Risks

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=rcqu20

Communication Quarterly

ISSN: 0146-3373 (Print) 1746-4102 (Online) Journal homepage: https://www.tandfonline.com/loi/rcqu20

On Measures of Message Elaboration in Narrative Communication

Lijiang Shen & Su-yeun Seung

To cite this article: Lijiang Shen & Su-yeun Seung (2018) On Measures of Message Elaboration in Narrative Communication, Communication Quarterly, 66:1, 79-95, DOI: 10.1080/01463373.2017.1334682

To link to this article: https://doi.org/10.1080/01463373.2017.1334682

Published online: 13 Jun 2017.

Submit your article to this journal

Article views: 908

View related articles

View Crossmark data

Citing articles: 3 View citing articles

On Measures of Message Elaboration in Narrative Communication Lijiang Shen & Su-yeun Seung

This study assesses and compares two forms of measurement instrument for message elaboration in narrative communication research: self-report vs. thought-listing. The valid- ity of the two forms was investigated in a nomological network consisting of need for cognition, message sensation value, identification, transportation, parasocial interaction, risk perception, and behavioral intention. Reliability was also examined. Results showed that the self-report form might be better when the focus was on entertainment-related information, while the thought-listing form might be more advantageous for persuasive information. Implications for narrative communication and future research were discussed.

Keywords: Construct Validity; Message Elaboration; Narrative Communication; Self-Report; Thought-Listing

The cognitive response tradition proposes that it is not the message per se but rather cognitive processing of the message that produces persuasion (Greenwald, 1968). Message elaboration is defined as the degree to which individuals cognitively process the message in such a way that they scrutinize the argument, weigh the evidence, analyze the logic, and assess the soundness of the claims presented. More or less rooted in the cognitive response tradition, several theories of persuasion concur that message elaboration plays an important role. The dual process models (Chaiken, Liberman, & Eagly, 1989; Petty & Cacioppo, 1986) suggest that the type and amount of message elaboration determine persuasion outcomes. Persuasion outcomes result- ing from high elaboration tend to be longer lasting, stronger, and more resistant to

Lijiang Shen (Ph.D., University of Wisconsin-Madison, 2005) is an associate professor in the Department of Communication Arts and Sciences at Pennsylvania State University. Su-yeun Seung is a graduate student in the Department of Communication Studies at the University of Georgia. Correspondence: Lijiang Shen, Depart- ment of Communication Arts & Sciences, Pennsylvania State University, 211 Sparks Building, University Park, PA 16802. E-mail: [email protected]

Communication Quarterly Vol. 66, No. 1, 2018, pp. 79–95

ISSN 0146-3373 print/1746-4102 online © 2017 Eastern Communication Association DOI: https://doi.org/10.1080/01463373.2017.1334682

counter persuasion (Barden & Petty, 2008; Petty, Haugtvedt, & Smith, 1995; Petty & Krosnick, 1995). On the other hand, persuasion out of low elaboration tends to be more flimsy, weaker, and less resistant to counter persuasion (Barden & Petty, 2008; Petty et al., 1995). The inoculation model suggests one type of message elaboration (i.e., counter-argument) is the source of resistance produced by the inoculation treatment (Compton, 2013; Pfau, 1997).

Message elaboration is also one of the key mechanisms that underlie the persuasive effect of narrative messages. Narratives can be defined as symbolic representation of cohesive and coherent events with an identifiable structure (Bilandzic & Busselle, 2013), which are bounded in space and time and contain implicit or explicit messages about the topics being addressed (Kreuter et al., 2007). Elaboration of narrative messages is enhanced by personal relevance and involvement as a result of various forms of narrative engagement (Bilandzic & Busselle, 2013). Different models of narrative communication, namely the extended-elaboration likelihood model (Slater & Rouner, 2002), the entertainment overcoming resistance model (Moyer‐Gusé, 2008; Moyer-Gusé & Nabi, 2010), and the transportation-imagery model (Green & Brock, 2000), concur on the importance of message elaboration, although they tend to have different conceptualizations (see Bilandzic & Busselle, 2013 for a review and discus- sion). Our review of the literature showed that existing empirical evidence had been rather mixed for the role of message elaboration. These inconsistencies in research findings might be attributed to the discrepancies in operationalization of the message elaboration construct in the models and the empirical studies as well.

In general, there are primarily three approaches to measuring message elaboration in persuasion research. The first approach is to infer from the association between attitude and argument quality (e.g., Bless, Mackie, & Schwartz, 1992; Wegener, Petty, & Klein, 1994) or from the effectiveness of different types of messages (e.g., Agrawal, Menon, & Aaker, 2007). Significant and substantial association between attitude and message quality/type indicates high message elaboration, while nonsignificant or near zero association suggests low or no message elaboration. The second approach is to use a form of self-report measure (e.g., Likert scale items) based on the definition of message elaboration (e.g., Moyer-Gusé & Nabi, 2010; Reynolds, 1997). The third approach uses the thought-listing technique (Cacioppo & Petty, 1981), and message elaboration is indicated by the number of evaluative thoughts (e.g., Escalas, 2004; Jones, Sinclair, & Courneya, 2003; Maheswaran & Meyers-Levy, 1990) or a thought- listing index (e.g., Das & Fennis, 2008). Such differences in measurement might have contributed to the inconsistent findings from individual studies (see Shen, 2013 for an example in research on incidental affect and message elaboration).

With form of measurement as a methodological factor, the mixed evidence for message elaboration in narrative communication might have been an artifact. Poten- tially, this issue can be resolved (a) if a meta-analysis can be conducted to test the moderating effect of measurement form, and/or (b) if researchers have knowledge of which measure might be better and shift toward using the better measure in future research. The primary goal of this article was to assess and compare how different measures of the message elaboration construct perform in narrative communication.

80 L. Shen & S.-Y. Seung

Given that argument strength is usually not a manipulated factor in narrative commu- nication research, the indirect measurement based on inference of the association between attitude and argument quality is less relevant. Similarly, recent studies in narrative communication have shifted away from comparing the impact of narrative vs. non- narrative messages to the underlying mechanisms of narrative effects. The perspective of inference based on the relative effectiveness of different message types is not so relevant either. Therefore, the comparison will focus on the other two forms: self-report measure vis-à-vis the thought-listing approach. First, criteria for measurement assessment and comparison will be discussed. Second, hypotheses involving external variables, including potential antecedents and outcomes of message elaboration, will be presented. These variables form a nomological network that enables measurement assessment and comparison. Third, empirical data will be presented to test these hypotheses, withmessage elaboration measured with different forms. Lastly, implications and directions for future narrative communication research will be discussed.

Criteria for Measurement Assessment and Comparison

Hunter and Gerbing (1982) proposed that measurement assessment and comparison should be based on content validity, construct validity, and reliability. Both the self- report and the thought-listing measures of message elaboration have enjoyed wide application in research, which suggests that scholars in the discipline agree that both measures exhibit good semantic correspondence with the construct and constitute reasonable sampling of the conceptual domain. That is, both forms are considered as high in content validity. Therefore, this article focuses on construct validity and reliability in scale assessment and comparison. To assess and compare the two different measures of the message elaboration, the construct will be situated in a nomological network that consists of its potential antecedents and outcome variables. Evidence in favor of a particular measurement form should come from its perfor- mance in hypothesis testing. For that purpose, we consider the following external variables: (a) need for cognition, which is a personality trait associated with message elaboration; (b) message sensation value, which is a message feature associated with elaboration; (c) narrative engagement, including identification, transportation, and parasocial interactions, which are the key mechanisms of narrative effects; and (d) persuasion outcomes, including risk perception and the association between risk perception and behavioral intention, which is considered as a function of message elaboration in the dual process models.

Need for Cognition

Need for cognition is “the tendency for an individual to engage in and enjoy thinking” (Cacioppo & Petty, 1982, p. 116). Individuals high in need for cognition tend to make more discriminating judgments and to be more motivated to think about message; consequently, they are more influenced by the argument quality than those low in

Communication Quarterly 81

need for cognition (e.g., Cohen, 1957; Cohen, Stotland, & Wolfe, 1955; Petty & Cacioppo, 1986; Sadowski & Gulgoz, 1996). The evidence for the impact of need for cognition on message elaboration has been well-documented in the dual process model literature. Therefore, it was predicted that:

H1: Need for cognition is positively associated with message elaboration.

Message Sensation Value

Message sensation value is “the degree to which formal and content audio-visual features of a televised message elicit sensory, affective and arousal responses” (Palm- green et al., 1991, p. 219). According to the activation model of exposure (Donohew, Palmgreen, & Duncan, 1980; Everette & Palmgreen, 1995), individuals would pay more attention to, and process in more depth, messages that provide enough and appropriate levels of arousal but turn away from, and process in less depth, messages that provide too much or too little stimulation. More recently, Niederdeppe and associates (Niederdeppe, 2005; Niederdeppe, Davis, Farrelly, & Yarsevich, 2007) found a positive main effect of message sensation value on message processing and recall, probably because the participants were young adolescents high in sensation seeking: 12–15 and 16–18 year olds in Niederdeppe (2005) and 12–17 year olds in Niederdeppe et al. (2007). Along this line of argument, when receivers (e.g., young adults) are high in sensation seeking, the following should hold:

H2: Message sensation value is positively associated with message elaboration.

Identification

Identification in the processing of narrative messages allows individuals to adopt the characters’ perspective, relate and connect with the character, and vicariously experience the emotions and events associated with the characters (Bilandzic & Busselle, 2013; Cohen, 2001). Once message recipients adopt the perspective of the media character through vicarious experiences and internalize the narrative, they tend to perceive the message as more relevant and related to their life (Cohen, 2001; Godlewski & Perse, 2010; Slater & Rouner, 2002). In other words, identification motivates message elaboration by increasing personal relevance and involvement (see Escalas, 2007). Therefore, it was predicted that:

H3: Identification is positively associated with message elaboration.

Transportation

Message recipients experience transportation when they converge with the story by decreasing their level of self-awareness and giving full attention to the narrative (Green & Brock, 2000), which can take place through a variety of mediums (Appel & Ricther, 2010; Green & Brock, 2000). Multiple models of narrative communication suggest that once

82 L. Shen & S.-Y. Seung

individuals are transported to the narrative messages, they comprehend the narrative as more connected to their lives (e.g., the transportation-imagery model, Green & Brock, 2000; the extended ELMmodel, Slater & Rouner, 2002; the entertainment overcoming resistance model, Moyer-Gusé, 2008; Moyer-Gusé &Nabi, 2010). Hence, they are more likely to focus on events of the story and put all mental capacities on the affairs currently taking place (Green & Brock, 2000). In addition, the narrative message can serve as exemplars (e.g., exemplification theory, Zillmann, 1999, 2002) which lead to more primary and concrete message processing and ultimately result in increased narrative-relatedmessage elaboration. Along this line of argument, it was predicted that:

H4: Transportation is positively associated with message elaboration.

Parasocial Interaction

Parasocial interaction is the interaction between narrative characters and message recei- vers (Moyer‐Gusé, 2008). Based on parasocial interaction, message recipients form a pseudo-relationship with characters in the narrative, which might have similar features of real social relationship (e.g., voluntary entry, companionship, and social attraction, Perse & Rubin, 1989) but is not reciprocal and might just be wishful (e.g., Hoffner & Buchanan, 2005). Given the focus on relational interaction, parasocial interaction should lead to more in-depth processing of the narrative message, at least to the entertainment compo- nent. Moreover, the affective, cognitive, and behavior processes involved in parasocial interaction can be intense (Schramm & Hartmann, 2008); hence, it should increase message elaboration. Along with this line of arguments, we predicted that:

H5: Parasocial interaction is positively associated with message elaboration.

Persuasion Outcomes

One distinctive feature of narrative messages is the combination of entertainment and education (i.e., persuasion) information; hence the intent to persuade is not explicit but rather disguised. In addition, the nature of narrative engagement suggests that cognitive responses generated in message elaboration (consciously or unconsciously) tend to be positive in valence. When the audience members identify with the narrative, their cognitive responses will be positive in that they tend to agree with and yield to the narrative, which reduces biased processing and decreases counterargument (Dal Cin, Zanna, & Fong, 2004; Escalas, 2004; Igartua, 2010). Once transported, individuals would perceive the narrative as more relevant and related to their own experiences/lives, which, according to the dual process models (Chaiken et al., 1989; Petty & Cacioppo, 1986), would increase narrative- related and positive responses. The social attraction, attachment, and affinity involved in parasocial interaction with characters make the message seem less authoritative, less con- trolling, and more acceptable (Moyer-Gusé & Nabi, 2010; Schiappa, Allen, & Gregg, 2007). In conclusion, the dominant cognitive responses to narrative messages should be positive, which increases persuasion. In the case of risk communication, the persuasion outcome

Communication Quarterly 83

would be increase in perceived personal risk. Also, according to the dual process models, message elaboration is one of the factors that makes attitudes and perceptions more persist over time,more resistant against counter-persuasion attempts, andmore influential on other forms of judgments and behaviors (Petty&Cacioppo, 1986; Petty et al., 1995). Therefore, we predicted that:

H6: Message elaboration is positively associated with risk perception. H7: The risk-behavioral intention relationship increases as message elaboration

increases.

Method

Stimuli Messages

The stimuli messages were four different video clips adopted from So and Nabi (2013). These clips were segments from Entourage, Sex and the City, and Grey’s Anatomy, and had been edited from the original aired version to specifically focus on the periods that characters were enacting or discussing sexual health. Each video was around nine minutes long.

Participants and Procedure

The participants were 374 undergraduate students enrolled in communication classes at the University of Georgia. Students participated in the study either to fulfill their course requirement or to earn extra credit for their class. The age of participants ranged from 18 to 27 years (M = 19.64, SD = 1.34), with 76.9% describing themselves asWhite/Caucasian, 9.4% as being of Asian descent, 3.0% as Hispanic descent, 8.9% as African descent, and 1.6% as other. Two participants failed to disclose their ethnicity. Fifty-four percent reported their sex as female and 46% as male. There were slight variations in the actual sample sizes used in data analyses due to missing responses.

The participants were randomly assigned to watch one of the four videos. First, participants signed and dated consent forms before they started to watch the clip. After watching the video clip, they completed the thought-listing task and then completed the rest of the questionnaire that contained the measurement instruments. They were debriefed about the study, and their questions were answered if they had any. This study was part of a larger project. Not all data were reported here. See So and Shen (2016) for other findings in the project.

Measures

Need for Cognition Need for cognition was assessed using Cacioppo and Petty’s (1982) 18 items on a 5- point Likert scale (1 = Extremely Uncharacteristic to 5 = Extremely Characteristic). This scale included questions such as, “I would prefer complex to simple problems”

84 L. Shen & S.-Y. Seung

and “I usually end up deliberating about issues even when they do not affect me personally.” Responses to these items were average into the composite score (M = 3.28, SD = 0.58, α = 0.87).

Message Sensation Value Message sensation value was measured by using Everett and Palmgreen’s (1995) 11 dichotomous items (1 = Yes and 0 = No). We asked participants whether the PSA has each feature or not, such as “Unexpected format” and “Background sound/sound saturation.” These items were averaged into an index for message sensation value such that higher scores indicated high in message sensation (M = 0.47, SD = 0.17).

Identification Identification was assessed by using Cohen’s (2001) Identification Scale. Ten items were used to measure identification using a 5-point Likert scale (1 = Strongly Disagree, 5 = Strongly Agree). This scale included questions such as, “I tend to understand the reasons why the characters felt the way they did” and “When the characters suc- ceeded, I felt joy; when they failed, I felt sad.” The items were averaged into a composite score (M = 3.15, SD = 0.72, α = 0.83).

Transportation Transportation was assessed by Green and Brock’s (2000) scale. Participants responded to nine items using a 5-point Likert scale (1 = Strongly Disagree, 5 = Strongly Agree). For example, individuals answered questions such as, “While I was watching the clip, I could picture myself in the scene of the clip,” and “While I was watching the video, activity around me was on my mind.” Responses to these items were average into the composite score (M = 3.29, SD = 0.59, α = 0.69).

Parasocial Interaction Parasocial interaction was assessed using nine items adapted from Schramm and Hartmann (2008) on a 5-point Likert scale (1 = Strongly Disagree, 5 = Strongly Agree), which distinguishes (a) perceptual-cognitive, (b) affective, and (c) behavioral responses. The perceptual-cognitive response comprises processes such as the persona perception, persona evaluation, activation of memories and own life experiences, or social comparisons between the persona and oneself. Examples of questions included, “Occasionally, I wondered if the characters were similar to me or not?” and “I kept wondering if I know persons that are similar to the characters?” The affective response, relates to positive and negative feelings towards the persona, as well as to emotions that are evoked by the persona. Sample questions included, “Sometimes I really loved the character for what s/he did” and “If the character felt bad, I felt bad as well.” The behavioral response covers users’ nonverbal behavior (mimics and gestures) and verbal and para-verbal behavior (e.g., groaning), as well as behavioral intentions. Examples of

Communication Quarterly 85

questions included, “Sometimes I felt like speaking out” and “Occasionally, I said something on impulse” (M = 3.15, SD = 0.68, α = 0.77).

Self-Report Scale of Message Elaboration Message elaboration was assessed by using Reynold’s (1997) scale. It was measured by twelve 5-point Likert scale items (1 = Strongly Disagree, 5 = Strongly Agree). Example items included: “While watching the video I attempted to analyze the issues in the message” and “While watching the video I was deep in thought about the message.” Confirmatory factor analysis showed that the scale was unidimensional. These items were averaged into an index for message elaboration such that higher scores indicated more elaboration (M = 3.38, SD = 0.65, α = 0.86).

Thought-Listing Measure of Message Elaboration The thought-listing approach measures message elaboration as the sum of evaluative thoughts, which depends on coded cognitive responses. Four coders, who were unaware of the hypotheses, received about nine hours of training before they started coding randomly selected 20% of the data. Reliabilities were checked for this portion of the data only.

Coding procedures took place in four steps. First, the coders divided the data into psychological thought units. Guetzkow’s U was around 0.02. Second, coders figured out whether the units are cognitive or affective. When coders coded for affective units, they used the list of emotional words of Shaver, Schwartz, Kirson, and O’Connor (1987). Whenever those words appeared, they coded that unit as affective. The intercoder reliability for this step was Krippendorff’s alpha = 0.78. Third, relevance/irrelevance of cognitive thoughts was coded. The purpose of this step was to screen irrelevant cognitions and thereby reduce the level of noise in the data. Thought units that were related to message (i.e., content or delivery), addressing particular health problem, or the message advocacy were coded as relevant thoughts, while irrelevant thoughts were unrelated to the message or task. For example, “I’m hungry” was coded as irrelevant. Krippendorff’s alpha = 0.73 for this step. Fourth, the valence of cognitive thought units was also coded either as positive, negative, or neutral thoughts. Positive thoughts were defined as responses expressing liking toward the message, agreement with message or message source, approval intention of wearing a condom or getting tested, disapproval of the risky sex behavior, approval with number of sex partners, or if participants have some impact from messages, etc. On the other hand, negative thoughts were coded when the units are about disliking toward the message, disagreement with message or message source, perceived message as irrelevance, derogation of the source, lack of intention of wearing a condom or getting tested, approval of risky sex behavior, disapproval with number of sex partners, etc. Neutral thoughts were defined as non- evaluative responses to the message. Krippendorff’s alpha was 0.74 for this step. A message elaboration index was created by taking the sum of positive thoughts and negative thoughts. Note that the valence cognitive responses were all censored in

86 L. Shen & S.-Y. Seung

distribution: Positive thoughts had 35.1% of the cases clustered at 0 (M = 1.28, SD = 1.28), and 76.1% of negative thoughts (M = 0.36, SD = 0.79) were clustered at 0.

Self-Risk Perception Self-risk perception was measured by a single question that asked respondents to provide a percentage estimate ranging from 0% (No risk at all) to 100% (Extremely high risk), with 10% increments (M = 20.88%, SD = 20.94%). The variable was divided by 10 to be at 0–10 scale to reduce heteroscedasticity.

Behavioral Intention Behavioral intention to get tested from STDs within the next six months was measured by a single question that asked participants to provide a percentage estimate ranging from 0% (Certain that I will not) to 100% (Certain that I will), with10% increments (M = 19.21%, SD = 22.00%). The variable was divided by 10 to be at 0–10 scale to reduce heteroscedasticity.

Controlled Covariates Several variables were also measured as controlled variables in addition to the demo- graphic variables. The participants were asked to report (a) if they had been sexually active in the past year; if yes, (b) how many sex partners they had.

Results

Recall that construct validity and reliability are used as criteria for measurement assessment and comparison. To assess and compare the construct validity of the two forms of elaboration message, namely self-report measure and thought-listing approach, each of them were subjected to hypothesis testing within a nomological network of external variables. Evidence in favor of a measurement form should come from its construct validity. It should be noted that the two forms of elaboration measure were positively and significantly correlated (r = 0.16, p = 0.004). Due to the censored distribution of cognitive response, we also examined the association via Tobit regression (regressing self-report onto thought-listing). The Tobit coefficient was substantially larger: β = 0.33, p = 0.01.

Hypothesis Testing and Construct Validity

Hypotheses 1–6 were concerned with bivariate correlations between message elabora- tion and each of the external variables. They served to assess and compare the convergent validity of each measurement form. Partial correlations with the external variables were estimated for each form of elaboration measure, with the three dummy variables for narrative messages, age, gender, sexually active and number of sex partners as controlled covariates. The first column of Table 1 presents the partial

Communication Quarterly 87

correlations between the self-report measure and each of the external variables. The second column reports those for the thought-listing measure. Given the censored nature of the thought-listing measure, the partial correlation coefficients tended to be under-estimates (see Pan, Shen, Paek, & Sun, 2006; Shen & Dillard, 2009). A set to Tobit regression models were estimated, using each of the external variables to predict the thought-listing elaboration measure, with the same covariates controlled for. The third column in Table 1 shows the Tobit regression coefficients. Note that these coefficients were unstandardized. Hence, their significance was meaningful, but their magnitude can’t be directly compared to the standardized partial correlations. With these parameters, assuming α = 0.05, two tailed, and n = 374, the statistical power to detect an effect size equivalent to r = 0.20 exceeded 0.97 but below 0.50 to detect an effect size equivalent to r = 0.10.

When the self-report measure was used, the partial correlation between message elaboration and need for cognition was positive but did not reach significance: rpartial = 0.09, p = 0.08; hence H1 did not receive support. That for message sensation value was positive and significant: rpartial = 0.14, p = 0.01; H2 received support. That for identification was positive and significant: rpartial = 0.36, p < 0.001; H3 received support. That for transportation was positive and significant: rpartial = 0.53, p < 0.001; H4 received support. That for parasocial interaction was positive and significant: rpartial = 0.41, p < 0.001; H5 received support. The partial correlation was positive but nonsignificant for risk perception: rpartial= 0.07, p = 0.21; therefore, H6 did not receive support.

When the thought-listing measure was used, the partial correlation between mes- sage elaboration and need for cognition was positive and nonsignificant: rpartial = 0.10, p = 0.07; but when the potential under-estimation was corrected in Tobit regression, the coefficient was positive and significant: β = 0.66, p = 0.04. Hence H1 was considered as supported. The partial correlation for message sensation value was negative and nonsignificant: rpartial = –0.07, p = 0.19. The Tobit regression coefficient was negative and nonsignificant as well: β = –0.68, p = 0.24. H2 was not supported. The partial correlation was positive but nonsignificant for identification (rpartial = 0.04, p = 0.42), transportation (rpartial = 0.07, p = 0.19), and parasocial interaction (rpartial = 0.04, p = 0.46). The corresponding Tobit coefficients were also positive but

Table 1 Partial Correlations Between Elaboration and External Variables (n = 374)

Self-report Thought-listing Thought-listing Tobit coefficient

1. Need for Cognition 0.09 0.10 0.66*

2. Message Sensation Value 0.14** −0.09 −0.68

3. Identification 0.36*** 0.03 0.08

4. Transportation 0.53*** 0.06 0.16

5. Parasocial Interaction 0.41*** 0.03 0.06

6. Risk Perception 0.07 0.13** 0.13**

Note. *p < 0.05, **p < 0.01, ***p < 0.001.

88 L. Shen & S.-Y. Seung

nonsignificant. Therefore, H3–H5 did not receive support. The partial correlation was positive and significant for risk perception: rpartial = 0.13, p = 0.01. The corresponding Tobit coefficient was positive and significant: β = 0.13, p = 0.004. Hence, H6 was supported. Combined, four out of the six hypotheses were supported when the self- report measure was adopted. On the other hand, two out of the six hypotheses were supported when the thought-listing measure was adopted. Therefore, the self-report measure seemed to have better convergent validity.

H7 predicted that risk perception would predict behavioral intention better when elaboration is high than when elaboration is low. This hypothesis served to assess and compare divergent validity of the two message elaboration measures. To test this hypothesis, simple slope analyses were performed following Aiken and West (1991). With the dummy variables for narrative messages, age, gender, sexually active, and number of sex partners as controlled variables, behavioral intention was regressed onto risk perception. The same regression model was estimated when message elaboration was high (i.e., above the mean) vs. when message elaboration was low (i.e., below the mean) across the two forms of measurement, respectively. For the self- report measure, the regression coefficient was β = 0.40, p < 0.001, n = 188 when elaboration was high (i.e., above the mean), and β = 0.29, p < .001, n = 168 when elaboration was low (i.e., below the mean). For the thought-listing measure, the partial correlation between risk perception and behavioral intention was β = 0.38, p < 0.001, n = 183 when elaboration was high, and β = 0.34, p < 0.001, n = 172 when elaboration was low. Z-tests following the procedure in Cohen and Cohen (1983, pp. 53–55) showed that neither difference reached statistical significance. However, given how message elaboration was operationalized in this study, this was still consistent with the dual process models. High versus low message elaboration, as operationalized in this study, were located on the same continuum of message elaboration, that is, they belonged to the same mode of processing (central route in ELM, or systematic processing in HSM). In other words, the high and low message elaboration as measured in this study did not qualify as qualitatively different modes of message processing as defined in ELM or HSM. We speculate that if high and low elaboration had been qualitatively different, the partial correlations might be significantly differ- ent. These results showed that H7 received support in both forms of elaboration measure. In other words, both self-report and thought-listing forms of the elaboration measure demonstrated divergent validity.

Reliability

Reliability is another important criterion in measurement assessment. Since the two forms of elaboration measures were operationalized in different procedures, their reliabilities were not directly comparable. The scale reliability of the self-report scale was Cronbach’s α = 0.86. The item-total correlations ranged from 0.41 to 0.62, which demonstrated good internal consistency. The thought-listing measure was created based on results from a four-step coding process that involved four coders. The

Communication Quarterly 89

Krippendorff’s alpha for these steps ranged from 0.73 to 0.78. Combined with the coding scheme, the coders became the measurement in the case of the thought-listing measure. The four coders in this study were pretty consistent and comparable across the four steps of coding.

Discussion

There have been inconsistencies between theories and models of narrative commu- nication and empirical data regarding the role of message elaboration as an underlying mechanism of narrative effects. Discrepancies in operationalization of the message elaboration constructs were considered as one plausible explanation for such mixed findings. This article set out to compare the two forms of message elaboration measure that are most widely applied in narrative communication research. The two forms of measures were assessed and compared in terms of their content validity, convergent validity, divergent validity, and reliability. While both forms were con- sidered as acceptable in terms of content validity and divergent validity, the assess- ment of convergent validity and reliability might have important implications for narrative communication research.

Convergent Validity

Based on the number of hypotheses supported, the self-report measure seemed to have fared better than the thought-listing measure. Further analyses of the pattern in hypotheses testing revealed a pattern such that there was no overlap among the external variables that produced favorable results for each measure. The two external variables that yielded support for the thought-listing measure, namely need for cognition and risk perception (i.e., persuasion), tended to focus on the education information (i.e., persuasive arguments and advocacies) rather than entertainment information in the narratives. Individuals high in need for cognition, by definition, are more motivated to evaluate arguments, logic, and make more evaluative judgments. This fitted better with the coding scheme used in this study to generate the thought- listing measure of elaboration, which was more persuasion oriented. Along with the same logic, elaboration measure out of this approach was likely to affect the valence of dominant cognition, hence persuasion outcomes.

The four external variables that favored the self-report measures tended to focus on the narrative as a whole, maybe leaning more toward the entertainment information in the narratives. Features that increase message sensation values are video production features, which are attention grabbing but could constitute as cues that distract receivers from processing the risk/persuasion information (e.g., Kang, Cappella, & Fishbein, 2006). Identification, transportation, and parasocial interaction all focus on the events that occurred to the characters. Elaboration on such information facilitates individuals’ constructing mental models to comprehend and represent the characters, situations, and events portrayed in the narrative (Green & Brock, 2000; Slater &

90 L. Shen & S.-Y. Seung

Rouner, 2002) and to produce coherent meaning. Such elaboration induced by narrative engagement does not tend to be evaluative but rather neutral instead. This is probably because allocation of cognitive capacity in processing of narratives tends to be prioritized for the entertainment information, especially when the narrative mes- sages were primarily entertainment related.

Given the inherent education-entertainment dialectic in narrative messages, the per- formance of either form of elaboration measures in hypothesis testingmight be a function of the entertainment vs. education ratio and integration (Bilandzic & Busselle, 2013). When there is more entertainment information than education information (e.g., the narratives used in this study), the self-report measure might fare better; when there is more education information than entertainment information, the thought-listing mea- sure might excel. The two forms might perform equally well if the entertainment and education information is very well integrated in the narrative messages. Due to the almost complete lack of conceptual overlap in the external variables that favored each form of measures in the current study, we recommend that it might be premature and inap- propriate to favor either one in future research on narrative communication. Such different performances from the two forms of elaboration measures have been documen- ted elsewhere. For example, Escalas (2007) found that the transportation reduced elabora- tion, in that strong vs. weak arguments produced no difference in persuasion when transportation was high. On the other hand, Krakowiak and Oliver (2012) found trans- portation increased elaboration in terms of cognitive enjoyment. Scholars have suggested that it might be fruitful to separate these two types of message elaboration that target education and entertainment information respectively and have both measures in the same study (e.g., Van Laer, Ruyter, Visconti, &Wetzels, 2013). In their meta-analysis, Van Laer et al. (2013) actually found a positive impact from narrative engagement on narrative-related elaboration but a negative impact on arguments-related elaboration. Our results were in a similar pattern, and we echo their suggestion. Another option might be to specify the targets (story vs. arguments) when the self-report measure is used and the target of cognitive response to be coded when the thought-listing approach is adopted.

Reliability

At least in the current study, both forms of elaboration measure demonstrated good reliabilities. However, the two forms also have different implications. The reliability of the self-report scale was assessed with Cronbach’s alpha, which is a function of inter- item correlation and number of items in the scale. This means that better reliabilities can be achieved through refinement of the items and/or the scale as a whole. On the other hand, the number of items can be reduced without affecting the scale reliability, which can be more cost-effective, reduce the length of data collection, and avoid participants’ fatigue. The thought-listing measure was created using the results from a four-step coding process and the intercoder reliabilities were assessed with Krippen- dorff’s alpha. Intercoder reliability is a function of quality of the coding scheme and the rigor in coder training. Enhancement in reliability depends on improvement of the

Communication Quarterly 91

coding scheme in addition to better coder training, which might be more costly and cumbersome as compared to the case for the self-report scale.

Limitations and Directions for Future Study

The findings and suggestions in the current study would be interpreted with its limitations in mind. The first limitation lies in the limited external validity due to two reasons. One reason was the use of college student sample. Given their life stage, college students might be unique regarding their media use and sexual activity, as well as risk perception. Hence, their elaboration of the video clips from popular TV shows that portray adults (and mostly at work setting) could have been very different than adults. The other reason was that only four video clips on the same topic (STDs) were used. Stronger evidence with better generalizability should come from future studies which utilize more video narratives on a variety of topics and with samples from the general public. A related issue in generalizability is that conclusions only apply to processing of narrative messages, rather than processing of all types of persuasive messages. The second limitation lies in the relatively low statistical power. It should be noted that the effects observed were rather small and evidence not strong. Given the observed effect sizes, the sample size in the current investigation (n = 374) did not provide strong power, which might create some uncertainties in the results. The third limitation lies in the invariability in the ratio of education vs. entertainment informa- tion, which means that our discussion remained speculative on advantages of the self- report measure on entertainment information and those of the thought-listing mea- sure when it comes to education information. To empirically test these possibilities, it would be imperative to have different types of narrative messages that vary in the proportion of education vs. entertainment information.

Conclusion

The current article strives to assess and compare two different forms of message elaboration measurement, the self-report approach and the thought-listing approach. Head-to-head comparisons were made between the two forms in terms of content validity, construct validity, and reliability. Results showed that the self-report measure was more appropriate when it comes to processing of non-educational information, while the thought-listing measure might be more advantageous when it comes to persuasive information presented in the narrative messages. It was recommended that it might be fruitful to include both forms or to specify and differentiate the target of message elaboration when either one form is used in future studies.

Acknowledgments

We thank Dr. Jiyeon So and Dr. Robin Nabi for their generosity in sharing the stimuli videos used in this study.

92 L. Shen & S.-Y. Seung

References

Agrawal, N., Menon, G., & Aaker, J. L. (2007). Getting emotional about health. Journal of Marketing Research, 44, 100–113. doi:10.1509/jmkr.44.1.100

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Thousand Oaks, CA: Sage.

Appel, M., & Richter, T. (2010). Transportation and need for affect in narrative persuasion: A mediated moderation model. Media Psychology, 13, 101–135. doi:10.1080/15213261003799847

Barden, J., & Petty, R. E. (2008). The mere perception of elaboration creates attitude certainty: Exploring the thoughtfulness heuristic. Journal of Personality and Social Psychology, 95, 489– 509. doi:10.1037/a0012559

Bilandzic, H., & Busselle, R. (2013). Narrative persuasion. In J. P. Dillard & L. Shen (Eds.), The Sage handbook of persuasion: Developments in theory and practice (2nd ed., pp. 200–219). Thou- sand Oaks, CA: Sage.

Bless, H., Mackie, D. H., & Schwartz, N. (1992). Mood effects on attitude judgments: Independent effects of mood before and after message elaboration. Journal of Personality and Social Psychology, 63, 585–595. doi:10.1080/08838159609364370

Cacioppo, J. T., & Petty, R. E. (1981). Social psychological procedures for cognitive response assessment: The thought-listing technique. In T. V. Merluzzi, C. R. Glass, & M. Genest (Eds.), Cognitive assessment (pp. 309–342). New York, NY: Guilford.

Cacioppo, J. T., & Petty, R. E. (1982). The need for cognition. Journal of Personality and Social Psychology, 42, 116–131. doi:10.1037/0022-3514.42.1.116

Chaiken, S., Liberman, A., & Eagly, A. H. (1989). Heuristic and systematic processing within and beyond the persuasion context. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp. 212–252). New York, NY: Guilford Press.

Cohen, A. R. (1957). Need for cognition and order of communication as determinants of opinion change. In C. I. Hovland (Ed.), The order of presentation in persuasion. New Haven, CT: Yale University Press.

Cohen, A. R., Stotland, E., & Wolfe, D. M. (1955). An experimental investigation of need for cognition. The Journal of Abnormal and Social Psychology, 51, 291–294. doi:10.1037/h0042761

Cohen, J. (2001). Defining identification: A theoretical look at the identification of audiences with media characters. Mass Communication & Society, 4, 245–264. doi:10.1207/ S15327825MCS0403_01

Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum.

Compton, J. (2013). Inoculation theory. In. In J. P. Dillard & L. Shen (Eds.), The handbook of persuasion: Developments in theory and practice (2nd ed., pp. 220–236). Thousand Oaks, CA: SAGE.

Dal Cin, S., Zanna, M. P., & Fong, G. T. (2004). Narrative persuasion and overcoming resistance. In E. S. Knowles & J. A. Linn (Eds.), Resistance and persuasion (pp. 175–191). Mahwah, NJ: Lawrence Erlbaum Associates.

Das, E., & Fennis, B. M. (2008). In the mood to face the facts: When a positive mood promotes systematic processing of self-threatening information. Motivation and Emotion, 32, 221–230. doi:10.1007/s11031-008-9093-1

Donohew, L., Palmgreen, P., & Duncan, J. (1980). An activation model of information exposure. Communication Monographs, 47, 295–303. doi:10.1080/03637758009376038

Escalas, J. E. (2004). Narrative processing: Building consumer connections to brands. Journal of Consumer Psychology, 14, 168–180. doi:10.1207/s15327663jcp1401&2_19

Escalas, J. E. (2007). Self-referencing and persuasion: Narrative transportation versus analytical elaboration. Journal of Consumer Research, 33, 421–429. doi:10.1086/502810

Everett, M. W., & Palmgreen, P. (1995). Influences of sensation seeking, message sensation value, and program context of effectiveness of anticocaine public service announcements. Health Communication, 7, 225–248. doi:10.1207/s15327027hc0703_3

Communication Quarterly 93

Godlewski, L. R., & Perse, E. M. (2010). Audience activity and reality television: Identification, online activity, and satisfaction. Communication Quarterly, 58, 148–169. doi:10.1080/ 01463371003773358

Green, M. C., & Brock, T. C. (2000). The role of transportation in the persuasiveness of public narratives. Journal of Personality and Social Psychology, 79, 701–721. doi:10.1037/0022-3514.79.5.701

Greenwald, A. G. (1968). Cognitive learning, cognitive response to persuasion, and attitude change. In A. G. Greenwald, T. C. Brock, & T. M. Ostrom (Eds.), Psychological foundations of attitudes (pp. 147–170). New York, NY: Academic Press.

Hoffner, C., & Buchanan, M. (2005). Young adults’ wishful identification with television characters: The role of perceived similarity and character attributes. Media Psychology, 7, 325–351. doi:10.1207/S1532785XMEP0704_2

Hunter, J. E., & Gerbing, D. W. (1982). Unidimensional measurement, second order factor analysis, and causal models. In B. M. Staw & L. L. Cummings (Eds.), Research in organizational behavior (Vol. 4). Greenwich, CT: JAI Press.

Igartua, J. (2010). Identification with characters and narrative persuasion through fictional feature films. Communications-European Journal of Communication Research, 35, 347–373. doi:10.1515/comm.2010.019

Jones, L. W., Sinclair, R. C., & Courneya, K. S. (2003). The effects of source credibility and message framing on exercise intentions, behaviors, and attitudes: An integration of the elaboration likelihood model and prospect theory. Journal of Applied Social Psychology, 33, 179–196. doi:10.1111/j.1559-1816.2003.tb02078.x

Kang, Y., Cappella, J., & Fishbein, M. (2006). The attentional mechanism of message sensation value: Interaction between message sensation value and argument quality on message effectiveness. Communication Monographs, 73, 351–378. doi:10.1080/03637750601024164

Krakowiak, K. M., & Oliver, M. B. (2012). When good characters do bad things: Examining the effect of moral ambiguity on enjoyment. Journal of Communication, 62, 117–135. doi:10.1111/ j.1460-2466.2011.01618.x

Kreuter, M. W., Green, M. C., Cappella, J. N., Slater, M. D., Wise, M. E., Storey, D., … Wooley, S. (2007). Narrative communication in cancer prevention and control: A framework to guide research and application. Annals of Behavioral Medicine, 33, 221–235. doi:10.1007/ BF02879904

Maheswaran, D., & Meyers-Levy, J. (1990). The influence of message framing and issue involvement. Journal of Marketing Research, 27, 361–367. doi:10.2307/3172593

Moyer‐Gusé, E. (2008). Toward a theory of entertainment persuasion: Explaining the persuasive effects of entertainment‐education messages. Communication Theory, 18, 407–425. doi:10.1111/j.1468-2885.2008.00328.x

Moyer-Gusé, E., & Nabi, R. L. (2010). Explaining the persuasive effects of narrative in an entertain- ment television program: Overcoming resistance to persuasion. Human Communication Research, 36, 25–51. doi:10.1111/j.1468-2958.2009.01367.x

Niederdeppe, J. (2005). Syntactic indeterminacy, perceived message sensation value-enhancing features, and message processing in the context of anti-tobacco advertisements. Communica- tion Monographs, 72, 324–344. doi:10.1080/0365=37750500206862

Niederdeppe, J., Davis, K. C., Farrelly, M. C., & Yarsevich, J. (2007). Stylistic features, need for sensation, and confirmed recall of national smoking prevention advertisements. Journal of Communication, 57, 272–292. doi:10.1111/j.1460-2466.2007.00343.x

Palmgreen, P., Donohew, L., Lorch, E. P., Rogus, M., Helm, D., & Grant, N. (1991). Sensation seeking, message sensation value, and drug use as mediators of PSA effectiveness. Health Communication, 3, 217–227. doi:10.1207/s15327027hc0304_4

Pan, Z., Shen, L., Paek, H., & Sun, Y. (2006). Mobilizing political talk in the 2000 presidential campaign—An examination of campaign effects in a deliberative framework. Communication Research, 33, 1–31. doi:10.1177/0093650206291478

94 L. Shen & S.-Y. Seung

Perse, E. M., & Rubin, R. B. (1989). Attribution in social and parasocial relationships. Communica- tion Research, 16, 59–77. doi:10.1177/009365089016001003

Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persuasion (pp. 1–24). New York, NY: Springer.

Petty, R. E., Haugtvedt, C. P., & Smith, S. M. (1995). Elaboration as a determinant of attitude strength: Creating attitudes that are persistent, resistant, and predictive of behavior. In R. E. Petty & J. A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 93–130). Mahwah, NJ: Erlbaum.

Petty, R. E., & Krosnick, J. A. (Eds.). (1995). Attitude strength: Antecedents and consequences. Mahwah, NJ: Erlbaum.

Pfau, M. (1997). Inoculation model of resistance to influence. In G. A. Barnett & F. J. Boster (Eds.), Progress in communication sciences: Advances in persuasion (Vol. 13, pp. 133–171). Green- wich, CT: Ablex.

Reynolds, R. A. (1997). A validation test of a message elaboration measure. Communication Research Reports, 14, 269–278. doi:10.1080/08824099709388670

Sadowski, C. J., & Gulgoz, S. (1996). Elaborative processing mediates the relationship between need for cognition and academic. Journal of Psychology, 130, 303–307. doi:10.1080/ 00223980.1996.9915011

Schiappa, E., Allen, M., & Gregg, P. B. (2007). Parasocial relationships and television: A meta- analysis of the effects. In R. W. Preiss, B. M. Gayle, N. Burrell, M. Allen, & J. Bryant (Eds.), Mass media effects research: Advances through meta-analysis (pp. 301–314). Mahwah, NJ: Lawrence Erlbaum Associates.

Schramm, H., & Hartmann, T. (2008). The PSI-process scales: A new measure to assess the intensity and breadth of parasocial processes. Communications-European Journal of Communication Research, 33, 385–401. doi:10.1515/COMM.2008.025,

Shaver, P., Schwartz, J., Kirson, D., & O’connor, C. (1987). Emotion knowledge: Further exploration of a prototype approach. Journal of Personality and Social Psychology, 52, 1061–1086. doi:10.1037/0022-3514.52.6.1061

Shen, L. (2013). Incidental affect and message processing: Revisiting the competing hypotheses. Communication Studies, 64, 337–352. doi:10.1080/10510974.2012.755642

Shen, L., & Dillard, J. P. (2009). Message frames interact with motivational systems to determine depth of message processing. Health Communication, 24, 504–514. doi:10.1080/10410230903104897

Slater, M. D., & Rouner, D. (2002). Entertainment-education and elaboration likelihood: Under- standing the processing of narrative persuasion. Communication Theory, 12, 173–191. doi:10.1111/j.1468-2885.2002.tb00265.x

So, J., & Nabi, R. (2013). Reduction of perceived social distance as an exploration for media’s influence on personal risk perceptions: A test of the risk convergence model. Human Com- munication Research, 39, 317–338. doi:10.1111/hcre.12005

So, J., & Shen, L. (2016). Personalization of risk through convergence of self- and character-risk: Narrative effects on social distance and self-character risk perception gap. Communication Research, 43, 1094–1115. doi:10.1177/0093650215570656

Van Laer, T., Ruyter, K. D., Visconti, L. M., & Wetzels, M. (2013). The extended transportation- imagery model: A meta-analysis of the antecedents and consequences of consumers’ narrative transportation. Journal of Consumer Research, 40, 797–817. doi:10.1086/673383

Wegener, D. T., Petty, R. E., & Klein, D. J. (1994). Effects of mood on high elaboration attitude change: The mediating role of likelihood judgments. European Journal of Social Psychology, 24, 25–43. doi:10.1002/ejsp.2420240103

Zillmann, D. (1999). Exemplification theory: Judging the whole by some of its parts. Media Psychology, 1, 69–94. doi:10.1207/s1532785xmep0101_5

Zillmann, D. (2002). Exemplification theory of media influence. In J. Bryant & D. Zillmann (Eds.), Media effects: Advances in theory and research (2nd ed., pp. 19–41). Mahwah, NJ: Erlbaum.

Communication Quarterly 95