The behavioural challenge of the COVID-19 pandemic: indirect measurements and personalized attitude changing treatments (IMPACT)

Abstract

Following the outbreak of COVID-19 pandemic, governments around the globe coerced their citizens to adhere to preventive health behaviours, aiming to reduce the effective reproduction numbers of the virus. Driven by game theoretic considerations and inspired by the work of US National Research Council's Committee on Food Habits (1943) during WWII, and the post-WWII Yale Communication Research Program, the present research shows how to achieve enhanced adherence to health regulations without coercion. To this aim, we combine three elements: (i) indirect measurements, (ii) personalized interventions, and (iii) attitude changing treatments (IMPACT). We find that a cluster of short interventions, such as elaboration on possible consequences, induction of cognitive dissonance, addressing next of kin and similar others and receiving advice following severity judgements, improves individuals' health-preserving attitudes. We propose extending the use of IMPACT under closure periods and during the resumption of social and economic activities under COVID-19 pandemic, since efficient and lasting adherence should rely on personal attitudes rather than on coercion alone. Finally, we point to the opportunity of international cooperation generated by the pandemic.

Keywords: COVID-19; attitude change; effective reproduction; pandemic.

Figure 1.

Depiction of the dependent variables and hypotheses for both the intervention and repetition condition. Initially, participants' attitude reflects a small difference between the number of people who will be infected if they adhere or if they do not adhere to health regulations. After the intervention (or within the second set of questions), the attitudes are expected to change, reflecting a lower number of infected people for those who adhere, and a higher, asymmetric, change for those who do not adhere to health regulations. (1) The difference between adherence and non-adherence in the first set of questions. (2) The difference between adherence and non-adherence in the second set of questions. (3) The shift in the number of estimated infected people who adhere to health regulations between the first (t1) and second (t2) set of questions. (4) The shift in the number of estimated infected people who do not adhere between the first (t1) and second (t2) set of questions. Table 2 shows the shift % for ‘1’ and ‘2’ differences. Table 3 shows shift % for ‘3’ and ‘4’ differences.

Figure 2.

Across languages (a, n = 3102) pre-intervention perceptions of the least and most efficient health regulation, followed by: English-speaking participants (b, n = 1214), Hebrew-speaking participants (c, n = 728), Arabic-speaking participants (d, n = 497), Polish-speaking participants (e, n = 466) and German-speaking participants (f, n = 126).

Figure 3.

Attitude change efficiencies with standard errors. The left panels show the changes in the perception of participants' estimates for the targeted attitude of each individual, i.e. the weakest personal pre-intervention attitude. This attitude is calculated by subtracting the estimated number of people who will contract the COVID-19 virus (out of 1000 people) if they adhere to a specific instruction, from the number of people who will contract the virus if they do not adhere to the same health instruction. After calculating the differences for all four health instructions, the attitude with the smallest difference, hence the attitude perceived as least efficient, is selected as the weakest attitude for each individual participant. The changes presented in the figure are calculated as the difference between pre- and post-intervention estimates, separately calculated for the numbers provided for the adherence and non-adherence questions. A successful intervention should show negative values for adherence (indicating that the post-intervention attitude reflects a reduced number of people who are likely to contract the virus) and positive values for non-adherence (indicating that the post-intervention attitude reflects a rise in the number of people who are likely to contract the virus). Each figure shows the same statistics for the repetition (only repeated estimates) and the full intervention conditions (comprising all practised attitude change procedures). The right panels show parallel non-personalized (non-targeted) indices, averaged across four health regulations. The upper two panels present cross-sample indices, lower panels present samples collected in five languages. The German sample was run only with the intervention treatment. *p < 0.05, **p < 0.01. Appendix B lists all respective ANOVA tests and effect sizes.