International climate adaptation assistance: Assessing public support in Switzerland

PLOS ONE | https://doi.org/10.1371/journal.pone.0317344 February 12, 2025 8 / 26 PLOS ONE Support for adaptation assistance in Switzerland Methods Empirically, we examine public support for bilateral climate adaptation assistance via a conjoint experiment nested in a survey experiment in Switzerland. As public policy is at least partly responsive to public opinion [79], surveys are an important tool to assess preferences about existing or proposed policies. Specifically, a conjoint exper- iment, contrary to other experimental designs, allows for a concurrent assessment of the causal effect of different policy attributes on public support for climate adaptation policies. Such conjoint experiments reduce social desirability bias because respondents can implicitly trade off attributes against each other. This makes them particularly suited to studying the drivers of policy preferences. Validation efforts of survey-experimental conjoint methods have shown that this experimental approach closely replicates the outcomes in popular votes [80]. If individuals are boundedly rational, their policy preferences might be influenced by how policies are framed. This is why we embedded the conjoint in a survey experiment where respondents are randomly assigned to two treatment frames that highlight aid and migration policies of peer countries as well Switzerland’s past policies on these subjects. Survey experiments are sometimes criticized for their drawbacks in terms of external valid- ity, they only offer a snapshot of a given case at one point in time, and they may not always

mirror real-world situations. The conjoint design of our experiment and its setting mitigate some of the above concerns. On the one hand, conjoint experiments are thought to have a better generalization potential compared to other survey experiments because they better approximate real-world situations where individuals are asked to evaluate policies with mul- tiple attributes [81]. Such evaluation exercise is, moreover, quite common in the semi-direct democratic context of Switzerland, where individuals often evaluate policies and their trade- offs when called to vote upon them [76]. Lastly, the method lends itself to rigorous replication and comparative understanding of the politics of climate aid, in this case concerning a prior analysis of Japan. Sampling strategy, sample, and survey administration We administered an online survey-embedded conjoint experiment to a sample of 876 Swiss residents aged 18 and above. Respondents were drawn from a Qualtrics panel. We established quotas on gender, age, and spoken language to ensure the sample was representative of the Swiss population, which resulted in a representative sample of 876 respondents. Qualtrics XM ran this survey between July 4 and July 19, 2023, in the three Swiss official languages (German, French, and Italian). Internet panels have their drawbacks, compared to population-based probability sam- pling, particularly regarding their representativeness of older age groups of society with lower internet access [82]. However, studies have shown that Qualtrics panels perform well

when compared to population-based random samples and better than those of other pro- viders [83,84]. Furthermore, the possibility of setting quotas – and their use – can mitigate these concerns. The data in our analysis is representative across key dimensions of the Swiss population. After Human Subject Approval was obtained by the authors, the experiment was pre- registered (https://osf.io/wahxb?view_only=73e3c88760ed478e87c90a4fcc86add2). More- over, given that the survey was to be administered in three different languages of Switzerland (German, French, and Italian), authors informally pre-tested the questionnaire with native speakers before the official launch to make sure that the questions were written clearly and understood by respondents across linguist groups.