiarity was limited. See the Supplementary Materials (S2) for the descriptions of the films’ synopses.

6.5. Measures

Implicit association test (IAT) of self-esteem.4 Based on the English versions of the IAT in general (Greenwald et al., 2003) and the IAT of self-esteem in particular (Verplanken et al., 2007), we devised a Hebrew version of the test to measure implicit self-esteem. As with the English version, people received higher self-esteem scores when they associated positive adjectives (e.g., good, proud) with first-person pronouns (e.g., I, me) and negative adjectives (e.g., bad, guilty) with third-person pronouns (e.g., they, them) faster than the opposite: negative adjectives with first-person pronouns and positive adjectives with third-person pronouns. Carpenter et al. (2019) presented a method for constructing and analyzing IATs for online survey software that has proven reliable and valid. Based on their open source tools (http://iatgen.org/), we constructed and analyzed the IAT for this specific study. The d-score data-cleaning and scoring algorithm was used, and IAT scores for participants with over 10% of trials shorter than 300 ms were not calculated, in line with Greenwald et al. (2003).

We conducted this test in the experimental groups twice: immediately after watching the movie clip (M = 0.497, SD = 0.399) and after a 15 min break (M = 0.417, SD = 0.403). In the control group, this test was conducted only once (M = 0.585, SD = 0.380).

Explicit measure of self-esteem. This measure was based on the performance and social sub-dimensions of the state self-esteem scale devised by Heatherton and Polivy (1991). Participants were asked to indicate the degree to which they felt that various statements were right for them at the moment. There were 10 statements such as “I feel confident about my abilities” and “I feel inferior to others at this moment” (reverse-scored). Responses to these statements were made on 1 (“strongly disagree”) to 7 (“strongly agree”) scales. We calculated a mean explicit self-esteem index (M = 4.991, SD = 1.030, Cronbach’s α = 0.878), for which higher values indicated greater self-esteem.

Identification. This measure was based on Cohen (2001). We asked the participants to indicate the degree to which they agreed with the following seven statements concerning the protagonist in the clip they had watched (P): ”I feel I have a good understanding of P”, “I understood the reasons for P’s deeds”, “While watching, I felt like P felt”, “While watching, I could really get into P’s head”, “At some moments while watching, I felt I knew exactly what P was going through”, “While watching, I wanted P to succeed in achieving his/her goals” and “When P succeeded, I felt joy, but when s/he failed, I was sad.” Responses to these items were given on 1–7 scales ranging from “strongly disagree” to “strongly agree.” We calculated this index based on the mean of the seven items (M = 5.638, SD = 1.091, Cronbach’s α = 0.916).

The psychometric properties of all study measures are described in the Supplementary Materials (S3).

The study data is publicly accessible at https://osf.io/rs3ce/?view_only=33f7e00e162846ad9e2e05ca6d1607ef (accessed on 9 July 2025).

7. Results

We first verified participants’ attention and understanding of the movie clips. Indeed, 94.4% of those in the experimental conditions answered at least one of the two comprehension questions correctly. Following this, we conducted an analysis of variance (ANOVA) for repeated measures with before and after IAT scores as the within-subjects factor. As H1 expected, IAT self-esteem scores were significantly higher before the break (M = 0.496, SD = 0.400) than after the break (M = 0.417, SD = 0.403), with F(1, 283) = 9.958, p = 0.002 and η²_p = 0.034.5

To examine H2, we conducted analyses of variances (ANCOVAs) in which gender was included as a covariate, because research has indicated that inspiring content has a different effect on men and women (Rieger & Klimmt, 2019). The group (experimental vs. control) was a between-subjects factor. As H2 expected, the IAT score of the control group (adjusted M = 0.584, SE = 0.043) was significantly higher than the experimental groups’ IAT scores after the break (adjusted M = 0.417, SE = 0.024), with F(1, 367) = 11.520, p < 0.001, MSE = 0.159 and η²_p = 0.030. Unexpectedly, however, the IAT score of the control group (adjusted M = 0.583, SE = 0.043) did not differ significantly from the experimental groups’ IAT scores before the break (adjusted M = 0.496, SE = 0.024), with F(1, 367) = 3.207, p = 0.074, MSE = 0.157 and η²_p = 0.009.

We also did not anticipate the pattern of correlations with identification that we found. Identification had a significant positive correlation with the IAT scores after the break, with r(284) = 0.119 and p = 0.046, and no significant correlation with the IAT scores before the break, with r(287) = 0.057 and p = 0.335. Thus, H3 was not supported6. To conclude this section on the main findings of Study 1, as hypothesized, the implicit self-esteem decreased from the first measurement to the second. However, unexpectedly, identification was positively correlated only with delayed self-esteem.

8. Discussion

As expected, the participants reported lower levels of implicit self-esteem after the break than immediately following the watching of the movie clips. We expected this pattern based on the premise that the assimilation process would be followed by a contrast. However, this expected finding was accompanied by the unexpected finding of a significant positive correlation of identification not with the immediate measurement of implicit self-esteem but with delayed self-esteem.

We expected that identification would enhance the assimilation process immediately following the watching of the movie clips, but would increase the contrast process after the break when the participants’ attentional focus shifted to include their own self and their own reality and when they could evaluate them relative to that of the protagonists. However, the positive correlation between identification and self-esteem after the break points to an alternative process. The positive effect of identification with the protagonist does not fade immediately following the return to one’s own reality. Indeed, this effect might actually increase over time. Thus, there might be some factor in the process of identification that makes the assimilation process last longer.

One potential explanation for this pattern of findings might lie in the conceptualization of retrospective imaginative involvement, which refers to the engagement of people with narratives after they end (e.g., Sethi et al., 2022). According to this idea, people continue their engagement with media characters long after being exposed to the narrative. This engagement might take various forms, including continued identification with the media characters (Slater et al., 2018).

Another explanation might lie in the conceptualization of the active self (Wheeler et al., 2007). The active self-concept is part of the chronic self-concept that is accessible at a particular time. According to Comello (2009), exposure to media stimuli might activate a particular self-concept. This result is similar but not identical to the priming effect of media exposure. It does not involve a simple activation of the concepts and attitudes raised in the media’s message but the activation of a particular aspect that already exists in the media consumer’s self. Indeed, a recent study documented that reading fiction tends to activate characteristics in the self that resemble those of the protagonist (Loi et al., 2023).

The pattern of findings we found might suggest that identification with a character activates a particular aspect already existing in the viewer’s self. Thus, following the viewing, instead of being reminded of their own self, which is distinct from that of the character, the viewers might keep on thinking about their own particular aspect of the self that is activated by exposure to the narrative. More specifically, they might think about their own struggles and their own successes. If identification indeed affects the active self, it might do so especially following a break in which the viewers had time to think about the relevance of the movie to their own self. Thus, based on the findings of Study 1, we expected identification to correlate with self-esteem specifically when it was measured after the break.

The stable effect of identification on the self might also stem from the type of movie clips that we used. The movie clips focused not only on the success of the protagonists, but also on their struggles to achieve it. Recent studies have demonstrated that perseverance and overcoming obstacles are among the most inspiring themes in media content (Chang, 2023). Moreover, research indicates that feeling inspired promotes the introduction of new ideas into one’s mind (e.g., Oliver et al., 2021). Thus, exposure to an inspirational narrative might result in changes in attitudes with regard to the self. Viewers who identify with a protagonist who succeeds in overcoming obstacles may form new attitudes about their own ability to overcome obstacles and succeed in achieving their goals. As in research on the effects of narratives on attitudes (e.g., Appel & Richter, 2007), this effect might be persistent and not disappear after the return to one’s own real world. If this is the case, we would expect that identification would correlate not only with self-esteem but also with self-efficacy and motivation. This expectation is based on previous research demonstrating a positive effect of inspirational narratives on these variables (e.g., Isberner et al., 2019; Nabi et al., 2024).

While the significant positive correlation between identification and implicit self-esteem at time 2 might hint at an assimilation effect that becomes even stronger with time, the reduction in implicit self-esteem from time 1 to time 2 might be seen as a contrast effect following the assimilation. One possible explanation for this inconsistency is that the reduction in implicit self-esteem with time is an artifact stemming from another reason that is not related to the response to the movie clip. Indeed, Greenwald et al. (2003) acknowledge a slight tendency for the second measurement of the IAT to produce smaller scores. To control for this possible artifact, in Study 2, we measured the IAT twice, before and after the break, not only in the experimental groups but also in the control group that was not exposed to any movie clip. This addition enabled us to examine whether the reduction in the IAT scores stems from a contrast effect related to the successful protagonist or from a phenomenon that is irrelevant to the movie itself. If the reduction in implicit self-esteem is indeed demonstrated in the control group as well, comparing the degree of reduction in implicit self-esteem between the control and experimental groups will help us distinguish between the artifact and the genuine change in self-esteem.

Most importantly, due to a fault in the design of the Qualtrics survey, in Study 1, the movie clips were not randomly assigned in each of the study’s conditions. Thus, the participants who reported their explicit self-esteem and motivation before the break necessarily watched different movie clips from those who reported their explicit self-esteem and motivation after the break. While this shortcoming did not impact the ability to appropriately examine the above hypotheses in relation to the implicit self-esteem measure, it prevented us from comparing the experimental conditions with respect to explicit self-esteem and motivation. Study 2 remedied this shortcoming by conducting a proper randomization of the movie clips so that in each experimental condition, the participants were randomly divided into four groups according to the movie they watched. This procedure enabled a comparison of the participants’ explicit self-esteem and motivation before and after the break because these conditions differed only in time of measurement and not in the type of clips watched.

Following Isberner et al. (2019), we also measured the participants’ self-efficacy either immediately after the exposure to the movie clip or after the break. If, indeed, the clips were inspiring and changed the viewers’ perceptions about their abilities and if, indeed, identification with the successful protagonist affected the active self and led the viewers to think during the break about the relevance of the movie to their own self, then we expected a stable and even intensifying assimilation effect. More specifically, we expected higher levels of explicit self-esteem, self-efficacy and motivation in the experimental groups than in the control group, especially when measured following the break.

Study 2 Hypotheses

To conclude, based on the findings of Study 1 and theory regarding inspiring media, identification and the active self, we hypothesized the following:

Hypothesis 4 (H4).

Implicit self-esteem would be lower after the break than before it, in both the experimental and control groups.

Hypothesis 5 (H5).

The level of identification with the protagonist would be positively correlated with the viewers’ self-esteem, self-efficacy and motivation after the break to a larger extent than immediately after viewing the clip.

Hypothesis 6 (H6).

Levels of explicit self-esteem, self-efficacy and motivation would be higher in the experimental groups than the control group, especially when measured following the break.

9. Study 2

9.1. Method

We designed a replication of Study 1 with several differences. In this study, we measured implicit self-esteem twice, before and after the break, not only in the experimental conditions but in the control condition as well. Moreover, in the current study, the films were randomly distributed in a way that enabled a comparison of the levels of self-esteem, motivation and self-efficacy in the experimental conditions.

9.2. Participants

Sample size planning for a design including three groups and two measurements indicated that a sample of 171 participants could reveal a small-size effect. We recruited 213 participants, 50% female, through Panel4All, an Israeli online crowdsourcing platform (https://www.panel4all.co.il). The participants were native Hebrew speakers, aged 20 to 40 years old. Participants were dropped from the study if they did not agree to the consent form (n = 6). We also excluded from the analysis participants who did not return from the break within 15-25 minutes after it began (n = 5). The remaining participants were rewarded in accordance with Panel4All’s policy. After the exclusion of the above-mentioned participants, the final sample included 202 participants: 103 women and 99 men, ages 20–40 (M = 30.35, SD = 5.748).

9.3. Procedure

As in Study 1, the participants connected to the study via a personal computer or a laptop. They clicked on the link to an online questionnaire that was prepared using Qualtrics software (Qualtrics©, http://www.qualtrics.com/uk/, 2019). After signing an informed consent form, the participants were randomly assigned to one of three conditions: two experimental conditions that differed in the order in which the measures of explicit self-esteem, self-efficacy and motivation were presented, and a control group in which the participants did not watch a movie clip (see Supplementary Materials—S1, Table S1b).

In each of the experimental conditions, the participants were first asked to watch a 15 min movie clip. They were randomly assigned to watch one of four clips. Following the movie clip, the participants were asked to complete an implicit association test (IAT) of self-esteem and respond to statements related to an identification scale. The participants in one of the experimental groups were also asked to respond to statements from an explicit self-esteem scale at that stage, as well as statements from a motivation scale and a self-efficacy scale.7 Then, the participants were asked to take a 15 min break. They were informed that they would not be able to continue the study if they did not return from the break within 25 min. After the break, the participants again took the IAT. The participants in the second experimental group who did not respond to the statements from the explicit self-esteem, motivation and self-efficacy scales before the break did so at this stage. Then, we asked the participants to answer two questions related to their understanding of the movie clip and a few demographic questions. In the control condition, the participants did not watch a movie clip. However, they took the IAT twice, before and after the break, and responded to the statements from the scales of explicit self-esteem, motivation and self-efficacy before the break. Lastly, we asked the participants whether they suspected anything during the experiment and what they thought the aim of the study was. After they completed the questionnaire, the participants received a written debriefing.

9.4. Materials

We used the same clips as in Study 1, to which the participants in each experimental group were randomly assigned. Only 13% of the participants reported having seen the movies before, suggesting that prior familiarity was limited.

9.5. Measures

Implicit association test (IAT) of self-esteem. We used the same test as in Study 1. We conducted this test twice: before the break (M = 0.522, SD = 0.380) and after a 15 min break (M = 0.440, SD = 0.404. As in Study 1, we did not calculate IAT scores for participants with over 10% of trials shorter than 300 ms, in line with Greenwald et al. (2003).

Explicit measure of self-esteem. We used the same measure as in Study 1. We calculated a mean explicit self-esteem index (M = 4.769, SD = 1.091, Cronbach’s α= 0.887), for which higher values indicated greater self-esteem.

Identification. We used the same measure as in Study 1 (M = 5.529, SD = 1.045, Cronbach’s α = 0.891).

Motivation. We devised a scale to measure general motivation to be a better person and live a better life. We asked the participants to indicate the degree to which they were motivated to behave in the following ways: to be a better person, to focus on what is really important in life, to live their lives better, to match their lives to their true wishes, to enjoy life, to work hard in order to succeed, to make an effort to overcome obstacles and to be persistent and not give up. Responses to these statements were made on 1 (“not at all”) to 7 (“to a very large extent”) scales. We calculated a mean motivation index (M = 6.011, SD = 0.903, Cronbach’s α = 0.930), for which higher values indicated greater motivation. Additional analyses provided initial validation of the motivation scale using factor analysis and correlations with related constructs (see Supplementary Materials, S4).

Self-efficacy. We used Schwarzer and Jerusalem’s (1995) scale translated into Hebrew. This scale has 10 items and four possible responses: 1, not at all true, 2, hardly true, 3, moderately true, and 4, exactly true. We calculated a mean self-efficacy index (M = 3.023, SD = 0.559, Cronbach’s α = 0.923), for which higher values indicated greater self-efficacy.

10. Results

We first confirmed that participants in the experimental conditions attended to and understood the movie clips, and indeed, all of them answered at least one of the two comprehension questions correctly. Subsequently, we conducted an analysis of variance (ANOVA) for repeated measures with before- and after-break IAT scores as the within-subjects factor. As H4 expected, the overall IAT self-esteem scores were significantly higher before the break (M = 0.522, SD = 0.381) than after the break (M = 0.440, SD = 0.404), with F(1, 192) = 7.712, p = 0.006 and η²_p = 0.039. Interestingly, this reduction in implicit self-esteem was especially prominent in the control group. Although in all of the groups the IAT score was higher before the break than after the break, only in the control group was the difference between the IAT score before the break (M = 0.548, SD = 0.366) and the IAT score after the break (M = 0.393, SD = 0.419) significant, with F(1, 70) = 8.701, p = 0.004 and η²_p = 0.111. However, there was no significant difference between the reduction in IAT scores in the control group (M = 0.155, SD = 0.0442) and that in the experimental groups (M = 0.040, SD = 0.386), with t(191) = −1.890, p = 0.06 and Cohen’s d = 0.277.

As H5 expected and Study 1 revealed, identification had a significant positive correlation with the IAT scores after the break, with r(122) = 0.248 and p = 0.006, and no significant correlation with the IAT scores before the break, with r(123) = 0.094 and p = 0.299. As H5 also expected, identification had a significant positive correlation with motivation when it was measured after the break, with r(63) = 0.298 and p = 0.018, but not when it was measured before the break, with r(64) = 0.200 and p = 0.113. Unexpectedly, there was no significant correlation between identification and explicit self-esteem and self-efficacy, either before or after the break, with p > 0.1. Thus, H5 was only partly supported (for bivariate correlations across experimental conditions, see Supplementary Materials—S5, Table S5).

Lastly, we conducted a multivariate analysis of covariances (MANCOVA) to examine whether the levels of explicit self-esteem, self-efficacy and motivation were higher in the experimental groups than the control group, especially when measured following the break (H6). As in Study 1, we controlled for the participants’ gender. There was a significant effect of the experimental condition on the dependent variables mentioned above (self-esteem, self-efficacy and motivation), with Hotelling’s trace = 0.083, F(6, 390) = 2.703, p = 0.014 and η²_p = 0.040. The univariate tests with the experimental group (measured after the clip, after the break and in the control group) as the independent variable were significant when the dependent variables were self-efficacy, with F(2, 198) = 5.159, p = 0.007 and η²_p = 0.050, and motivation, with F(2, 198) = 5.298, p = 0.006 and η²_p = 0.051. When the dependent variable was explicit self-esteem, the groups did not differ significantly, with F < 1. Post hoc analyses, depicted in

Table 1. The effect of the time of measurement on motivation, self-efficacy and self-esteem.

Dependent Variable	Before Break	After Break	Control
Motivation	6.113 (0.913) ab	6.200 (0.763) a	5.765 (0.957) b
Self-efficacy	3.103 (0.523) a	3.137 (0.485) a	2.860 (0.612) b
Explicit self-esteem	4.777 (1.164) a	4.865 (1.158) a	4.683 (0.969) a

Note. Different letters in a row denote a significant difference between groups; p < 0.05.

, revealed that for motivation, only the post-break measure was significantly different from that of the control group. Self-efficacy was higher than in the control group both before and after the break. Thus, H6 was also partly supported (see note 6).

To conclude this section on the main findings of Study 2, the reduction in implicit self-esteem between the two measurement times that was found in Study 1 was replicated in the control group. Like in Study 1, identification was positively related to implicit self-esteem and motivation only when these constructs were measured after the break and not when they were measured immediately following the watching of the movie clips. Moreover, the self-efficacy of the experimental groups who watched the movie clips was higher than that of the control group, irrespective of the time of measurement. However, motivation was higher in the experimental groups compared with the control group only in the delayed measurement, after the break.

11. General Discussion

Our study investigated how the effect of narratives on the self changes from the time of exposure to a later point in time. Previous research measured these effects immediately following narrative exposure, revealing an assimilation effect (e.g., Richter et al., 2014; Sestir & Green, 2010). We initially assumed that these assimilation effects would diminish when the viewers returned to their own world outside the narrative. We based our hypotheses on psychological research demonstrating that when people fantasize about positive or negative events occurring for them, a contrast effect can follow assimilation when people are reminded of their own reality (Dermer et al., 1979).

The findings of Study 1 seemingly contradicted each other. In line with our expectations, implicit self-esteem was higher immediately following the watching of the video clip than after the 15 min break in which the participants had the time to be reminded of their own reality. On the other hand, identification with the protagonist was positively correlated with self-esteem only when this measure was taken after the break. This last finding prompted us to rethink the temporal change in the effects of narratives on the self. There might be something in the effect of the narrative on the self that is more consequential to the self than simple fantasy. The assimilation effect following exposure to the narrative might not be superficial, and the identification with the protagonist might not involve only a temporary adoption of the character’s traits. In line with the idea of the active self (Wheeler et al., 2007), identification with a protagonist who demonstrates particular positive traits might prime those traits that already exist in the viewer’s self. This explanation is in line with recent research that found that reading fiction tends to activate characteristics in the present self that are similar to those of the protagonist (Loi et al., 2023).

Moreover, the narratives we used depicted protagonists who succeeded in overcoming difficulties. Previous research has indicated that such narratives are inspiring (Chang, 2023). When viewers are inspired, they believe that they have the ability to overcome obstacles just as the protagonist did and are motivated to act in order to be the person they want to be (e.g., Isberner et al., 2019; Nabi et al., 2024). Thus, the effect of inspiring narratives on the self might be similar to the effects of narratives on attitudes. Narratives actually change one’s attitudes about oneself.

In order to examine the new hypotheses, we devised Study 2 in which we measured implicit self-esteem twice, not only in the experimental groups but also in the no-treatment control group. We also measured self-efficacy and motivation in the experimental groups either before or after the break. This design allowed us to determine that the reduction in implicit self-esteem observed in Study 1 was not the result of contrast following assimilation but rather an artifact, as it was found also for control group participants who did not watch any movie clips. Most importantly, this reduction in implicit self-esteem was significant only in the control group, and not in the experimental groups. If the impact of the artifact were diminished, it would reveal the genuine effect of the movie clip on self-esteem at the two different time points. We can speculate that under these conditions, implicit self-esteem following the break might have been higher than immediate implicit self-esteem.

As expected, and in line with the findings of Study 1, identification with the protagonist was positively correlated with implicit self-esteem and motivation only when these measures were taken after the break. Taken together, these findings suggest that assimilation persists after watching a narrative and returning to one’s own real world thanks to the viewer’s enduring identification with the protagonist. These findings add to the accumulated knowledge on media narratives, which are based on studies examining the effects of these narratives in the form of newspaper articles (e.g., Thier et al., 2021), books (Wimmer et al., 2021), social media (e.g., Rieger & Klimmt, 2019), commercials (e.g., Thomas & Grigsby, 2024), TV programs (Tchernev et al., 2023) and films (e.g., Igartua, 2010). Our findings strengthen past studies that demonstrated a tendency of ascribing to the self the attributes of the character and measured this effect immediately following exposure (e.g., Richter et al., 2014; Sestir & Green, 2010). They also add to our understanding of the effect of identification on the self, as previous research concentrated on the immediate effect of identification (e.g., Sestir & Green, 2010).

Thus, we cautiously conclude that the effect of media narratives on the self is not limited to the time of exposure. Still, it is important to consider the generalizability of our conclusions. Although we used four movie clips with both male and female protagonists, in all of them, the protagonists succeeded in overcoming various obstacles before they achieved their goals. The initial hypotheses regarding the contrast effect that occurs when viewers return to their real world might manifest itself in less-inspiring narratives that focus only on the protagonists’ wealth, beauty and/or successes. In these narratives, the viewers might indeed experience a drop in their self-view when returning to their real world. Future research might compare the temporal change in the effects on the self of various types of narratives that differ in the degree to which they inspire the viewers. Similarly, our study is limited in its conclusions about the persistence of the narrative and identification effects on the self, as it measured the delayed effect only 15 min after exposure to the narrative. Future studies should examine the persistence of the effects over longer periods of time. Lastly, as is common these days (e.g., Igartua et al., 2025; Myrick & Bilandzic, 2025), our study was conducted online. Consequently, we could not control for factors such as the environment in which the participants were located. Although random assignment to conditions helps protect against threats to internal validity that might stem from these uncontrolled factors, they might have enhanced the within-group variance, which may have reduced the likelihood of detecting additional significant effects.

Note that while the findings of both studies generally support our conclusions about the stable effects of narratives on the self and the positive contribution of identification to this persistent effect, not all of the findings completely fit with this inference. Specifically, there was no significant correlation between identification and explicit self-esteem and self-efficacy, either before or after the break. Moreover, while motivation levels were significantly higher in the experimental groups than in the control group only in their post-break measure, the level of self-efficacy in the experimental groups was higher than that in the control group both before and after the break, and the experimental groups’ explicit self-esteem did not differ from that of the control group. It seems that the movie clips were very influential on self-efficacy to the degree that identification did not matter, indicating a ceiling effect of self-efficacy. On the other hand, they were not influential at all on explicit self-esteem, to the degree that, again, identification did not matter, indicating a floor effect of explicit self-esteem. The movie clips focused specifically on overcoming obstacles. Therefore, they affected perceived self-efficacy, which was measured by asking the participants about the degree to which they sensed that they could achieve their goals and overcome various difficulties. Conversely, explicit self-esteem referred to a more generalized sense of self-worth that was not affected by the short and specific movie clips.

Conclusions

Despite its limitations, we believe that the current study makes an important contribution to the media psychology literature. It demonstrates that narratives can change people’s attitudes about themselves and that this effect persists beyond the time of exposure to the narrative. It also indicates that identification with the narrative’s protagonist plays an important role in this process. There might also be a practical application for our findings. People who want to increase their self-efficacy and motivation levels might be advised to expose themselves to narratives in which there are protagonists with whom they tend to identify and who succeed in overcoming obstacles on the way to success.