Understanding Narrative Effects: The Role of Discrete Negative Emotions on Message Processing and Attitudes Among Low-Income African American Women

Abstract

This study tests the processes through which breast cancer narrative messages are effective by taking a functional approach. We explore how discrete negative emotions (i.e., sadness, fear, and anger) induced by breast cancer survivor stories affect African American women’s message processing, recall of message content, and attitudinal outcomes. Structural equation modeling was performed for narrative and informational versions of a breast cancer screening video shown to 489 low-income African American women ages 40 years and older. The model was well fitted. Sadness enhanced the persuasive process, while fear inhibited it. Sadness also helped participants recall more message-relevant content, while fear inhibited recall. Anger was not related to the persuasive process. Implications of these findings for narrative research and application are discussed.

ACKNOWLEDGMENT

This research was funded by a grant from the National Cancer Institute’s Centers of Excellence in Cancer Communication Research (CECCR) program (CA-P50-95815; PI: Kreuter).

Notes

1. ¹Narrative messages have been called base rate messages (Baesler & Burgoon, 1984) or example evidence (Allen & Preiss, 1997), while informational messages have also been called statistical evidence messages (Allen & Preiss, 1997) and logical argument messages (Kopfman et al, 1998).

2. ²Dillard and Nabi (2006) claimed that cancer-related messages have the potential to arouse various negative emotions as well as various positive emotions. In this article, we specifically focus only on negative emotions because of the following reasons. First, it is common that cancer-related messages predominantly elicit negative emotions, including fear, sadness, and anger in particular. Second, unlike consumer marketing campaigns where the primary goal is to elicit positive feeling to a product, public messages often elicit negative emotions because these messages are purported to highlight the negative consequences of risky behaviors and to promote healthy behaviors (Dunlop et al., 2008). Negative emotions are generally known to be more capable of motivating individuals to perform desirable behavioral outcomes than positive outcomes (Bagozzi & Moore, 1994). We also focus on unintentionally aroused negative emotions in this study that these messages were not designed to elicit: specific emotions such as fear appeals or anger appeals.

3. ³We used message type as a dummy variable (narrative = 1).

4. ⁴The potential reason why these indicators (i.e., sad, concerned, worried) clustered as one factor (i.e., sadness) is due to the high correlations among sadness, sympathy, and personal distress (Eisenberg et al., 1991). According to Eisenberg and colleagues (1991), the adjective “concerned” was used to measure one’s sympathy level while “worried” was used to measure one’s personal distress level. It should be noted that sympathy and personal distress are similar in that these both originate from sadness; however, if sympathy is other-oriented concern or sorrow, then personal distress is self-oriented discomfort. To elaborate, when a person feels sad for another person, it increases the likelihood of sympathy, which is known as an affective response of sorrow or concern for the troubled or needy person (Eisenberg, 2000; Eisenberg & Miller, 1987). In contrast to sympathy, when a person becomes sad for him- or herself after reading a sad message, it sometimes causes “personal distress” (Eisenberg, 2000). Personal distress also stems from empathy, but it refers to the unpleasant feelings of personal anxiety and worry regarding one’s own well-being (Eisenberg & Miller, 1987). For the analysis purpose, we use all three indicators as the factor “sadness.”

5. ⁵We also examined which type of message would predict message recall by conducting univariate ANOVA. Participants who read the narrative message (M = 3.84, SD = .56) significantly had a higher score on message recall (i.e., closed-ended format) than participants who read the informational message (M = 2.90, SD = .51), F(1, 408) = 317.33, p < .001. For the specific contents of the message, the logistic regression analysis revealed that participants who read the narrative message recalled the breast cancer risk content (β = .66; OR, 1.94; 95% CI, 1.08–3.49; p < .05) significantly better than those who read the informational message.