ABSTRACT
Even though the immediate urgency of the COVID-19 pandemic seems to have passed, many countries did not reach the vaccination rates they initially aimed for. The stagnation in vaccine uptake during the height of the pandemic presented policy makers with a challenge that remains unresolved and is paramount for future pandemics and other crises: How to convince the (often not insubstantial) unvaccinated proportion of the population of the benefits of a vaccination? Designing more successful communication strategies, both in retrospect and looking ahead, requires a differentiated understanding of the concerns of those that remain unvaccinated. Guided by the elaboration likelihood model, this paper has two objectives: First, it explores by means of a latent class analysis how unvaccinated individuals might be characterized by their attitudes towards COVID-19 vaccination. Second, we investigate to what extent (i) varying types of evidence (none/anecdotal/statistical) can be employed by (ii) different types of communicators (scientists/politicians) to improve vaccination intentions across these subgroups. To address these questions, we conducted an original online survey experiment among 2145 unvaccinated respondents from Germany where a substantial population share remains unvaccinated. The results suggest three different subgroups, which differ regarding their openness towards a COVID-19 vaccination: Vaccination opponents (N = 1184), sceptics (N = 572) and those in principle receptive (N = 389) to be vaccinated. On average, neither the provision of statistical nor anecdotal evidence increased the persuasiveness of information regarding the efficacy of a COVID-19 vaccine. However, scientists were, on average, more persuasive than politicians (relatively increase vaccination intentions by 0.184 standard deviations). With respect to heterogeneous treatment effects among the three subgroups, vaccination opponents seem largely unreachable, while sceptics value information by scientists, particularly if supported by anecdotal evidence (relatively increases intentions by 0.45 standard deviations). Receptives seem much more responsive to statistical evidence from politicians (relatively increases intentions by 0.38 standard deviations).