Telling people to “rely on their reasoning” increases intentions to wear a face covering to slow down COVID‐19 transmission

Abstract Finding messaging to promote the use of face masks is fundamental during a pandemic. Study 1 (N = 399) shows that telling people to “rely on their reasoning” increases intentions to wear a face mask, compared with telling them to “rely on their emotions.” In Study 2 (N = 591) we add a baseline. However, the results show only a non‐significant trend. Study 3 reports a well‐powered replication of Study 2 (N = 930). In line with Study 1, this study shows that telling people to “rely on their reasoning” increases intentions to wear a face mask, compared to telling them to “rely on their emotions.” Two internal meta‐analyses show that telling people to “rely on their reasoning” increases intentions to wear a face mask compared (1) to telling them to “rely on their emotions” and (2) to the baseline. These findings suggest interventions to promote intentions to wear a face mask.


| INTRODUCTION
The coronavirus disease  pandemic is one of the greatest health threats of the last century. At the time of writing (January 11, 2021), more than 90 million people have tested positive and more than 1.9 million are dead (Worldometers, 2021) and these are probably substantial underestimations (Burn-Murdoch et al., 2020).
The large impact of COVID-19 is partly due to its transmissibility by asymptomatic people, who often are unaware of their infection, through viral droplets in coughs or sneezes (Bai et al., 2020;Mizumoto et al., 2020;Nishiura et al., 2020). For this reason, epidemiologists and health experts have recommended the use of face coverings, with the aim of minimizing the number of infected droplets spread by asymptomatic people, thereby reducing the risk of infecting others. In line with these experts' suggestions, a study based in Germany found that the use of face masks reduced the daily growth rate of reported infections by approximately 40% (Mitze et al., 2020), whereas a study based in Beijing, China, exploring transmission in families with at least one laboratory confirmed COVID-19 case found that "face mask use by the primary case and family contacts before the primary case developed symptoms was 79% effective in reducing transmission" ).
Yet, we might expect that people may be reluctant to wear a face covering since it represents a significant change in their habitual behavior. It follows that developing mechanisms that favor the use of face masks is crucial to slow down COVID-19 transmission and "flatten the curve" of the spread. Several national or local governments have taken the difficult decision of making the use of face coverings mandatory in a number of contexts (Javid, 2020). However, since it is impossible to monitor the behavior of every person, even in places where wearing a face covering is mandatory, explicit laws should be complemented by implicit behavioral "nudges" aimed at directing people's behavior towards desired outcomes. In particular, appeals and messages can be effective at promoting desired behavioral changes, because they reach people both inside their homes, through television and social media, as well as outside their homes, through screens, posters, and megaphones. This raises the important question of which types of messaging are effective in promoting the use of face coverings (Van Bavel et al. 2020).
Little is known about this question. Several papers have explored the effect of appeals and messaging on intentions to engage in COVID-19 preventive behaviors (Bilancini et al., 2020;Capraro & Barcelo, 2020;Everett et al., 2020;Falco & Zaccagni, 2020;Heffner et al., 2020;Jordan et al., 2020;Lunn et al., 2020;Pfattheicher et al., 2020). However, with the exception of one paper, none of these works explored the effect of messages on intentions to wear a face covering; the only exception is Capraro and Barcelo (2020), which found that telling subjects that the coronavirus (COVID-19) is a threat to their community increases intentions to wear a face covering, relative to the baseline. In the current paper, we contribute to this area of the literature by exploring the effect of telling people to "rely on their reasoning" versus telling people to "rely on emotion" versus the baseline with respect to intentions to wear a face covering. This is an important practical question: if one of these messages has a positive effect, it would offer a simple scalable intervention to promote intentions to wear a face covering.
We report three pre-registered experiments (total N = 1920). The experiments were conducted on a heterogeneous, although not representative, sample of people living in the US and surveyed using Amazon Mechanical Turk (Paolacci et al., 2010). The main results are: combining Studies 1-3, we find that telling people to "rely on their reasoning" increases intentions to wear a face covering relative to telling them to "rely on their emotions"; putting Studies 2 and 3 together, we find that the effect is primarily driven by reasoning, meaning that, compared to the baseline, promoting reasoning and logic significantly increases intentions to wear a face covering, whereas promoting emotions does not significantly change intentions to wear a face covering. Participants were randomly assigned to one of two conditions: in the promoting emotion condition, they were shown a message highlighting the positive consequences of making decisions based on feelings; in the promoting reasoning condition, they were shown a message highlighting the positive consequences of making decisions based on reasoning. These messages were taken from previously published work (Capraro et al., 2019;Caviola & Capraro, 2020;Levine et al., 2018). See Table 1 for the exact messages.

| Dependent variables
After reading the message, all participants took the following scale.
Intentions to wear a face covering. Participants were asked to: "answer the following questions by relying on emotions [reasoning].
When the shelter-in-place rules are relaxed, I intend to … 1. Wear a face covering any time I leave home. All answers were collected using a 10-line "snap to grid" slider with three labels: "strongly disagree" at the extreme left, "neither agree nor disagree" at the center, "strongly agree" at the extreme right.

| Demographics
After the scale, participants were asked the following set of demographic questions: sex, age, race, political views, religiosity, whether they live in an urban area, whether wearing a face covering is mandatory in their county, whether they live in an area where shelter-in-place rules apply, whether they previously tested positive, whether they believe they will contract coronavirus and, if so, whether they believe they will recover from it relatively easily. At the end, there was a control question to prevent the potential intrusion of bots.

| Pre-registration
The design, the analysis and the sample size were pre-registered at: https://osf.io/hfjpw/?view_only= cc5aa039b96d4075a3c834c408091992. For this and for the following studies, we report all measures and conditions.

| Results
The experiment was conducted on May 28, 2020. The raw data of this and the following studies may be found at: https://osf.io/hfjpw/?
Promoting emotion Sometimes people make decisions by using feelings and relying on their emotions. Other times, people make decisions by using logic and relying on their reasoning. Many people believe that emotions lead to good decisionmaking. When we use feelings, rather than logic, we make emotionally satisfying decisions. Please answer the following questions by relying on emotions, rather than reasoning.

Promoting reason
Sometimes people make decisions by using logic and relying on their reasoning. Other times, people make decisions by using feelings and relying on their emotions. Many people believe that reason leads to good decision-making. When we use logic, rather than feelings, we make rationally satisfying decisions. Please answer the following questions by relying on reasoning, rather than emotions.
Note: Between-subjects random assignment.
view_only=cc5aa039b96d4075a3c834c408091992. The analysis code can be easily replicated by the reader following the analysis below.

| Demographic characteristics of the sample
As pre-registered, we eliminated from the analysis subjects who did not pass the attention check and, for each multiple IP address or Turk ID, we kept only the first observation and discarded the rest. This meant deleting about 1% of the observations; our main results remain qualitatively similar when including these observations. In doing so, we were left with 399 subjects. A posteriori sensitivity analysis shows that this sample size is sufficient to detect an effect size of d = 0.28, with power of 0.80 and with α = 0.05, two-tailed. In Table 2

| The effect of promoting emotion versus reasoning on intentions to wear a face covering
We first build the composite variable "intentions to wear a face covering" by taking the average of its three items (α emotion = 0.932, α reason = 0.924). The average intention to wear a face covering when promoting reasoning is M reason = 7.38 (SD reason = 3.00); the average intention to wear a face covering when promoting emotion is M emotion = 6.61 (SD emotion = 3.24). Wilcoxon rank-sum shows that the distribution of intentions to wear a face covering when reasoning is promoted is statistically different from the corresponding distribution when emotion is promoted (z = 2.366, p = .018). Note: Political view goes from 1 = "very left-leaning" to 7 = "very right-leaning," with 4 = "center." In the table we classified as "center" only those subjects who answered "center."

| STUDY 2
the baseline condition, participants are not presented with any messaging before taking the "intentions to wear a face covering" scale.

| Pre-registration
The design, the analyses, and the sample size were pre-registered at:

| The effect of promoting emotion versus reason on intentions to wear a face covering
We first build the composite variable "intentions to wear a face covering" by taking the average of its three items (α emotion = 0.914, α baseline = 0.937, α reason = 0.941). The average intention to wear a face covering when promoting reasoning is M reason = 6.89 (SD reason = 3.34); the average intention to wear a face covering in the baseline is M emotion = 6.71 (SD emotion = 3.24); the average intention to wear a face covering when promoting emotion is M emotion = 6.65 (SD emotion = 3.01).
A one-way ANOVA with Bonferroni correction 1 reveals that there are no statistically significant differences across conditions (F(2,588) = 0.31, p = .731).

| STUDY 3
Study 2 finds a non-significant trend in the same direction as Study 1. One possibility is that Study 1 was a false positive. Another possibility is that Study 2 failed to find an effect for some reason. To clarify this, we conducted a third study with a sample size large enough to detect a small effect of d = 0.20 with power 0.80 and alpha = 0.05.
This sample size was determined by the a priori power analysis reported in the pre-registration.

| Conditions
Study 3 is identical to Study 2.

| Demographic characteristics of the sample
This experiment was conducted on June 1, 2020. People who participated in either of the previous two studies were not allowed to participate in this study. As pre-registered, we eliminated from the analysis subjects who did not pass the attention check and, for each multiple IP address or Turk ID, we kept only the first observation and discarded the rest. This corresponds to deleting about 7% of the observations; our main results remain qualitatively similar when including these observations. In doing so, we were left with 930 subjects. The demographic characteristics of the sample are reported in Table 2.

| The effect of promoting emotion versus reason on intentions to wear a face covering
We first build the composite variable "intentions to wear a face covering" by taking the average of its three items (α emotion = 0.933,

| Pooling the three studies together
As pre-registered in Study 3, we pooled all the data together to increase the power and test which of the three effects are most significant with a larger sample size. Since the three conditions are identical across studies, we simply pooled the data together by condition (Curran & Hussong, 2009  See Figure 1. If we repeat the ANOVA by adding an interaction term condition*study we find that the interaction is not significant (F(3,1916) = 1.00, p = .390), suggesting that the effect of condition is similar across studies. Also, the meta-analysis found no significant heterogeneity across conditions (see Footnote 2).

| Exploratory analysis looking at potential moderators of the effect
As exploratory analysis, we added each demographic variable as a separated moderator, in order to test whether the effect of the treatment is particularly strong on any subset of participants. In doing so, we found

| DISCUSSION
Here we reported three pre-registered studies exploring the effect of promoting emotion versus reasoning on intentions to wear a face covering. Study 1 shows that telling people to "rely on their reasoning" increases intentions to wear a face mask, compared with telling them to "rely on their emotions." Study 2 attempts to replicate Study 1, with the addition of a baseline. However, the results show only a nonsignificant trend, albeit in the anticipated direction. Study 3 reports a well-powered replication of Study 2. In line with Study 1, this study shows that telling people to "rely on their reasoning" increases intentions to wear a face mask, compared with telling them to "rely on their emotions." An internal meta-analysis shows that telling people to "rely on their reasoning" increases intentions to wear a face mask both compared with telling them to "rely on their emotions" and with the baseline, whereas compared to the baseline, promoting emotion has no effect on intentions to wear a face covering. The latter finding should be taken with caution because the data trend in the direction that promoting emotion decreases intentions to wear a face covering, compared to baseline. Therefore, it is possible that we failed to detect the effect of promoting emotion versus baseline due to insufficient statistical power.
These results contribute to the emerging literature on messaging that increases engagement in preventative COVID-19 behaviors. Bilancini et al. (2020) found that nudging the personal, or the descriptive, or the injunctive norm has no effect on understanding COVID-19 related governmental rules. Capraro and Barcelo (2020) reported that telling people that "the coronavirus is a threat to your community" increases intentions to wear a face mask, compared to the baseline. Everett et al. (2020)  Of course, our results have some limitations. One regards the sample. Our results were obtained with a heterogeneous, but not representative, sample of people living in the US. We did not find evidence that our results were driven by a particular subset of the population: we included each demographic variable as a potential moderator into separate regression models and found that none of the demographic variables moderated the effect of the messages on intentions to wear a face covering. However, future research should test whether our results can be generalized to the American population at large. Of course, our results cannot be readily generalized to other countries. We suggest that non-American policymakers who might be interested in using these messages to promote the use of face coverings outside the USA test their effect on intentions to wear a face mask in their countries before implementing them on a large scale. A major limitation of our study is the fact that it focuses on intentions, rather than actual behavior. A recent study found that intentions to practice physical distancing are correlated to actual behavior (Gollwitzer et al., 2020). Although this certainly does not imply that intentions to wear a face covering correlate with actual behavior, it does give some hope that it could actually be the case. Future work should test whether messages of the form used in this paper impact people's actual use of face coverings.
From a theoretical perspective, our results raise the question of why promoting reasoning increases intentions to wear a face covering. There are several possibilities: one is that reasoning deactivates the negative emotions that people feel when wearing a face mask (Capraro & Barcelo, 2020); another is that people generally tend to underestimate their likelihood of infection and reasoning makes them more realistic about their personal health risks (Sjåstad & Baumeister, 2020); another is that reasoning makes people introspect and reflect on their motivations (Wilson & Schooler, 1991;Wilson et al. 1993). Future work could disentangle these potential explanations. Related to these issues, another question is whether the subjects truly acted under emotion/reason or acted as if they were under emotion/reason. Disentanglement of these issues could be a worthwhile subject of future research.