**Replication Model:** Consistent with the originating lab's analysis: we will run an OLS regression on the primary DV 'rbelief', testing whether seeing the researcher using the term climate change denier makes people think the researcher believes global warming is happening; controlling for female, age, Hispanic, race, education, income, and region. We will first re-code all responses of 'Cannot tell whether researcher thinks global warming has or has not been happening' as zero, consistent with the originating lab's analysis. We will then run the following code in STATA regress rbelief i.label i.female age i.hispanic i.race i.ed i.income i.region **Follow-up analyses:** 1. *Demographic Analysis*: As a follow-up series of analyses, we will first run a model with only the demographics, dropping those that do not predict beliefs. Then, retaining those, we will add the treatment variable. 2. *Just Treatment*: We will also test whether the treatment caused a difference in people's beliefs. We will run an unordered logistic regression on all three response options, only controlling for the demographics that come out in Analysis 1. 2. *Multiple DVs*: We will run a path model on all five DVs, allowing them to correlate with on another. The rbelief variable will be treated as an unordered categorical variable, all others will be treated as ordered categorical variables. This will be done with no covariates.