Multilevel predictors of climate change beliefs in Africa

perception or believing climate change must be stopped. The limited emotional salience of cli- matic information compared with personal experience or motivated reasoning may account for this divergence [29–31]. Nevertheless, it must be noted that the importance of information is contingent on language (Fig 1D). Not speaking French, English or Portuguese relates to a hindered understanding of climate terminology, which frequently lacks accurate translations to African languages [16]. Greater efforts should be made to translate the nature, causes, and impacts of climate change to African languages. Local changes in climate conditions are among the most important predictors across all models. Overall, they are more important than access to news, political ideology, or demo- graphics (see Table 1). Previous research finds that perceived changes in local temperatures were the most important predictor of climate risk perception in some African countries [10]. Building upon these findings, this study shows that actual long-term anomalies in tempera- ture, rainfall, and drought at second-order boundaries are key predictors of individual climate change beliefs. Attribute substitution and emotional salience may explain the importance of personal experience with local weather conditions for CCBs [29, 39]. In fact, qualitative evi- dence suggests that many communities in Africa understand climate change not as a global Fig 2. Key predictors of need to stop climate change and self-efficacy. Top 15 predictors of need to stop climate change (A) (79.5%

prediction accuracy) and self-efficacy (C) (74.1% prediction accuracy). (B) Partial dependence plot (PDP) of mean temperature anomalies (temp. anom.), belief in human causation (CC human cause), perceived worsening of agricultural conditions (agric. cond.), and perceived risk from climate change (CC risk percep.) over need to stop climate change. (D) Partial dependence plot of mean temperature anomalies (temp. anom.), belief in human causation (CC human cause), perceived risk from climate change (CC risk percep.) and being favourable to one-man rule (authoritarian) over self-efficacy. https://doi.org/10.1371/journal.pone.0266387.g002 PLOS ONE Multilevel predictors of climate change beliefs in Africa PLOS ONE | https://doi.org/10.1371/journal.pone.0266387 April 5, 2022 8 / 14