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 9 / 20 Policy Capturing Analyses Unconditional model. As an initial step, an unconditional model (i.e., no predictors at within-individual or between- individual levels) was used to decompose the total variance in climate change motivation into within- and between-person components. The intraclass correlation from the unconditional model was rI = .72, indicating that almost three quarters of the variance in motivation was attributable to individual differences (between-subjects variance), and just over a quarter reflected within-subject variance across messages viewed. Level 1 models: Which message/messenger attributes motivate viewers? Average unstandardized coefficients and robust standard errors for the intercept and for all three Level 1 analyses are presented in Table 2. Our initial Level 1 analysis involved regressing viewers’ motivation to act on climate change (the criterion variable) on four LIWC summary variables: Analytic thinking, clout, authenticity, and positive emotional tone. None of the four LIWC summary variables significantly predicted viewers’ intentions, and together explained less than 1% of the variance in the criterion variable. Our second Level 1 analysis regressed viewers’ climate change motivation on LIWC’s five drive variables: affiliation, achievement, power, reward focus, and risk focus. Of the drive variables, only risk focus significantly predicted motivation

to act. The effect was negative, indicating that MTS videos that emphasized climate change risks were, in general, demo- tivating for viewers. But as with the LIWC summary variable analysis, the amount of explained within subject variance in viewers’ climate change motivation was small, less than 1%. Our final Level 1 analysis used viewers’ climate change motivation as the dependent variable and five messenger attributes as predictors: sex, age, attractiveness, whether other family members were present, and whether the presenter was in a natural setting. All messenger variables, except present with other family members, explained significant vari- ance in viewers’ motivation to act on climate change. Viewers expressed stronger motivation to act if the MTS presenter was male, older, attractive, and presenting in a natural setting. Together these predictors explained 21.46% of the within subject variance in viewers’ motivation to act. Level 2 models: Do message/messenger effects vary as a function of viewers’ political orientation? A second major aim of the study was to determine whether the effects of message and messenger attributes on viewers’ motivation to act varied according to their political orientation. Specifically, we examined whether left-leaning and right-leaning Table 1. Descriptive statistics for all continuous variables. Variables Minimum Maximum Mean Standard Deviation Participant Variables Political Orientation 1.0 5.0 3.1 0.9 LIWC Summary Variables Analytic Thinking 4.9 94.9 44.67 25.3 Clout 18.3 99.0 65.99 23.1

Authentic Style 5.0 98.9 61.7 26.1 Positive Emotional Tone 1.00 99.0 50.0 29.7 LIWC Drive Variables Affiliation 0.00 9.9 3.9 2.7 Achievement 0.00 4.8 1.4 1.1 Power 0.00 5.0 2.1 1.3 Reward 0.00 3.2 0.9 0.8 Risk-Focus 0.00 4.2 0.8 0.9 Motivation to Act 1.00 5.00 2.7 1.3 Note: Summary variable scores are percentiles; drive variable scores are percentages. Political Orientation and Motivation to Act scores are 5, 4, and 5-point Likert-scale ratings, respectively. N = 581 for participant variables and 3,486 for all other variables. https://doi.org/10.1371/journal.pone.0331672.t001