Native Americans’ Responses to Obesity Attributions and Message Sources in an Obesity Prevention Campaign

ABSTRACT

This study investigated the effectiveness of Native American (NA) targeted obesity prevention messages. The researchers manipulated obesity attributions (internal vs. external) and message sources (NAs vs. non-NAs) in a 2 × 2 mixed experimental design to examine the way these message attributes influence NAs’ emotional, attitudinal, cognitive, and behavioral responses. One-hundred and eighteen Cheyenne and Arapaho (C&A) tribal citizens read two paper-based obesity prevention PSAs and then answered questions that assessed their message responses. The key findings demonstrated that the match between participants’ ethnicity and the message source’s ethnicity had a significant effect, as it reduced anger and promoted positive message attitudes and favorable source evaluations. Implications, limitations, and suggestions for future research and NA targeted health campaigns are discussed.

Acknowledgement

The authors want to express our greatest appreciation to the Cheyenne and Arapaho Health Board Members, Arapaho Chief Elvin Kenrick, Chief Allen Sutton, and their families for providing insightful advice and support related to this research. We also thank Dr. Derrell Cox and Lori Cox (Integrated Research Institute), Dr. Adam Pitluk (Coastal Carolina University), Dr. Brian Petrotta (University of Nebraska-Lincoln), and Dr. Loarre Andreu Perez, and Dr. Yoon Hi Sung (University of Oklahoma) for providing advice and guidance, as well as assistance in conducting data collection and manuscript preparation.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

¹ Historical Trauma (Brave Heart et al., 2011; Brave Heart & DeBruyn, 1998) ascribes the current health disparities prevalent among indigenous communities to historical assaults that NAs have experienced. The theoretical framework proposes that past tragedies that external forces (i.e., European explorers and settlers, the U.S. federal government) inflicted on indigenous societies remain in contemporary NA communities in various trauma responses, including emotional distress, self-destructive behaviors, and physical illnesses, (Brave Heart, 2000; Sotero, 2006; Walters et al., 2011; Whitbeck, Adams, Hoyt, & Chen, 2004). These external pressures are presumed to have changed NAs’ body weight and composition in destructive ways and continue to worsen their obesity status and related health problems (Hill & Peters, 1998).

² Because the messages targeted a specific ethnic minority and tested their effects on the group, rather than individuals, we operationalized internal attributions as primary causes of obesity that are controllable factors within the NA community. We selected these three internal obesity attributions from relevant literature that has focused more on individual-level factors. C&A Health Board members endorsed each internal obesity attribution’s relevance based upon observations of the factors that cause and exacerbate their community members’ obesity. As a counterpart to the manipulation logic, the three external obesity attributions were chosen as factors that affect NAs’ obesity rate, but are known to be uncontrollable social determinants at the NA community level.

³ White American doctors were chosen to represent non-NA doctors in the messages. With the guidance of the tribal health board members who endorsed the adequacy of our message source manipulations, Whites were considered the doctors’ encountered most commonly compared to other races and ethnicities (i.e., Black or African American, Hispanic, and/or Asian) when tribal members visit clinics and/or hospitals.

⁴ The study was designed and conducted in a multicultural/intercultural study context. The researchers and study population shared little in common in many respects, including race and ethnicity, social status, and cultural/historical backgrounds. We planned and designed the study primarily with the guidance of theories and factual data on ethnic and racial disparities in obesity in U.S. adults. However, we acknowledge that the limitations of our outsiders’ viewpoints of the specific illness in the study population should be resolved by reflecting their voice in the research process overall. Therefore, we invited the tribal leadership to be co-investigators in the study. During the year that we spent building collaborative relationships with them, we had opportunities to engage in multiple social gatherings and Health Board meetings. These opportunities allowed us to make direct observations of the community’s obesity status and learn about tribal members’ experiences and perceptions of obesity. In addition, the process helped us ensure the adequacy of applying theories and the appropriateness of selecting specific internal and external obesity attribution types. We learned further about other significant aspects (i.e., their average health literacy levels and attitudes toward academic researchers overall) that would affect our study materials’ preparation and the data collection process.

⁵ The Labor Day Powwow participants received paper-based materials because there was no internet service at the location. To maintain consistency between the two study platforms, we used screenshots of each page of the Qualtrics displays and ensured that the same randomization was applied to both study platforms.

⁶ A power analysis was conducted with G*Power to determine the sample size necessary to compare levels of between-groups factors for repeated measures. Assuming a moderate effect size (f = 0.25), a total of 98 participants in two between-subject conditions (49 for each of message source’s group) with two repeated measures (internal and external obesity attributions) was the minimum sample required (alpha = 0.05, 1- β = 0.80). A total of 118 NA participants were recruited (Head-Start Health fair: n = 4; Barefoot Park Powwow: n = 56; Labor Day Powwow: n = 58) and assigned randomly to one of two experimental conditions based upon the source factor (59 for each message source condition). Six participants’ data were excluded from the final data analysis because they did not complete the study.

⁷ The power analysis set the alpha level of 0.05 that this study used to reject the null hypotheses. Accordingly, we computed 95% bootstrap CIs with 10,000 bootstrap samples to test our findings’ statistical significance (Tables 4, 5, 6).

Additional information

Funding

This work was supported by the University of Oklahoma [Robberson Research and Creative Endeavors Grant and the Gaylord College of Journalism and Mass Communication [Dissertation fellowship].