Table 3.

Classification of tweets used for training the model as ‘for’ or ‘against’ climate change.

‘For’ (belief): N = 3300 tweets

‘Against’ (denial): N = 3200 tweets

Climate change concern: The user believes climate change is real and worries about its negative consequences Advocate for action: The user calls for collective actions and supports adaptation and mitigation policies Scientific consensus: The user advocates for the scientific evidence on climate change and recognizes the role of greenhouse gas emissions caused by human activities

Trend denialism: The user shows disbelief that the Earth is warming and climate change is real Attribution denialism: The user believes climate change is happening, but it is a natural, unpreventable process and anthropogenic greenhouse gases are not the dominant driver Impact denialism: The user believes climate change will not have significant negative impacts on the environment and humanity Evidence denialism: The user doubts there is trustworthy scientific consensus on climate change