The effect of spokesperson attribution on public health message sharing during the COVID-19 pandemic

Abstract

It is urgent to understand how to effectively communicate public health messages during the COVID-19 pandemic. Previous work has focused on how to formulate messages in terms of style and content, rather than on who should send them. In particular, little is known about the impact of spokesperson selection on message propagation during times of crisis. We report on the effectiveness of different public figures at promoting social distancing among 12,194 respondents from six countries that were severely affected by the COVID-19 pandemic at the time of data collection. Across countries and demographic strata, immunology expert Dr. Anthony Fauci achieved the highest level of respondents' willingness to reshare a call to social distancing, followed by a government spokesperson. Celebrity spokespersons were least effective. The likelihood of message resharing increased with age and when respondents expressed positive sentiments towards the spokesperson. These results contribute to the development of evidence-based knowledge regarding the effectiveness of prominent official and non-official public figures in communicating public health messaging in times of crisis. Our findings serve as a reminder that scientific experts and governments should not underestimate their power to inform and persuade in times of crisis and underscore the crucial importance of selecting the most effective messenger in propagating messages of lifesaving information during a pandemic.

Fig 1. Temporal context of the study.

Shown are the dates at which key social distancing measures were announced on a national level by countries in the study. The time frame of data collection (March 24–30, 2020) is highlighted in gray. Empty bars indicate that no action was announced or taken by the national government (for comparability between federal states and unitary states, we only considered announcements by the federal government in federated countries, even though there may have been actions on a local, city, or state level). For a detailed list of these government announcements, see SC Appendix in S1 File.

Fig 2. Correlation matrix of all study variables.

Significance threshold is Bonferroni-corrected, p < 3.3x10^-4. Empty cell = non-significant correlation; SD = Social distancing; Num. = Number.

Fig 3. Main effects of spokesperson, country and age group on message sharing.

(a) Message sharing score on a 1–7 Likert scale. Error bars represent 95% CIs. (b) Corresponding, color-coded significant pairwise comparisons, accounting for multiple comparisons via sequential Bonferroni correction. Only significant comparisons are shown. The results show that the medical spokesperson Dr. Fauci achieved the highest level of respondents’ willingness to reshare a call to social distancing, whereas celebrity spokesperson Kim Kardashian was least effective. Celebrity spokesperson Tom Hanks, the Government, and the no-spokesperson conditions took a middle ground. The likelihood of message resharing increased with age and when respondents expressed positive sentiments towards the spokesperson.

Fig 4. Country-by-spokesperson interaction on message sharing.

(a) Message sharing score on a 1–7 Likert scale. Error bars represent 95% CIs. (b) Corresponding, color-coded significant pairwise comparisons, accounting for multiple comparisons via sequential Bonferroni correction. Only significant comparisons are shown.

Fig 5. Spokesperson-by-age-group interaction on message sharing.

(a) Message sharing score on a 1–7 Likert scale. Error bars represent 95% CIs. (b) Corresponding, color-coded significant pairwise comparisons, accounting for multiple comparisons via sequential Bonferroni correction. Only significant comparisons are shown.

Fig 6. Spokesperson-by-likeability interaction on message sharing.

(a) Message sharing score as the standardized residual of the message sharing scores (on a 1–7 Likert scale), adjusted for all demographic and attitudinal measures (see Materials and methods: Statistical analyses). Error bars represent 95% CIs. The dashed black line (95% CI, gray band) represents the effect for the no-spokesperson condition, for which a likeability could not be elicited. (b) Corresponding, color-coded significant pairwise comparisons, accounting for multiple comparisons via sequential Bonferroni correction. Only significant comparisons are shown.