Understanding Blame in the Context of Childhood Obesity

ABSTRACT

This paper explicates blame as a psychological construct in communication processes, with a focus on its underlying structure and its mediating role between message characteristics and the public’s issue engagement. Data were collected from Amazon Mechanical Turk (N = 373) via a Web-based experiment, where we manipulated a news story about a child suffering an obesity-related health risk (asthma attack vs. heart attack) with different levels of preventability (high vs. low) and severity (high vs. low). Findings showed that blame should best be conceptualized and operationalized as a latent construct comprising both cognitive and affective components. Blame mediated message effects on social responses related to obesity prevention, including punitive attitudes, policy support, and intentions toward interpersonal communication and civic participatory behaviors.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Estimates from the sensitivity tests using G*power (Faul et al., 2009) suggested that given our experimental design, this sample size would help obtain a statistical power of .80 or above for a small to medium effect size (f = .15 and .18 for 80% and 95% statistical power, respectively, assuming an alpha of 0.05).

2. Discriminant validity of these factors was examined in two ways. First, CFA results showed that a five-factor model was substantially better than the one-factor model. The final five-factor measurement model: χ²(125) = 424.38, RMSEA = .080, CFI = .93, SRMR = .06. Second, using the average variance extracted (AVE) approach (Fornell & Larcker, 1981), for all pairs of variables, the square root of their respective AVE was higher than the factor correlation between the two, meeting the condition for discriminant validity.

3. The variables where there were such statistically significant differences included the following– (1) Perceived outcome preventability: heart attack message, M = 4.16, SD = 0.73; asthma attack message, M = 3.95, SD = 0.75; t(371) = 2.76, p < .01. (2) Perceived outcome severity: heart attack message, M = 4.31, SD = 0.73; asthma attack message, M = 3.93, SD = 0.88; t(365.85, equal variance not assumed) = 4.55, p < .001. (3) Indignation: heart attack message, M = 3.20, SD = 1.12; asthma attack message: M = 2.85, SD = 1.09; t(371) = 3.05, p < .01. (4) Punishment: heart attack message, M = 3.23, SD = 1.27; asthma attack message: M = 2.85, SD = 1.28; t(371) = 2.88, p < .01. (5) Interpersonal communication: heart attack message, M = 3.21, SD = 1.32; asthma attack message, M = 3.54, SD = 1.13; t(371) = 2.65, p < .01. (6) Civic participation: heart attack message, M = 2.34, SD = 1.22; asthma attack message, M = 2.61, SD = 1.27; t(371) = 2.11, p < .05. On responsibility attribution and two variables of policy support, there was no statistically significant difference.

4. Perceived preventability was measured by three items: “Ryan’s tragedy was entirely avoidable,” “What happened to Ryan could have been prevented,” and “Ryan’s situation would not have occurred to other children with more responsible parents” (1= “strongly disagree;” 5= “strongly agree;” α = .85). Perceived severity of consequence was measured by four 5-piont semantic differential items (“trivial/serious,” “severe/slight,” “negligible/dreadful,” and “grim/mild;” α = .85). Preventability manipulation induced significantly different levels of perceived preventability [Asthma attack message: F(1, 193) = 44.27, p < .001, M = 4.26, SD = 0.59 vs. M = 3.62, SD = 0.76 for high vs. low preventability conditions; Heart attack message: F(1, 180) = 13.43, p < .001, M = 4.44, SD = 0.56 vs. M = 3.88, SD = 0.76]. Similarly, perceived severity of consequence showed statistically significant differences as intended by message manipulation [Asthma attack message: F(1, 193) = 38.06, p < .001, M = 4.29, SD = 0.67 vs. M = 3.57, SD = 0.92 for high vs. low severity conditions; Heart attack message, F(1, 180) = 19.12, p < .05, M = 4.53, SD = 0.61 vs. M = 4.09, SD = 0.77].

5. Analyses with other outcome variables (i.e., interpersonal communication, parent-oriented policy support, public-oriented policy support, civic participation) also consistently showed that the intertwined model had a better fit than the dual process model.

6. In the dual process model, cognition and affect were corrected for measurement error by fixing the error term of the corresponding manifest variables to 1-$\mathit{\hspace{0.17em}}$α² times its variance. In both dual process and intertwined models, punishment was corrected for measurement error by fixing the error term of the corresponding manifest variable at 1-$\mathit{\hspace{0.17em}}$α² times its variance. For both the dual process and the intertwined models, preventability (1 vs. 0), severity (1 vs. 0), and the interaction term between preventability and severity were specified as observed exogenous variables.

7. Direct effects were not included in the reported models, given that our theoretical focus was on the structure of blame as a mediating construct. Our stimuli were manipulated to elicit varying levels of blame, not the outcome variables in any direct manner. To check whether direct effects were empirically relevant, we did additional analyses by including direct effects in the model, which resulted in a substantially worse model fit: the BIC difference between the models with vs. without direct effects was 43.02 for the asthma message and 58.19 for the heart attack message, both strongly in favor of the indirect-effects-only model (Raftery, 1995).

8. We thank a reviewer for raising this point.

Additional information

Funding

The authors reported there is no funding associated with the work featured in this article.