More than scientists: How message and messenger attributes influence viewers’ climate change intentions

PLOS One | https://doi.org/10.1371/journal.pone.0331672 September 10, 2025 8 / 20 Political orientation. The seven-item, Conservatism-Liberalism Scale (CLS; [77]) assessed political orientation. The Likert format CLS ranges from 1 (strongly disagree) to 5 (strongly agree). Three items endorse conservative orientation and four endorse liberal orientation. Liberal endorsing items are reverse scored for interpretation of higher scores to indicate conservative orientation (α = .84). Statistical Analyses We used policy capturing, in conjunction with multilevel modeling, to determine which message and messenger attributes of the MTS videos would predict viewer motivations to act and seek out more information about climate change. Pol- icy capturing is a method used in applied psychology to explore the relationships between people’s judgments and the information used to make those judgments [78], and multilevel modeling is a flexible statistical procedure for quantifying the magnitude of these relationships. In combination, these two tools have been applied in a range of domains including: environmental sustainability [79], job performance [80], academic decision-making [81], and climate-change communica- tion [13]. In the current study, all of the policy capturing analyses were conducted using HLM-6 [82]. Separate analyses were conducted for: (1) LIWC summary variables (analytical thinking, clout, authenticity, and emotional tone), (2) LIWC drive variables (affiliation, achievement, power, reward focus and risk focus), and (3) messenger attribute variables (sex, age,

attractiveness, whether additional family members were present, and whether the presenter was situated in a natural setting). Given that each respondent provided motivation ratings in response to six MTS videos, randomly selected from a pool of 41, ratings were nested within participants. In the Level 1 (within-person) analyses, regression equations were com- puted for each participant using motivation to act on climate change as the criterion variable and either the LIWC sum- mary variables, LIWC drive variables, or messenger attributes as the predictor variables. The Level 2 (between-person) analyses employed a restricted maximum likelihood approach in which the inter- cept and beta coefficients from the Level 1 analyses were regressed on one Level 2 variable, viewers’ political orientation. The Level 2 analysis enabled us to assess whether the MTS message/messenger effects on climate change motivation varied systematically as a function of viewer’s political orientation; that is whether certain types of messages and messengers were more effective for viewers with left-leaning or right-leaning political orientations. Sample size is an important consideration in multilevel studies. Maas and Hox [83] ran a series of simulations in which they varied Level 1 and Level 2 sample sizes. They found that only simulations with small Level 2 sam- ples (consisting of 50 or fewer cases) produced biased Level 2 standard errors. All other simulations, including those with Level 1 sample sizes as small as five, produced accurate and unbiased regression coefficients, variance

components, and standard errors at both Level 1 and Level 2. Given the current study had over 500 respondents at Level 2 and six ratings per respondent at Level 1, it exceeded Maas and Hox’s [83] recommended sample size guidelines. Results Descriptive Statistics Means, standard deviations and, where appropriate, frequencies are reported for the continuous message, mes- senger, and motivation to act variables in Table 1. For the categorical messenger variables, 58% of the presenters were male, 37% were female, and 5% had both male and female presenters. Just under a third of presenters (30%) presented in natural (as opposed to indoor) settings, and 9% included family members in their presentations. Cor- relations for the message and messenger variables used in the study are provided in the supplementary materials for the article.