When Correlation Does Not Imply Causation – No Evidence for Causal Effects of Trust in Science on Pandemic-Related Protection Intentions

Drawing the right lessons from the COVID-19 pandemic will be essential for effective future policy responses. One central finding from the past pandemic is that trust in science predicts health-related protection intentions and behaviors, such as social distancing and vaccination. Many researchers and policymakers thus believe that interventions targeting trust in science will be key for curbing the spread of future pandemics. Yet, it is unclear whether the observed correlation between trust in science and protection intentions does indeed imply causation. In a series of studies (total N = 5,311), we replicated this correlation between trust in science and protection intentions (Studies 1, 3, and 4). At the same time, when experimentally manipulating trust in science, we found no evidence for causal effects on protection intentions (Studies 2 to 4). This absence of meaningful effects was further confirmed by equivalence tests and by an internal meta-analysis (N = 3,761). Finally, a machine learning algorithm likewise found no evidence that manipulating trust in science affects protection intentions. While it is inherently difficult to prove the absence of an effect, these results cast doubts on the causal importance of short-term changes in trust in science for protection intentions.