The current project will combine psychology and data science to investigate if and how psychological traits and political identity can be used to predict susceptibility to COVID-19 misinformation.
- Machine Learning
- data science
- Cognitive Psychology
- Behavioral Economics
Social media platforms have been used as tools for large-scale dissemination of factually incorrect and/or absurd COVID-19 information. The volume of misinformation regarding how the virus spreads, how it can be treated, the dangers of vaccines and who "created" the virus prompted the World Health Organization to warn of an information pandemic happening in parallel to the COVID-19 pandemic (WHO, 2020). There is now evidence showing that online misinformation campaigns related to COVID-19 can lead to psychological distress, being associated with poorer knowledge of the disease and fewer preventive behaviors (Lee et al., 2020). In their study, Hornik et al. (2020) showed that belief in COVID-19 misinformation was negatively associated with both mask-wearing and social-distancing behaviors in a cross-sectional analysis of a nationally representative US sample. However, they showed that only beliefs about behavioral outcomes for face mask wearing and social distancing significantly predicted these behaviors over time.
Considering the impact of COVID-19 misinformation on the compliance with public health guidance, two questions are of central importance: 1) which psychological traits and characteristics predict susceptibility to misinformation about COVID-19, and 2) is it possible to mitigate susceptibility to COVID-19 misinformation using cognitive training strategies? Roozenbeek et al. (2020) showed that numeracy and trust in scientists were significant predictors of COVID-19 misinformation beliefs, but they did not investigate the role of personality traits, scientific reasoning, metacognition and emotional reactions to complex problems in their study, traits that are of central relevance to the formation/maintenance of belief systems in complex scenarios. In another seminal study, Roozenbeek and Van der Linden (2019) showed that a game to train people to master six documented techniques commonly used in the production of misinformation (polarization, invoking emotions, spreading conspiracy theories, trolling people online, deflecting blame, and impersonating fake accounts) in a large-scale randomized trial improved people's ability to spot and resist misinformation irrespective of education, age, political ideology, and cognitive style. Roozenbeek and Van der Linden (2019) were inspired by the "inoculation metaphor" that is based on the assumption that exposing, warning and familiarizing people with the strategies used in the production of fake news can boost a "cognitive immunity" when exposed to real misinformation.
The current project has two central goals. The first is to identify psychological traits and characteristics (including numeracy skills, personality traits, scientific reasoning, metacognition, emotional reactions to complex problems, cognitive control and cognitive flexibility) as well as political identity and demographic/educational characteristics can predict the susceptibility in COVID-19 misinformation. To address this research question, we'll use state-of-the-art, non-parametric and non-linear predictive modeling techniques from machine learning, to map the psychological predictors to the outcome variable (scores on a COVID-19 misinformation scale developed by Roozenbeek et al. (2020). The second goal is to verify the impact of a brief cognitive training protocol in the susceptibility of COVID-19 misinformation. This cognitive intervention will have three phases. In the first phase, participants will learn about our susceptibility to errors caused by intuitive answers to logical problems. They will be asked to answer to logical problems designed to elicit intuitive answers, and will learn strategies that can improve their capacity to control these automatic intuitions. In the second phase, the participants will learn about another relevant cognitive bias known as the "affect heuristic", in which people substitute the answer to a complex question (e.g. "how does COVID-19 spreads?") to a much easier question (e.g. "how do I feel about the information I received regarding the spread of COVID-19?"). In the third and last phase of the training program, participants will learn basic definitions of decision making under risk and the evaluation of prospects from Prospect Theory. Kahneman and Tversky's prospect theory (that granted the Nobel Prize in Economics to Kahneman in 2002) shows how people are risk-averse in uncertain conditions involving gains, but are risk-seeking under conditions involving losses. Kahneman and Tversky (1979) demonstrated for the first time that people do not behave consistently and rationally under uncertain conditions involving gains and losses. Together, intuitive thinking in complex problems, the affective heuristic and our tendency to seek certainty in uncertain conditions, to be averse to losses and to focus on differentiators while comparing prospects might play a role in the development and consolidation of beliefs in misinformation. This hypothesis seems reasonable since the COVID-19 pandemic involves a number of highly uncertain conditions (how the virus work, how it spreads, how to control it, etc) in which the possible losses are unprecedented (death, unemployment, financial struggle, emotional and psychological distress, etc).
The design of our study involves recruiting a large sample of college students from UVA (n = 5,000) that will be randomly assigned to one of two groups: 1) a passive control group and 2) an experimental group. Both groups will answer to several psychological scales and tests to assess their numeracy skills, personality traits, scientific reasoning capacity, metacognition, emotional reactions to complex problems, cognitive control, cognitive flexibility and susceptibility to COVID-19 misinformation. After answering the psychological scales/tests, only those assigned randomly to the experimental group will go through a brief training session (phases 1, 2 and 3 of the cognitive intervention mentioned above). At the end, participants from both groups will answer the COVID-19 misinformation scale (Roozenbeek) one more time. In this experiment, we'll investigate if the cognitive intervention can decrease the beliefs in COVID-19 misinformation more than could be expected by chance alone (i.e. compared to the passive control group).
1. World Health Organization. 2020 Novel Coronavirus (2019-nCoV) Situation Report, 13. See https://www.who.int/docs/default-source/ coronaviruse/situation-reports/20200202-sitrep- 13-ncov-v3.pdf.
2. Hornik, R., Kikut, A., Jesch, E., Woko, C., Siegel, L., & Kim, K. (2021). Association of COVID-19 misinformation with face mask wearing and social distancing in a nationally representative US sample. Health communication, 36(1), 6-14.
3. Roozenbeek, J., Schneider, C. R., Dryhurst, S., Kerr, J., Freeman, A. L., Recchia, G., ... & Van Der Linden, S. (2020). Susceptibility to misinformation about COVID-19 around the world. Royal Society open science, 7(10), 201199.
4. Roozenbeek, J., & van der Linden, S. (2019). Fake news game confers psychological resistance against online misinformation. Palgrave Communications, 5(1), 1-10.
5. Kahneman, D. & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 2(47).
1) Identification of psychological traits and characteristics that are predictors of susceptibility to misinformation about COVID-19. Once identified, the predictors can help us develop mitigating strategies to decrease susceptibility to COVID-19 misinformation.
2) Verify the feasibility of a brief cognitive intervention program focused on cognitive biases and heuristics as potential mitigators of beliefs in COVID-19 misinformation. If the cognitive intervention program is effective to decrease beliefs in COVID-19 misinformation, it can be used as a tool to mitigate the effects of misinformation campaigns.
Students from UVA will be recruited to participate in this study. Also, we'll hire undergraduate and graduate research assistants to help with the study.