The negative consequences of failing to communicate uncertainties during a pandemic: The case of COVID-19 vaccines

Abstract

Uncertainties pervade our health choices, particularly in the context of a novel pandemic. Despite this, rather little is known about when and how to effectively communicate these uncertainties. The focus in the medical literature so far has been on how patients respond to mentions of uncertainty relating to diagnosis or treatment, showing that these can have detrimental effects on trust and satisfaction. On the other hand, how patients are affected by these communications over time, particularly in the face of conflicting information, has received little attention. This is particularly important in the context a novel pandemic where uncertainty is rife and information changes over time. To fill this gap, we conducted an online study with UK participants on hypothetical communications relating to COVID-19 vaccines. Participants first read a vaccine announcement, which either communicated with certainty or uncertainty, and then received information which conflicted with the announcement. Those who were exposed to the certain announcement reported a greater loss of trust and vaccination intention than those who were exposed to the uncertain announcement. This shows that communicating with unwarranted certainty can backfire in the long-term, whereas communicating uncertainties can protect people from the negative impact of exposure to conflicting information.

The effects of communicating uncertainty around statistics on public trust: an international study

“…While a single message with uncertainty may not increase trust, it might buffer against future damage to credibility if figures are revised. Batteux et al (2021) report that the inclusion of a range (vs point estimate) in COVID-19 vaccine efficacy communications had no immediate impact on trust. However, when participants were later presented with conflicting evidence (updating the previous estimate), those who had first received a point estimate with no uncertainty reported lower trust in the communicator, compared to those receiving a range.…”

Training Calibration-based Counterfactual Explainers for Deep Learning Models in Medical Image Analysis

“…All images were resized to 224×224 pixels, and treated as single channel images. With the latent space dimensionality fixed at 100, the encoder model was comprised of 4 convolutional layers, with the number of filters set to [16,32,64,32], followed by two fully connected layers with hidden units as 512 and 100. All convolutional layers used the kernel size (3,3) and stride 2.…”

“…Meaningful uncertainties can play a crucial role in supporting practical objectives that range from assessing regimes of over (or under)-confidence and active data collection, to ultimately improving the predictive models themselves 14 . However, in practice, uncertainties are known to be challenging to communicate to decision-makers 15 , and the robustness of decisions with respect to uncertainties can vary considerably between use-cases 16 . Consequently, it could sometimes be more beneficial to implicitly leverage uncertainties when performing introspective analysis of machine-learned models.…”

Narrative expectations in financial forecasting

“…Here, we restrict ourselves to testing two of these predictions, concerning the structure of narratives, though we test several other predictions of CNT elsewhere (Batteux et al, 2020, 2021; Bilovich et al, 2020; Johnson, Matiashvili, & Tuckett, 2019a, 2019b; Nyman et al, 2018; see Johnson, Bilovich, & Tuckett, 2020 for a review), including predictions about narrative content . Here, we examine the consequences of narratives being goal‐oriented and temporally oriented .…”