Abstract
This study examines if accounting for the causal location of information within a narrative can improve the predictability of narrative persuasion. Using perceived realism as a variable of narrative persuasion and environmental communication as a context, results reveal a significant moderating influence of the location of information relative to the cause-and-effect structure of the narrative. Specifically, external realism increased the acceptance of narrative information, but only after accounting for the additional variance of location and only for the subset of information on the causal line of the narrative. Future studies should continue to explore how the variance associated with this narrative causality can be leveraged toward a more nuanced understanding of narrative persuasion.
Narratives represent a promising tool for communication practitioners to influence the acceptance of information. Narratives represents a format of communication using a causally linked, temporal sequence of events involving specific humanlike characters and are often contrasted with other formats of communication (Longacre, 1983), most notably with that of logical or evidence-based communication (Fisher, 1984). Examples of narrative range from short exemplars or anecdotes that may be contained within a larger message (Zillmann, 2006) to the detailed and lengthy stories common in entertainment media (Slater, Rouner, & Long, 2006).
Information communicated through narratives often result in greater information acceptance through their ease of processing and comprehension (Graesser, Olde, & Klettke, 2002; Schank & Abelson, 1995), the generation of fewer counterarguments (Green, 2006; Niederdeppe, Shapiro, & Porticella, 2011; Slater, 2002) and the increase in self-efficacy through modeling and identification (Bandura, 1986, 2002; Moyer-Guse, 2008). This influence would obviously benefit communication areas with a persuasive intent, but narrative persuasion could also benefit more informative communication areas, such as science or risk communication that often promote rational attitudes or behaviors regarding contentious issues.
Unfortunately, narrative persuasion remains difficult to predict, often failing to influence audiences as intended or contradicting previous findings (Prentice, Gerrig, & Bailis, 1997; Slater, Buller, Waters, Archibeque, & LeBlanc, 2003; Slater, et al., 2006; Wheeler, Green, & Brock, 1999). As a consequence, researchers have called for a more nuanced study of narratives (Davin, 2003; Kreuter et al., 2007) and even suggested completely avoiding the use of narratives in health-related situations until their effects are more clearly understood (Winterbottom, Bekker, Conner, & Mooney, 2008).
One claim is that because most studies of narrative persuasion use the entire narrative as the unit of analysis, there remains significant overlooked variance within narratives that may help improve the predictability of narrative persuasion (Dahlstrom, 2010). As evidence of this claim, Dahlstrom (2010) found that the acceptance of narrative information as being accurate in the real world depended on the information’s location relative to the internal cause-and-effect structure of the narrative. Specifically, information that caused future narrative events to occur were more persuasive than the same information presented with no causal link to future narrative events within the same narrative (Dahlstrom, 2010, 2012). Because a cause-and-effect structure is present in all narratives, it was claimed that narrative causality might serve an important moderating role with previously studied variables of narrative persuasion.
One such variable that can be used to explore this claim is perceived realism, or how similar a narrative is perceived to be with regard to reality. It is often suggested that narratives with a higher degree of perceived realism should be more persuasive than narratives with less perceived realism. However, empirical evidence for this assumption has been mixed, potentially due to a poorly defined conceptualization of the construct (Busselle & Greenberg, 2000; Potter, 1986). Perceived realism is a useful variable for exploring the potential moderating impact of narrative causality for two reasons: (a) causality effects are already related to realism through their impact on perceptions of information being accurate relative to the real world (Dahlstrom, 2010, 2012) and (b) the inconsistent effects of perceived realism suggest a significant portion of variance may yet be unaccounted for.
Therefore, the aim of this study is to use perceived realism to begin investigating if narrative causality does serve as a significant pool of untapped variance that could improve the predictability of narrative persuasion. If so, leveraging this variance could offer the field a method of improving existing theory and practice. The measured outcome of narrative persuasion generally falls into two categories. The first focuses on acceptance of specific, belief-level assertions stated within the narrative as true in the real world (Appel & Richter, 2007; Dahlstrom, 2010; Marsh, Meade, & Roediger, 2003) while the second focuses on the alignment of global, normative-level attitudes toward the moral of the narrative (Green & Brock, 2000; Slater, et al., 2006). The former stands to benefit more directly from leveraging this variance as assertions represent specific points within a narrative that can more easily be moved with regard to causality. The latter is more complex as acceptance of a moral requires assessing the aggregate meaning of multiple events within a narrative against existing beliefs and experiences. While these attitude studies may benefit indirectly from leveraging this variance by weighting specific events toward a stronger aggregate, this article will begin exploring the moderating impact of narrative causality by focusing on the potential direct effects of assertion acceptance.
Narrative Persuasion and Narrative Causality
The field of narrative persuasion examines how narrative communication, often in the form of fictional stories consumed for entertainment purposes, can persuade about real world topics. Many of the models proposed to explain this persuasive influence, including the transportation-imagery model (Green & Brock, 2000), extended elaboration likelihood model (Slater & Rouner, 2002) and entertainment overcoming resistance model (Moyer-Guse, 2008), claim that transportation, or engagement with the narrative, as well as identification with characters serve to increase persuasive impact through reducing the formation of counterarguments, lessening message scrutiny and/or inhibiting psychological reactance.
However, the persuasiveness of a particular narrative and its direction of influence have proven difficult to predict. Narrative persuasion has been found to lose much of its influence once the reader becomes aware of the persuasive intent (Moyer-Guse, 2008) and many factors associated with studies of narrative persuasion are psychologically based, which provide little guidance for how to create specific narratives that can better activate those psychological states (O’Keefe, 2003). As such, there have been multiple calls within the field for a more nuanced understanding of narrative persuasion (Davin, 2003; Kreuter, et al., 2007; Winterbottom, et al., 2008). Yet difficult to predict.
Narrative causality represents a relatively new construct developed to explore some of these inconsistencies (Dahlstrom, 2010). Based on studies of mental models in discourse psychology that examines how the mind comprehends narrative texts (Fletcher, Hummel, & Marsolek, 1990; Trabasso & Sperry, 1985), narrative causality refers to the placement of information within a narrative relative to its underlying cause-and-effect structure. Information is termed causal if the content is the direct cause of later narrative events, whereas noncausal information describes content with no direct influence on narrative events that could be removed without disrupting the storyline of the narrative (Dahlstrom, 2010). While all narratives contain a causal chain by definition, the ratio of causal to noncausal information within any single narrative depends on its creation and subsequent telling.
These causal and noncausal areas of information are referred to as “locations” within the particular structure of an individual narrative because the content of the information does not depend on its causal or noncausal categorization. For example, the statement, “Frequent hand washing is the best method of stopping the spread of disease,” could serve as a causal statement if a character altered his or her actions based on the statement and therefore avoided catching a disease. The same statement could serve as a noncausal statement if the character heard or spoke the statement but it caused no alterations of future actions for the characters involved.
It was found that when information was inserted into a narrative at a causal location, the information was perceived as more likely to be true in the real world than the same information inserted in a noncausal location in the same narrative (Dahlstrom, 2010). This narrative causality effect appears to be a consequence of the intrinsic importance of causality in the comprehension of narratives (Dahlstrom, 2012; Magliano, 1999; Trabasso & Sperry, 1985).
Because a cause-and-effect structure is intrinsic to all narratives, it is claimed that the effect of narrative causality would likely interact with other variables of narrative persuasion, providing an untapped source of variance that may improve the predictability of narrative persuasion (Dahlstrom, 2010). Likewise, such a relationship could offer communication practitioners message-level recommendations toward leveraging the causal structure of a particular narrative toward their persuasive goals. The purpose of this study is to explore the usefulness of this variance through its potential interaction with the related variable of perceived realism.
Perceived Realism
Perceived realism refers to the level of similarity associated between a narrative and the real world and it is commonly assumed that more realistic narratives should be more persuasive than less realistic narratives (Busselle & Greenberg, 2000). However, the ambiguity of defining “similarity” has been a common critique of scholars who have argued for a multidimensional conceptualization of perceived realism for decades (Potter, 1986).
Because of this lack of conceptual clarity, perceived realism has appeared under many different terms and operationalizations. For instance, perceived realism is sometimes defined as the perceived fictional status of a narrative (fiction versus nonfiction). This fictionality construct has been subject to multiple tests, and does not seem to influence narrative persuasion to a large degree (Green & Brock, 2000; Green, Garst, Brock, & Chung, 2006). Likewise, the familiarity of a narrative, or the degree of overlap between the content of the narrative and the audience’s previous knowledge, also represents a definition of perceived realism but has also received mixed results with regard to an influence on narrative persuasion (Prentice, et al., 1997; Wheeler, et al., 1999). Busselle and Greenberg (2000) provide an extensive review, and critique, of many of these perceived realism conceptualizations, claiming that such inconsistencies potentially impede future research into the construct.
Even though perceived realism is weakly defined and inconsistently measured, the consistent focus suggests a belief that the realism of a narrative does have some effect on its persuasiveness. A recent model of perceived realism attempts to address these critiques through a mental models approach and divides perceived realism into three dimensions: fictionality, external realism, and narrative realism (Busselle & Bilandzic, 2008).
The dimension of fictionality remains similar to what has been described in the literature, perceptions of a narrative as either being fictional or nonfictional. External realism represents the similarity of aspects in the narrative world to the real world. Examples of high external realism could include settings (a living room, a forest, New York City), actions (driving a car, paying taxes, brushing hair) or characters (a postal worker, a parent, an academic researcher) that might reasonably be observed outside the narrative world. In contrast, narrative realism represents the similarity of aspects in the narrative world to what would be expected with regard to the narrative itself. For example, a mixed-alien-race character fighting giant insects within a floating city would have very little external realism, but may exhibit high narrative realism if the settings, actions, and characters are coherent relative to the rules set forth in the narrative world (Busselle & Bilandzic, 2008). Fictionality can still vary independently of external realism, with a fictional narrative about externally real characters or events, or a nonfictional historical narrative that no longer represents an externally valid representation of reality.
This model is relevant for the current study because its foundation in mental models matches the theoretical foundation on which narrative causality is based. In addition, many of the earlier concepts of perceived realism were designed for television, but this model has been developed for narratives in general, lending itself greater generalizability.
Study Objectives
The purpose of this study is to investigate if narrative causality represents a meaningful, yet overlooked, pool of variance that may improve the predictability of narrative persuasion. Perceived realism represents a previously studied and related variable that can serve to test the usefulness of such an interaction.
To explore this relationship, this study will use science communication as its context, specifically the acceptance of environmental assertions. The effects of environmental communication have long been a focus of science communication for representing a complex interplay between science and society (Ader, 1995; Besley & Shanahan, 2004; Boykoff & Boykoff, 2004; Corbett & Durfee, 2004; Griffin & Dunwoody, 1997). In general, it is suggested that narratives can have an effect on environmental beliefs, and that the environment, like many other science issues, may be especially susceptible to mediated effects because individuals have less opportunities for direct experience on which to base their beliefs. The literature suggests that an individual’s environmental understanding relies disproportionately on either the mass media (Ader, 1995; Shanahan & McComas, 1999) or on public opinion leaders (Nisbet & Kotcher, 2009). This susceptibility of environmental beliefs on mediated communication offers a strong context with potentially large effect sizes in which to investigate the larger questions of interest.
The design of this study mirrors belief-based narrative persuasion methodologies (Appel & Richter, 2007; Marsh, et al., 2003) that insert assertions into a narrative and measures the perceived truthfulness of those assertions after exposure to the narrative. The first hypothesis attempts to replicate the narrative causality effect that finds assertions placed at causal locations result in greater acceptance than the same assertions placed at noncausal locations.
Hypothesis 1(H1): Assertions inserted at causal locations of a narrative will be perceived as more truthful with regard to the real world than the same assertions inserted at noncausal locations within the same narrative.
The other main effect examines the influence of perceived realism on information acceptance. Of the three perceived realism dimensions in the Busselle and Bilandzic model, the first, fictionality, has received previous attention in the literature with little relevant effects (Green, et al., 2006). Therefore, this study will focus on the remaining two dimensions, external realism and narrative realism. Because there have been few effect studies contrasting these dimensions, the potential influence on information acceptance will be investigated using a research question.
Research Question 1(RQ1): How do the dimensions of external and narrative realism influence the acceptance of information from a narrative?
The main question of interest for this study regards the potential interaction between narrative causality and perceived realism. Such an interaction would support the claim that narrative causality can account for meaningful, overlooked variance within the current understanding of narrative persuasion. Because there are yet no directional expectations on which to base a hypothesis, this potential interaction will also be investigated with a research question.
Research Question 2 (RQ 2): How does the causal placement of information within a narrative interact with external and narrative realism to influence information acceptance?
Method
Subjects
Two hundred and seventy five undergraduate students served as participants for this study. 1 Participants were offered extra credit as compensation. The mean age of the participant pool was 19.91 (SD = 1.73) and the group comprised of 72% females.
Experimental Design and Procedure
This study used a mixed-factorial design consisting of three levels of within-subject assertion placement (causality: causal, noncausal, control) crossed within a between-subjects 2 (external realism: high low) × 2 (narrative realism: high low) design. Participants were informed that they would be involved in a study addressing how information is presented in the media. They were initially contacted by e-mail and provided with a web link to an online experiment. Participants were presented with a set of questions measuring individual differences and standard demographics and were then randomly assigned to read one of 12 stimulus narratives. Participants were then presented with a set of questions designed to measure the constructs of interest.
Stimulus
The stimulus consisted of a fictional narrative in the style of a short story that would be read for entertainment. This style was chosen to align with previous studies examining the relevant variables. Creation of the stimulus narrative required the development of (a) a list of assertions and (b) a template narrative within which the assertion locations could be manipulated regarding its causal location.
The assertions were created to represent false, antienvironmental statements. The falsity helped to ensure that any acceptance was due to the stimulus and not to the reinforcement of preexisting environmental knowledge. An alternative method would have been to measure and then control for previous knowledge, but this introduces potential priming effects that could be avoided by using false assertions. The antienvironmental bias of both provided a consistent ideological view from which to measure acceptance without introducing an additional variable of bias within the assertions and avoids the artifact of agreeing with environmentally supportive information based on normative expectations.
Therefore, a pool of 12 false, antienvironment assertions was created such as, “More energy is saved by leaving your computer on overnight than by turning it off and starting it back up in the morning,” and “Urban and agricultural areas currently account for less than 15% of available land area.” Four proenvironmental assertions were also included into the narrative to help disguise the persuasive intent of the narrative and avoid the appearance of a completely antienvironmental narrative as narrative persuasion has been found decrease if the persuasive intent becomes salient (Moyer-Guse & Nabi, 2010). All four proenvironmental assertions were not manipulated nor included in any analyses.
The template narrative into which these assertions could be manipulated had to meet certain requirements. First, the structure of the template narrative needed to contain multiple slots, each of which could be filled with various assertions without disrupting the logic of the narrative world. Second, the topic of the template narrative also needed to conceal its persuasive attempt for the same reason as with the assertions; the narrative needed to state antienvironmental assertions without the theme of the narrative relating to those assertions.
To meet these requirements, the template narrative told the story of Lucas, a science center employee in competition with a nearby museum to attract more visitors to its Earth Day exhibits. Lucas was in charge of creating the exhibits and he repeatedly focused on the artistic presentation of each exhibit at the expense of scientific accuracy. His competitor attempted to reveal these inaccuracies, thereby increasing traffic to his competing museum, by handing out flyers of environmental facts that continually contradicted Lucas’s exhibit. After repeatedly being reprimanded by the director for his incorrect environmental facts, Lucas finally accepted the importance of accuracy and vowed to improve his performance for future exhibits. The neutral setting of a museum was selected over a more controversial environmental setting to better hide the persuasive intent, and the conflict between Lucas and his competitor allowed for environmental assertions to be inserted according to the logic of the narrative without moralizing about environmental themes.
Measures
Independent Variables
External realism and narrative realism served as the two between-subjects factors and were arranged in a 2×2 high/low design resulting in four initial versions of the stimulus narrative that differed on these variables.
External realism was manipulated by altering the characters and settings of the narrative template to create a story that was either possible or impossible with regard to the real world. The events and core narrative structure remained unchanged. The high external realism narrative retained the narrative template’s characters and setting of a young college intern named Lucas designing Earth Day exhibits for the Metropolitan Museum of Science. The low external realism narrative transformed the characters from humans to speaking animals to represent a fantasy impossible outside the narrative world.
Narrative realism was manipulated by altering behaviors or descriptions of characters to create a story that was either consistent or inconsistent with regard to the narrative world. Again, the events and core narrative structure remained unchanged. The high narrative realism narrative retained the consistent behaviors and descriptions from the narrative template, such as the protagonist acting happy when he succeeded or upset when he failed. The low narrative realism narrative reduced the consistency of these behaviors and descriptions, where the protagonist may act upset with success or relieved upon failure, or where descriptions changed, such as a minor character switching from male to female with no explanation.
Care was taken to ensure that there was no confusion of the participant as to the fictional nature of the narrative to avoid introducing the confounding realism dimension of fictionality. Therefore, the stimulus narrative was also always referred to as a “fictional short story” and its fictional nature was highlighted again before exposure to questions of perceived truthfulness.
Causal placement was manipulated by inserting assertions at one of three locations relative to the narrative: causal, noncausal, or absent. Causal locations represent events that lie on the main cause-and-effect chain within the narrative and therefore cause future events to occur. Noncausal locations represent statements that lie off the main cause-and-effect chain within the narrative and have no impact on future events. Absent locations represent a control that were not included in the text of that particular narrative.
The template narrative was written to contain four causal and four noncausal slots that could be logically filled with various assertions. To account for any differences of memorability or perceived truthfulness between the assertions, three versions of each narrative were created to allow each assertion to be measured in each location.
To illustrate, at one point in the story, the protagonist creates an exhibit showing the dangers of water pollution. The exhibit displays a large fish kill, long-term damage from an oil spill and massive amounts of chemical runoff dumping into rivers. Visitors then enter the museum carrying flyers of environmental facts distributed by the protagonist’s competitor. Throughout the narrative, the protagonist overhears visitors reading assertions from their flyers. These assertions represent noncausal slots because their declaration in the narrative does not cause the protagonist or any other character to alter their actions. However, one of the causal slots occurs when the protagonist overhears a visitor complain about his water pollution exhibit because it does not align with an environmental fact as read from the flyer. This causes the protagonist to change his course of action and address the shortcoming in the following exhibit.
Associated with this particular causal slot were three assertions that the visitor could read to contradict the water pollution exhibit and serve as a causal assertions: (a) Fish populations are now increasing in the wild; (b) The effects of an oil spill typically last less than a 1 year; (c) Over 90% of runoff contaminants are eliminated by microbes before entering water systems. In the first version of the narrative, the first assertion was placed into this causal slot, the second assertion was placed into a noncausal slot later in the narrative when the water pollution exhibit was no longer the focus and thus had no impact on the narrative’s events. The third assertion was left out of the narrative as a control.
The second version of the narrative rotated the order so that the second assertion was placed into this causal slot, the third assertion was placed into the noncausal slot and the first assertion was left out of the narrative. The third version completed this rotation so that each assertion was measured at each possible location. The final narrative contained four causal and four noncausal slots and these slots alternated to control for order effects. The three versions of each narrative were created to control for variability in the assertions and do not represent meaningful manipulations in of themselves. Of interest is comparing the within-subject influence of causal placement regardless of the assertions that were placed in those locations.
Taken together, these dependent variables resulted in the creation of 12 stimulus narratives: three narratives based on rotations of the three within-subject narrative causality locations for each of the narratives designed for the four between-subject perceived realism categories. Final word counts were similar across all 12 stimuli with the largest deviation being between 3,015 and 3,060 words.
Dependent Variable
Perceived truthfulness was measured by asking participants to rate the likelihood of each assertion being true in the real world on a 7-point scale with possible responses ranging from “absolutely false,” “probably false,” “possibly false,” “don’t know,” “possibly true,” probably true” to “absolutely true.” Responses were coded so larger numbers represented greater levels of perceived truthfulness (M = 4.20, SD = 0.94). There was no significance difference in average perceived truthfulness between the three versions of the narrative within each of the four perceived realism treatments, F(2, 57) = 2.37, p = .10, ηp2 = 0.08, F(2, 54) = 0.37, p = .69, 10, ηp2 = 0.01, F(2, 67) = 0.89, p = .42, 10, ηp2 = 0.03, F(2, 67) = 0.65, p = .52, 10, ηp2 = 0.02, allowing for a collapse of all narrative versions within the four perceived realism treatments.
Perceived truthfulness for each of the three possible causal locations was calculated as the average of the four specific assertions each individual saw that represented those locations in their narrative version.
Manipulation Checks
Manipulation checks were conducted to ensure that the message-level alterations used to manipulate the two dimensions of perceived realism were successful. 2
Perceived external realism served as the first manipulation check. While there have been overall measures of perceived realism used in previous research, it is not clear which items specifically measure perceived external realism rather than other dimensions of perceived realism. Therefore, 10 questions were constructed based on the description of external realism from Busselle and Bilandzic (2008) and the difference between possibility and typicality of external realism (Busselle & Greenberg, 2000). The questions asked the participants to rate both the typicality and the possibility of five aspects of the narrative occurring in real life: (a) the story as a whole, (b) the events that occurred in the story, (c) the setting used in the story, (d) the type of characters used in the story and (e) the behavior of characters in the story. Each of the 10 questions was rated on a 7-point scale ranging from agree to disagree. Even though the possibility ratings averaged a full point higher than the typicality ratings, the two groups were combined because of the similar pattern between them and the resulting larger reliability of the measure (α = .84). Responses were coded so larger numbers represented greater levels of perceived external realism.
Perceived narrative realism served as the second manipulation check and its measure was constructed as the average of six questions. All questions were rated on a 7-point scale ranging from agree to disagree. Three of the questions were taken from a narrative realism measure of Busselle and Bilandzic (2009): (a) I understood why the events unfolded the way they did, (b) at some points in the story, it was not quite clear why something happened (reverse coded) and (c) At points, I had a hard time making sense of what was going on in the story (reverse coded). Three additional questions were added to mirror the form of the external realism questions: (d) The events that occurred in the story seemed to fit with the fictional setting, (e) The characters that occurred in the story seemed to fit with to the fictional setting and (f) The behavior of the characters in the story seemed to fit with the fictional events that happened. Because the first and last three items came from different sources, each group was assessed for internal consistency. Reliability for the first three items (α = .73) and the last three items (α = .69) were both less than the combined measure (α = .79). Therefore, all six items were treated as measuring the same construct of interest. Responses were recoded so larger numbers represented greater levels of perceived narrative realism.
An ANOVA analysis confirms that both manipulations successfully influenced perceptions as expected. The mean of the perceived external realism of the high external realism manipulation (M = 5.40, SD = 0.82) was significantly higher than the perceived external realism of the low external realism manipulation (M = 4.35, SD = 1.11), F(1, 250) = 69.71, p < .01, 10, ηp2 = 0.22, with no significant difference across the narrative realism manipulation, F(1, 250) = 0.74, p = .39, 10, ηp2 < 0.01. Likewise, the mean of the perceived narrative realism of the high narrative realism manipulation (M = 5.71, SD = 0.93) was significantly higher than the perceived external realism of the low narrative realism manipulation (M = 5.42, SD = 1.15), F(1, 262) = 5.23, p = .02, 10, ηp2 = 0.02, with no significant difference across the external realism manipulation, F(1, 262) = 0.50, p = .48, 10, ηp2 < 0.01.
Results
The first hypothesis predicted that assertions placed at causal locations within the narrative would be perceived as more truthful than assertions placed at noncausal locations within the narrative. The omnibus test of perceived truthfulness of assertions between the three within-subject levels of narrative causality was significant, F(2, 512) = 55.51, p < .01, 10, ηp2 = 0.18. The mean of the perceived truthfulness of assertions at causal locations (M = 4.67, SD = 1.49), was significantly higher than the assertions at noncausal locations (M = 4.28, SD = 1.28), F(1, 256) = 20.94, p < .01, 10, ηp2 = 0.08, and both causal and noncausal assertions were significantly perceived as more truthful than the control assertions not present in the narrative (M = 3.69, SD = 1.06), F(1, 256) = 88.26, p < .01, 10, ηp2 = 0.26, F(1, 256) = 43.44, p < .01, 10, ηp2 = 0.15. Therefore, Hypothesis 1 was supported.
The first research question addressed the effects of perceived realism on the perceived truthfulness of the assertions. In response, neither the main effect of external realism, F(1, 253) = 2.80, p = .10, 10, ηp2 = 0.01, nor narrative realism, F(1, 253) = 0.46, p = .50, 10, ηp2 < 0.01, significantly influenced overall perceived truthfulness of the assertions. Likewise, the interaction term also failed to reach significance, F(1, 253) = 0.01, p = .91, 10, ηp2 < 0.01.
The second research question addressed the interaction of perceived realism and narrative causality on perceived truthfulness. The omnibus test of the interaction between external realism and the three within-subject levels of narrative causality was significant, F(2, 252) = 3.12, p = .05, 10, ηp2 = 0.02. As seen in Figure 1, the mean of the perceived truthfulness of causal assertions in the high external realism treatment (M = 4.93, SD = 1.39) was significantly higher than the perceived truthfulness of causal assertions in the low external realism treatment (M = 4.45, SD = 1.54), F(1, 266) = 7.04, p < .01, 10, ηp2 = 0.03. There was no significant difference between the perceived truthfulness of noncausal assertions, F(1, 261) = 0.07, p = .80, 10, ηp2 < 0.01, or control assertions, F(1, 262) = 0.33, p = .56, 10, ηp2 < 0.01, between external realism treatments. In response to RQ2, the data suggests that the acceptance of assertions placed at causal locations is dependent on perceptions of external realism whereas other causal locations have no moderating dependence.

The influence of external realism on information acceptance is limited to causal locations within the narrative.
In contrast to external realism, the interaction of narrative realism and narrative causality on perceived truthfulness failed to reach significance, F(2, 252) = 0.56, p = .57, 10, ηp2 < 0.01, supporting the proposed meaningful differences between the two dimensions of realism as did the relevant three-way interaction, F(2, 252) = 1.07, p = .35, 10, ηp2 < 0.01.
Discussion
The purpose of this study was to investigate if the hypothesized interaction between narrative causality and a previously studied variable of narrative persuasion could serve as a pool of untapped variance to improve both the theory and practice of narrative persuasion. Using perceived realism as such a variable, the data suggests that narrative causality does meaningfully moderate the influence of external realism and accounting for this variance in future studies of narrative persuasion may help to explain some of the inconsistencies within the field.
Narratives that exhibited greater external realism resulted in greater information acceptance but only for information that had a causal effect on future narrative events. In other words, the effect of external realism only applied to a subset of the information contained within the narrative. This partially explains why the data found no main effect of external realism on information acceptance; ignoring the differences between causal locations reduced the combined effect below a level detectable by the study’s parameters. Likewise, the fact that causal information was the location most influenced by external realism aligns with the expectations of the narrative causality effect that assigns heavier importance to the influence of causal statements (Dahlstrom, 2010, 2012) and further supports the categorical difference between causal and noncausal narrative information.
In contrast to external realism, the dimension of narrative realism showed no interaction with narrative causality, supporting both the proposed categorization between perceived realism dimensions (Busselle & Bilandzic, 2008), but also suggesting that the “quality” or “cohesion” of a narrative plays a smaller role in information acceptance than the content choices, at least as defined by the manipulations of this study. On the other hand, it is important to note that the perceived difference between the high and low narrative realism manipulations was, while significantly different, relatively small. This raises two possibilities. First, the differences in perceived narrative realism between the stimuli may have not been large enough to either elicit or detect any relevant effects. Second, because the low narrative realism stimulus was constructed to contain glaring errors to hinder narrative reality, this lack of a perceived difference may represent an intrinsic resilience of narrative audiences to maintain narrative realism even in the face of errors. This assumption has some support in the literature (Appel & Richter, 2007; Marsh, et al., 2003) and may suggest that narrative realism should not be thought of as a linear function, but perhaps as a tipping point above which a narrative audience will try to maintain narrative realism at any cost and below which the coherence of the narrative becomes so poor that the narrative may become unintelligible. The “quality” of a narrative is often noted as a potentially important factor in narrative persuasion even though it is rarely operationalized (Green, 2006; Green & Brock, 2000; Kreuter, et al., 2007), so this lack of influence should be explored further.
While manipulation checks ensured that participants were influenced as expected, the measurement and manipulation of perceived realism remains poorly defined. It is possible that the measurement scales used did not capture the full extent of the constructs. Likewise, the sample population used in this study consisted of college students, which may be more open to the believability of information than more experienced adults, resulting in a more pronounced effect than may exist in the larger public.
This study focused on the acceptance of specific assertions within a narrative and is likely generalizable to similar contexts interested in belief-based outcomes of narrative persuasion. However, an alternative outcome of narrative persuasion is the alignment of normative attitudes with the moral of the narrative and the relationship between these two outcomes has not yet been clearly explored. While both outcomes are referred to as “narrative persuasion”, they represent different processes. A single narrative can contain many opportunities for belief-based effects and acceptance requires assessing the certainty of previously held beliefs to arrive at an evaluation of truth. In contrast, a single narrative often has only one moral constructed from its aggregate of events whereas its assessment requires aligning ideologies and related attitudes to arrive at an evaluation of good or bad.
This comparison mirrors the difference between first-order and second-order cultivation effects where the first level represents acceptance of information as true and the second represents attitudes about what to do about it (Nabi & Sullivan, 2001; Potter, 1991). It seems likely that the relationship between belief-based and attitude-based narrative persuasion effects may be similar. In such a case, assertion acceptance would be a prerequisite for attitudinal persuasion, but not a guarantee. A narrative using an aggregate of causal assertions that fails to persuade an individual to accept a particular normative attitude would not contradict the persuasive power of causal locations because it would be measuring a second-order outcome and not directly comparing causal and noncausal impacts on the first-order outcome.
Therefore, while belief-based outcomes will likely be directly affected by accounting for the variance of narrative causality, attitude-based outcomes will likely only be indirectly affected through influencing the acceptance of specific assertions that together support the implied moral. Future research should continue to explore this proposed relationship between belief-based and attitude-based narrative persuasion and the additional complicating factors inherent in the latter.
Regardless, the findings of this study can offer recommendations to communication practitioners on how to more effectively craft narrative messages to increase the potential for acceptance of target information within narrative communication. Specifically, target information should be integrated into the narrative so that it has a direct, causal effect on future narrative events for increased acceptance, and the selection of settings and characters that closely mirror the real world as experienced by the target audience may strengthen this effect.
Potentially more transformative is the fact that these findings also offer communication researchers the first empirical evidence that narrative causality represents a valuable pool of untapped variance for the more nuanced understanding of narrative persuasion as called for in the literature. Future studies should both explore the bounds of its influence and examine how accounting for the variance explained by narrative causality can enhance known relationships and existing theories within the field of narrative persuasion.
Footnotes
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
