medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020.

The copyright holder for this

preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

How to Make COVID-19 Contact Tracing Apps work: Insights

From Behavioral Economics
Ian Ayres,1 , Alessandro Romano1 ,2 , Chiara Sotis,3
1 2 3
Yale Law School, Bocconi Law School, London School of Economics and Political Science

Abstract

Due to network effects, Contact Tracing Apps (CTAs) are only effective if many people
download them. However, the response to CTAs has been tepid. For example, in France
less than 2 million people (roughly 3% of the population) downloaded the CTA. Against
this background, we carry out an online experiment to show that CTAs can still play a key
role in containing the spread of COVID-19, provided that they are re-conceptualized to
account for insights from behavioral science. We start by showing that carefully devised
in-app notifications are effective in inducing prudent behavior like wearing a mask or
staying home. In particular, people that are notified that they are taking too much risk
and could become a superspreader engage in more prudent behavior. Building on this
result, we suggest that CTAs should be re-framed as Behavioral Feedback Apps (BFAs).
The main function of BFAs would be providing users with information on how to minimize
the risk of contracting COVID-19, like how crowded a store is likely to be. Moreover,
the BFA could have a rating system that allows users to flag stores that do not respect
safety norms like wearing masks. These functions can inform the behavior of app users,
thus playing a key role in containing the spread of the virus even if a small percentage of
people download the BFA. While effective contact tracing is impossible when only 3% of
the population downloads the app, less risk taking by small portions of the population
can produce large benefits. BFAs can be programmed so that users can also activate a
tracing function akin to the one currently carried out by CTAs. Making contact tracing
an ancillary, opt-in function might facilitate a wider acceptance of BFAs.
Keywords: COVID-19, Contact Tracing, Framing, Media
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

1. Introduction

Contact tracing, defined by the World Health Organization (WHO) as “the process of
identifying, assessing, and managing people who have been exposed to a disease to prevent
onward transmission” [1] is unanimously considered a key tool to contain the spread of
COVID-19. Traditionally, contact tracing is carried out by public health authorities who
interview people that have tested positive to an infectious disease. In these interviews,
respondents are asked to identify the people that they have been in contact with and to
determine some characteristics of these encounters (e.g. duration, proximity, etc.). After
the interview, contact tracers attempt to inform the people that have been in contact
with the person who tested positive and give them recommendations on how to behave.
However, the COVID-19 pandemic has proven to be a challenge for traditional contact
tracing due to the speed at which the virus spreads [2], and to the fact that patients can
be contagious while asymptomatic [3]. Under these circumstances, human contact tracing
becomes a highly inaccurate process. On the one hand, often people cannot remember
who they were in close contact with during the period in which they were infectious. On
the other hand, for people it is virtually impossible to identify strangers that they have
been in contact with in places like stores and public transportation. Additionally, contact
tracing can only be effective if people are informed in timely manner that they have been
exposed to the virus, so that they can self-isolate and avoid spreading the virus further
[4]. But human contact tracing is a slow and expensive process that can take many days,
if not weeks [3].
To address these shortcomings of human contact tracing, policymakers around the
world have looked at Contact Tracing Apps (CTAs). In a nutshell, the basic idea is
that people would download a CTA on their cellphone that can inform them when they
have been in contact with a person that has tested positive for COVID-19. In principle,
CTAs could be the perfect complement to human contact tracing, as they can record
contacts among people that do not know each other and can send notifications almost
instantaneously. Consequently, it is not surprising that almost 50 countries have either
launched or are planning to launch a CTA [5].
Nevertheless, the results produced by CTAs have been disappointing in virtually all

2
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

places in which they have been introduced. As they are currently conceived, CTAs’
main function is to notify people when they have entered in contact with a person who
tested positive to COVID-19. To perform this task, it is necessary that both the infected
person and the person she entered into contact with have installed the app. Therefore, if
50% of the people download the app, only 25% of the encounters between two randomly
picked members of the population would be registered. Consequently, epidemiologists
have estimated that an adoption rate of 90%-95% is required in order to effectively control
COVID-19 spread [6, 7]. Additionally, they have estimated that for contact tracing in
general to be effective it is necessary to identify and isolate 60% of cases within a few
days [4, 8].
However, even in countries characterized by high level of trust towards institutions
like Singapore and Norway, governments have struggled to convince their citizens to
download the CTA. In Singapore only 20% of the residents have installed the CTA [9] and
in Norway the percentage of people that downloaded the app two weeks after its release
was even lower (16%) [10]. Recent surveys highlight privacy, data storage and battery use
concerns as the prime issues preventing people from downloading CTAs [11, 12]. Against
this background, a lively debate sparked on how useful CTAs can be in preventing the
spread of COVID-19. Some scholars suggest that they are an invaluable tool to contain
COVID-19 diffusion [4], whereas others argue that they are largely ineffective [13, 8].
Most of these studies, however, consider people’s behavior as exogenous. That is, they
implicitly assume that how often people wear a mask, how often they attend gatherings
and how often they leave their home will not be affected by the existence and adoption of
a CTA. In this vein, they compare the status quo with a world in which people behave in
exactly the same manner, except for having downloaded a CTA. However, not anticipating
behavioral responses is an important shortcoming of these studies as a large body of
research shows that whether people wear a mask, attend large gatherings and practice
social distancing is a fundamental determinant of how fast COVID-19 can spread [14].
In this article, we challenge this assumption and suggest that CTAs are likely to affect
how people behave.
In particular, we suggest that the effect of CTAs on behaviors could go in two direc-

3
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

tions. On the one hand, a person might engage in more risky behaviors when there is a
CTA that can mitigate the diffusion of the virus and inform the person when she enters
in contact with someone who tested positive. This is the so-called Peltzman effect [15].
The beneficial effects of strategies aimed at reducing the level of risk to which a certain
population is exposed would be offset by increased risk-taking by the target population.
In other words, after installing the app, people might feel safer and hence wear masks
less often or meet more people. If this hypothesis is verified, CTAs might even contribute
to the diffusion of the virus.
On the other hand, in app-notifications from the CTA could reduce risk taking. In
fact, insights from behavioral science suggest that carefully devised messages can induce
people to engage in pro-social behavior [16, 17, 18]. The CTA could, for instance, notify
the individuals when they are getting in contact with too many people, and hence might
be contributing to the spread of the virus. If this hypothesis is correct, then CTAs can
become more than just a complement to human contact tracing. They could become a
precious communication channel between health authorities and the general public that
can be used to promote prudent behaviors.
While our results offer no support to the Peltzman effect, we find that in-app notifica-
tions can be really effective tools to promote pro-social behaviors and to induce people to
take more precautions. Consequently, we suggest that CTAs should be re-conceptualized
as Behavioral Feedback Apps (BFAs) to account for this finding. The main function
of BFAs would be providing users with information and feedback on how to behave to
minimize the risk of contracting COVID-19. To carry out this function the app would
not need to track the movements of the users but should provide users with useful infor-
mation. Users, however, should be able to activate and deactivate the tracing function
at any time. In this vein, contact tracing would become an ancillary service provided by
BFAs, and only for users that decide to activate this function.
In particular, we suggest that BFAs should inform users on the level of risk of areas in
which she is interested, based on how likely it is that the area will be crowded and on the
number of cases. Some CTAs already provide similar information [19]. Additionally, the
BFAs could allow users to rate businesses they visit based on whether such businesses

4
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

are complying with mask-wearing and social distancing norms. Many apps that have
introduced a rating system (e.g. TripAdvisors) are widely used, thus suggesting that
people are inclined to engage with rating systems. From this perspective, the app would
serve five purposes. First, it would directly protect the app user by giving information on
which areas and which businesses to avoid. Second, it would incentivize business owners
and their employees to respect social distancing rules to avoid negative reviews. Third,
it would facilitate government enforcement, since public authorities would be warned
about businesses that are recidivist violators of norms aimed at minimizing the spread of
COVID-19. Fourth, the app would protect the community because people following its
advice will reduce the potential for contacts in crowded areas. Last, using the app in this
way could help creating a bond between the users and the app, which might increase the
likelihood that the user will activate the tracing function.
The tracing function would work like present-day CTAs, but would also include tai-
lored in-app notifications like the ones in Fig. 3. Consequently, the app could send
information as “You are taking less risks than your peers. You are protecting your com-
munity” or “You are among the users that took the most risks. You could become a
superspreader”.
Importantly, BFAs would provide useful guidance and feedback even if the percentage
of people that download it and activate the tracing function is relatively small. In fact,
behavioral guidance does not require a critical mass of downloads comparable to that
required by contact tracing.

2. Background Information

2.1. Contact Tracing Apps Around the World

Singapore was the first country to introduce a CTA. In March 2020 it released Trace-
Together, which uses Bluetooth technology to inform users when they have been in contact
with someone who tested positive for COVID-19 [20]. The results produced by TraceTo-
gether have not been fully satisfying. Only 20% of the residents downloaded it [9], and
the app has some problems with Apple operating systems (iOS) [20]. Despite these issues,
many countries followed Singapore’s lead and implemented their own app. For instance,

5
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Australia introduced the app CovidSafe, which shares many features with TraceTogether.
To this date, over 6 million people – or about one quarter of the population – have down-
loaded it. But many have questioned its utility, given that the numbers of cases that it
helped identifying appears to be small. One of the reasons for this is that the app does
not register proximity between two cellphones if the screen is locked [12].
Also many European countries have embraced enthusiastically the idea of using CTAs
to complement the efforts of their healthcare workers. In June, France, Germany and
Italy, the three biggest economies of continental Europe, all launched their version of a
CTA.
France released the app StopCovid in June, but two months after its release less than
two million people – or roughly 3% of the population – had downloaded it [21]. In an
attempt to increase the number of users, French authorities attempted to inform the
public about the importance of the app. For instance, Olivier Ve’ran, France Health
Ministry, tweeted that people can use StopCovid to protect themselves and the people
they meet during their holidays. Meanwhile, however, hundreds of thousands of people
seem to have uninstalled the app because it consumes too much battery [12].
In Germany the introduction of a CTA was sponsored from the very beginning by the
most important political leaders. In an official video-message, Angela Merkel said that
the Corona-Warn app is a companion and protector and an important helper that will
benefit the community [22]. The spokesman of the German government also stressed that
downloading the app is both in the interest of the individuals and of the community and
that the app can be an effective tool to help containing the spread of COVID-19 [23]. In
response to these calls roughly 16 million people have downloaded Corona-Warn, which
is about 20% of the population.
In a similar fashion, key figures of the Italian government spoke in support of the
Italian CTA Immuni. The Italian prime minister Giuseppe Conte, in an official message
to the citizens, noted that the app is safe and that it respects the user’s privacy [24]. Like
his European counterparts he emphasized that downloading the app is entirely voluntary,
but that by downloading the app citizens can help containing the spread of the virus [25].
By the 5th of August 2020, Immuni had been downloaded 4.6 million times. Moreover,

6
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

according to the Italian Ministry of Innovation it had helped breaking the chain of conta-
gions in two key circumstances by allowing health authorities to identify 63 people that
were positive to COVID-19 [26].
In the United States, the Federal government did not show the same kind of support
to CTAs. There have been a number of attempts but, even when they were promoted
by tech giants like Google, CTAs failed to reach a critical mass of users [27]. At the
state level, however, there is more support for these apps. For instance, the Governor
of Virginia has strongly endorsed COVIDWISE, which is the first statewide app to use
the technology developed by Google and Apple. He argued that the app “can really help
us catch new cases early before they spread as far” [28]. Alabama, North Dakota and
South Carolina also expressed interest in the technology developed by Google and Apple
as a means to contain the spread of the virus [29]. Rhode Island went in a different
direction, instead, and supported the use of the “homegrown app” Crush COVID RI,
which, according to Governor Raimondo, is a “a tool that helps everybody in Rhode
Island get through the crisis” [30]. The Governor also emphasized that the app “protects
people’s privacy and data in an ironclad way” [30]. About 60.000 people decided to
download the app (about 5% of the population), while Raimondo is hoping to reach
100.000 downloads. Other States like North Dakota, South Dakota, Utah and Wyoming
also developed their version of the app.
However, citizens’ response has been tepid at best: in North Dakota and in Utah
respectively less than 5% and 3% of the population downloaded the app [31], a far cry
from the critical mass epidemiologists suggest is needed to make it useful. The result was
so disappointing that Utah authorities decided to turn off the location tracking function,
and to use the app only as a way to communicate COVID-19 related information with
the population.
This brief overview reveals that policymakers are keenly aware of the importance of
reaching a critical mass of downloads, but reaching that critical mass has proven to be
problematic. For this reason, most of the communication from political leaders and health
authorities has been aimed at increasing the number of downloads by emphasizing the
role that the CTA can play in containing the spread of the virus and by reassuring the

7
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

public that CTAs respect users’ privacy.

2.2. Literature Review

The literature on CTAs has mushroomed during the COVID-19 pandemic. Studies
have focused around three main issues. First, many investigated how to best develop
and improve these apps from a technical perspective e.g. [32]. A second group of studies
analyzed the privacy problems posed by CTAs [33], and how to minimize them [34]. A
last strand of literature investigated what influences people decision to download the app.
For instance, Abeler et al [35] find that people might resist downloading the app because:
(i) they fear that the government might keep monitoring citizens also after the pandemic
is over, (ii) they are concerned that it might make them more anxious and (iii) they are
afraid that their cellphone might get hacked. Frimpong and Helleringer [36] study how
financial incentives influence the decision to download the app and find that they can be
effective. It is unclear, however, how to ensure that people do not uninstall the CTA after
receiving the financial incentive. Kaptchuck et al. [37], instead, carry out a survey with
4500 Americans to investigate which attributes of the app they value the most. They find
that ensuring that the app is “perfectly private” greatly increases the number of people
who are willing to install it. They also find that Americans value how much the app
can contribute to public health, and that false negatives have a larger effect than false
positives on reducing the willingness to download the app. Last, using a sample of 1117
Australians, Bradshaw et al. [11] find that data safety is an important determinant of
whether people download the app.
Notwithstanding the importance of these works, here we focus on a different issue that
to the best of our knowledge has been largely overlooked. Specifically, we study whether
people’s behavior is affected by the information governments give about the app in the
pre-download phase, and by the in-app notifications in the post-download phase.
Thinking about the communication by public authorities in the pre-download phase,
it is possible that excessively positive descriptions of the app could induce too much
optimism in the general public. Consequently, people might feel protected by the app
and engage in activities that increase the risk of contracting the virus like attending small
and large gatherings. This would be in line with the risk compensation theory, which
8
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

posits that when people feel more protected they will engage in more risky behaviors. For
example, [38] and [39] find that the presence of pre-exposure prophylaxis – a strategy that
helps protecting against HIV – induces people to engage in more condomless sex. There is
some evidence in support of the risk compensation theory also in the context of COVID-
19. Cartaud et al. [40] find that people wearing mask preferred a lower interpersonal
distance than people without masks. Similarly, Yan et al. [41] find that people living
in American States that mandated the use of masks spent less time at home and visited
more commercial businesses. If the risk compensation theory holds true in this context,
it would pose a dilemma for health authorities: promoting the app emphasizing only
its advantages, like the German government did, may induce more people to download
it. However, people might then engage in more risky activities. Vice versa, if health
authorities flag also the downsides of the app, people might refrain from engaging in
more risky activities, but the number of downloads might be too small for the app to
become an effective tool in containing the spread of the disease.
Additionally, the app opens a communication channel between the health authorities
and the app users. Health authorities can then use in-app messages to induce people to
engage in pro-social behavior. For example, Ayres et al. [42] and Allcott [43] find that
informing people when they are consuming more energy than their neighbors can lead to
considerable savings. Additionally, Ferraro and Price find that a similar nudge can help
reducing water consumption [44]. Therefore, we hypothesize that informing people about
how much risk they are taking in comparison to other users might induce app users to
engage in more pro-social behaviors like wearing mask or staying home. In the context
of COVID-19, researchers are converging on the idea that 20% of the people cause 80%
of all COVID-19 cases (i.e. superspreaders) [45, 46]. Consequently, we hypothesize that
informing people that they might be a superspreader could be an effective way to induce
them to engage in more pro-social behavior.

3. Materials and Methods

We devised a double-blind experiment approved by the Yale IRB divided in two

phases. Participants were informed at the beginning of Phase I that the experiment was

9
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

composed of two phases. Only those who participated in Phase I would be allowed to
take part in Phase II. Participants were paid $1 for taking part in Phase I and $0.75 for
taking part in Phase II. In addition, we also informed them that by participating in both
Phases they would be automatically entered into a lottery with three prizes of $100 and
one of $200.
Phase I aimed at studying the impact on behaviors of various kinds of communica-
tion strategies in the pre-download phase. Consequently, the underlying assumption is
that participants do not have the CTA on their cellphone. Participants in the treatment
groups are then provided with information about the app, devised to mimic governments’
campaigns that were run around the world. Phase II was carried out one week after Phase
I and aimed at studying the impact of in-app notifications on behaviors. Therefore, par-
ticipants were asked to imagine that they had downloaded the app, and were exposed to
different kinds of in-app notifications. To improve the degree of realism of the experi-
ment, all the in-app notifications use a similar graphic to the Italian CTA Immuni. We
built on this CTA because its developers have shared all their code and screenshots of
the notifications on GitHub.

3.1. Phase I: Design

We recruited a sample of n = 1500 U.S. residents on Prolific. In Phase I respondents

were randomly assigned to one of three different groups: Pros , Pros and Cons and
Control I. The respondents included in the Pros group were shown information on the
advantages of CTAs (Fig. 1). The respondents in the Pros and Cons group were shown
information on both the advantages (Fig. 1) and the problems that characterize CTAs
(Fig. 2). Last, respondents in Control I group were not given any information on CTAs.
We then asked the same questions to all participants.
To begin with, we asked respondents how likely they were to engage in some behaviors
(outside of the workplace) in the coming week (hereinafter “behavioral questions”). In
particular, we asked how likely they were to: (i ) attend small gatherings, (ii ) attend large
gatherings, (iii ) see older relatives or people at risk, (iv ) stay home unless they need to go
for groceries/shopping for essential items, and (v) wear a mask. In the following screen
we asked them how worried they are of contracting COVID-19. All these questions were
10
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Figure 1: Treatment I, Phase I: Participants see some of the app’s pros.

asked on a Likert scale from 1 to 10 where 1 meant “Absolutely certain that you will not
engage in this activity” and 10 “Absolutely certain that you will engage in this activity”.
After this, we asked respondents how likely they were to download the app (5-point Likert
scale from “Extremely Unlikely” to “Extremely Likely”). To the people that said that
they were neither likely nor unlikely, unlikely, or extremely unlikely to download the app
we also asked the amount of money which would make them willing to accept downloading
the app. We then asked which percentage of people they thought were going to download
the app, and whether they agreed that CTAs were important to contain the spread of
COVID-19 (5-point Likert scale from “Strongly Disagree” to “Strongly Agree”).
Lastly, we asked standard demographic questions and other questions related with
11
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Figure 2: Treatment II, Phase I: Participants see some of the pros and cons of the app.

how much respondents trust various actors, and whether they thought that their State
and the Federal Government had handled the COVID-19 crisis properly. We asked trust
towards: people known personally, people not known personally, US government, health
insurance companies, the media, the CDC and the healthcare system. As these measures
are all related and can affect the response to the treatments, we used principal component
analysis to construct a trust index.

Phase II: Design

A week later we carried out Phase II. Out of the initial 1500 participants from the
first round 1303 also completed Phase II. In this part of the experiment we first asked
participants how they behaved the previous week using the same behavioral questions
12
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Figure 3: Treatment I, Phase II: Participants are informed that they are taking less risk than the
average user and protecting their community (Left Panel). Treatment II, Phase II: Participants are
informed that they are taking more risk than the average person and endangering their community
(center panel). Treatment III, Phase II: Participants are informed that due to their behavior they might
become superspreaders (right panel)

13
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

asked in Phase I. Then we asked participants to imagine that they had downloaded the
app. Participants were then randomly assigned to four groups: Good Behavior, Bad
Behavior, Superspreader and Control II. Fig. 3 shows the three treatments.
The participants in the Good Behavior group were shown an in-app notification that
read “This week you have taken less risks than average. You are protecting yourself
and your community” on a green screen with a smiling face. The participants in the
Bad Behavior group were shown an in-app notification that read “This week you have
taken more risks than average. You are endangering yourself and your community” on
an orange screen with an unhappy face. The participants in the Superspreader group
received a notification that read “Due to your behavior you could be a superspreader
for Covid-19 and endanger your community. Remember 20% of the people cause 80% of
transmissions” on a red screen with a worried face. Last, the respondents in Control II
group were not shown any notification. After this screen, we asked to all participants
how they intended to behave in the coming week using the same behavioral questions
asked in Round I. We also asked them how long for they were planning to keep the app.
Afterwards, we showed to all participants a notification that they had exposed to
the virus (Fig. 4 left panel), and another screen suggesting how to behave following
the contact (Fig. 4 right panel). In particular, the app suggested to: (i ) wash hands
frequently with soap, (ii ) cough and sneeze into a a tissue or in the crook of the elbow,
(iii ) assess symptoms and take temperature twice per day and (iv ) stay home and practice
social distancing for 14 days since the contact. These recommendations were all based on
those given by the Italian app Immuni and are in line with suggestions from the WHO and
US Center for Diseases Control and Prevention (CDC) [47]. We then asked respondents
on a Likert scale from 1 to 10 how likely they were to follow each recommendation of the
app.
Last, we turned to questions related to the prevention optimism based on the CTA
and asked respondents how much they agreed with certain statements (see Table 1).
These statements were adapted from the medical literature on prevention optimism for
HIV related to Pre-exposure Prophylaxis [38].

14
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Figure 4: Warning message for a potential contact (left panel). The recommendations given by the app
after the warning (right panel)

15
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

4. Results of Phase 1

In Phase 1 respondents are randomly assigned to three groups: Pros, Pros and Cons
and Control I. To the Pros group we showed the pros of the app (Fig. 1), to the Pros
and Cons we showed both the strengths and the weaknesses of the app (Fig. 2), while
we provided no information on the app to Control I group.
We observe that participants in the treated groups are less worried about the pandemic
(Table 5). The result is robust and significant at 1% for the Pro group, whereas it is
smaller, less robust and significant only at 10% for the Pros and Cons group. These
results are consistent with the idea that a CTA can be reassuring. In fact, the effect is
stronger if only the advantages of the app are emphasized.
We observe that the two treatments have no significant effects on behaviors. This
suggests that communicating just the pros of the app, or also its cons, will not influence
how people behave after having downloaded the app.
At the same time, we find that both treatments increase the likelihood to download
the app (Table 6). Notably, the result is more robust for the Pros and Cons treatment,
suggesting that communicating also the downsides of the app can promote downloads.
This suggests that governments that only emphasized the virtues of CTAs might have
chosen the wrong communicative strategy if their goal was to maximize the number of
downloads.
In line with previous papers analyzing the drivers of adoption [48, 49, 33], we find that
people with higher levels of trust are more willing to download the app. We also find that
treatments have different effects depending on the political ideology of the respondent.
In particular, we find that republicans in the treatment groups are less willing download
the app.
Notably, we find that about 56% of our respondents are willing to download the app
even without a payment. Moreover, we observe that respondents stating that they are
less likely to download the app want more money to download the app.

16
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

5. Phase 2

In Phase 2 the participants were randomly assigned to three treatment groups (Fig.
3 Good Behavior, Bad Behavior, Superspreader and Control II group.

5.1. Behaviors after Phase I

Respondents are first asked about their behaviors in the previous week, with the same
behavioral questions asked the week before. In addition we asked two more questions on
how many contacts they had during the last 7 days. We do not find differences in behavior
among groups, with the exception of masks use. People who were in the treatment groups
wore masks more often than the control group (significant at 5%, or 1% depending on
the sets of controls) (Table 7). Overall, the main finding is that communication in the
pre-download phase affects how worried people are, but has a small impact on how they
behave.

5.2. In-app Notifications

Behavioral feedback messages have a significant effect on how people behave. Here we
have three treatments and one control group. To one group we show a green screen saying
that they have taken less risks than average (Good Behavior), to one group we show an
orange screen saying that they have taken more risks than average (Bad Behavior) and
to the last group we show a red screen saying that due to their behavior they might be
super-spreaders (Superspreaders) and explain what super-spreaders are (Fig. 3).
We find that:

• Large Gatherings: Respondents in the Superspreaders group attend less gatherings

than any other group. The result is significant at 5% (Table 8). There are no
significant differences between Respondents in Good and Bad Behavior and Control
group.

• Small Gatherings: Respondents in the Superspreaders group attend less small gath-
erings than any other group (significant at 1 %) (Table 9). No significant impact
of the Good and the Bad behavior treatments.

17
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

• Seeing people at risk : Respondents in the Superspreaders group see less people at
risk than any other group (significant at 1 %) (Table 10). No significant impact of
the Good and the Bad behavior treatments.

• Stay Home: Respondents in the Superspreaders group stay at home more often
than any other group (significant at 1 %) (Table 11). No significant impact of
Good and the Bad behavior treatments.

• Masks: The Superspreaders treatment has a positive effect on mask wearing, but
the statistical significance depends on the set of controls used (ranging from 1 % to
not significant) (Table 12).

Overall, these results point very strongly in the direction that behavioral feedback
messages can be effectively used to induce prudent behavior among app users. In partic-
ular, informing the people who are moving the most or are having the most contacts that
they could be superspreaders can induce them to stay home more often, to attend less
large and small gatherings and to see less people at risk. To the best of our knowledge, no
app gives this information, therefore our findings provide a concrete suggestion on how
to improve the effectiveness of CTAs in containing the spread of COVID-19.
Comparative messages (Good and Bad behavior) might also be helpful, as the people
who receive a Good behavior message keep the app for longer (Table 13). This effect is
relatively weak and not always significant, but including comparative messages should be
virtually cost free, hence it might be worthwhile even if the benefit is small. Moreover,
app users might respond to this type of feedback by improving their behavior in order to
be recognized as being better than average.
In line with the rest of our results we find that the effect of in-app notification depends
on the characteristics of the respondents. For example, males that are in the Good
behavior group will attend less small gatherings. Similarly, males that receive the Good
and the Bad behavior treatment are less likely to see people at risk than males in the
control group. Last, respondents in the Superspreaders group with a Professional or
Master’s degree use mask much more.
However, in-app notifications sometimes triggered counterproductive responses. For

18
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

instance, Republicans that see the Bad behavior treatment will see more people at risk.
This result underscores the importance of tailoring in-app notifications to the character-
istics of the app user.

6. COVID-19 Prevention Optimism

We now turn to analyze the results on the questions related with COVID-19 prevention
optimism based on the CTA (Tables 1 and 14–17).

Table 1: The table summarizes the effects of the various treatments on the level of agreements with
statements aimed at identifying prevention optimism. * significant at 10%, ** significant at 5% and ***
significant at 1%.

Statement Agrees More Agrees Less

Because of the contact trac- Good Pros and Cons**
ing app I am less likely to Behavior**
contract COVID-19
It is safe for me to be around - Pros and Cons**,
people because of the con- Bad Behavior**,
tact tracing app Superspreaders**
Because of the contact trac- Good -
ing app less people will be Behavior**
infected by COVID-19
If many people download - Pros and Cons*
the contact tracing app,
wearing a mask becomes
less important

These results offer some support to the idea that CTAs can cause prevention opti-
mism for respondents that are included in the Good Behavior treatment. Vice versa, the
respondents that were included in the Pros and Cons, Bad Behavior and Superspreaders
treatment agree less than the Control group with some of the statements that capture
19
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

prevention optimism. Consequently, a communication that is transparent about the cons

of the app and in-app notifications signaling dangerous behavior create prevention pes-
simism, which could result in more prudent behavior. This confirms that in-app notifica-
tions can be effectively used to induce prudent behavior among app users. Additionally,
we find that the treatments impact essential workers differently. More specifically, essen-
tial workers that are in the Bad Behavior and in the Superspreader groups agree less that
the app reduces the amount of infected people and the need for wearing masks.

6.1. Post warning

Last, we study how much people comply with the recommendations given by the CTA
after it notifies them that they have been in contact with a person who tested positive
(Tables 2 and 18–20).
We find that our results are consistent with standard findings of the literature. For
instance, male subjects reported that they would be less likely to sneeze into the crook of
their elbows (p < 0.01) and generally speaking reported that they would take more risk.
Also in this case the impact of different messages depends on the characteristics of
the app user. We find that essential workers in the Bad Behavior group are more likely
to wash their hands often, to measure more frequently their temperature and to respect
social distancing. The level of trust that participants declare also affects how they react
to these recommendations. For instance, respondents with high trust state that they
would respect more social distancing, would assess their temperature more often and
would sneeze more frequently in the crook of their elbow. We find also that younger
people in the Good Behavior, Bad Behavior and Superspreaders group will assess their
temperatures less often and those in the Good Behavior group stay home less.

7. Discussion

In most countries the penetration rate of CTAs is extremely low, and hence their
success has been limited. Is this the end of the story for CTAs? Not necessarily. Our
results suggest a possible way forward based on the insights of behavioral economics.
To begin with, information on the CTA reduces how worried people are about the
pandemic, without increasing the amount of risky activities in which they engage (Table
20
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 2: The table summarizes the effects of the various treatments on the intention to comply with
the recommendations given by the app after it notifies the user of a contact with a person who tested
positive to COVID-19. * significant at 10%, ** significant at 5% and *** significant at 1%.

Recommendation Follows More Follows Less

Wash hands frequently Good Pros, Pros and
with soap and water Behavior** Cons**
Cough and sneeze directly - Pros**
into a tissue or in the crook
of your elbow?
Assess your temperature at - -
least twice daily?
Stay home for at least 14 - -
days since your last expo-
sure?
Maintain social distance (at - Pros**
least six feet) from others at
all time?

21
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

5). Given the devastating impact that COVID-19 is having on the mental well-being of
many people [51, 52], the importance of this effect should not be underestimated. Re-
ducing how worried people are about the pandemic is in itself a desirable goal, especially
when this reduction is not accompanied by increased risk taking.
Second, we find that a more transparent communication that also emphasizes the
downsides of the CTA can increase the number of downloads (Table 6). One potential
explanation for this finding is that people find more credible a communication that also
touches on the weaknesses of the app. To put it differently, authorities might be more
credible when they guarantee that CTAs protect privacy if they have been transparent
about problems like false positive and false negatives that are likely to constitute an issue
[19].
However, the most important finding of our study is that providing users with useful
information, such as in-app notifications on their current level of risk-taking, can signif-
icantly alter people’s behaviors. Respondents that were included in the Superspreader
group were significantly less likely to attend large and small gatherings and to see people
at risk, while they were significantly more likely to stay home. This suggests that CTAs
can play an important role in promoting pro-social behavior during the pandemic. While
also this function is best carried out when many people download the app, it does not
require a minimum number of downloads. For instance, consider the case of France. As
roughly 3% of the population (about 2 million people) downloaded the CTA, the number
of contacts that it can identify is likely to be minimal because the likelihood that two
people (infected and potentially infected) that meet both have the app is extremely small.
In fact, in the first three weeks of its existence the app only notified 18 people that they
had been exposed to COVID-19 [12]. This number is too small to have a large impact on
how the virus spreads. However, inducing 2 million people to take less risk via tailored
in-app messages could help containing the spread of COVID-19.
Consequently, we argue that CTAs should be re-framed as Behavioral Feedback Apps
and their main function should be providing users with useful information on how to
behave to minimize the risk of being infected with COVID-19. Contact tracing should
then become an ancillary function of these BFAs, and it should be performed only for

22
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

users that decide to activate it.

The BFA could indicate how crowded an area is likely to be based on information like
the percentage of hotels and bed and breakfasts booked on platforms like Booking.com
and AirBnB.com. Another important function that the BFA could perform is a rating
system. Similarly to what apps like TripAdvisor do, the BFA might allow users to rate
businesses based on how much they respect the rules on mask use, social distancing, and
so on. This would leverage the wisdom of the crowd to identify the safest businesses, and
might push business owners and employees to follow safety norms more closely. Online
reviews have already been proposed as a tool to determine in which restaurants hygiene
inspections should be carried out [50]. The wisdom of the crowd might be even more
effective in this context since it is very easy to determine whether people inside a store
are wearing a mask, whereas for a customer it is hard to assess the level of hygiene of
a restaurant’s kitchen. Therefore, businesses’ ratings and information on which areas
are safer to visit can offer precious guidance to the app users, and could help public
authorities identifying the businesses that systematically violate safety norms. Moreover,
receiving valuable information from the app could contribute to create a bond between
the users and the app, which might increase the trust that the user has in the app.
BFAs should also have a tracing function akin to the one that CTAs currently have,
and the user should be allowed to activate it at any moment. If the user chooses to do so,
the app should also keep track of how much a user has moved – or how many contacts
she has had – compared to similar users. This way, the app would be able to give tailored
advice such as “You are taking less risks than your peers. You are protecting your
community” or “You are among the users that took the most risks. You could become
a superspreader”. Our results show that behavioral feedback messages of this kind can
effectively induce people to engage in more prudent behavior.

8. Future Research

CTAs are failing in virtually every country in which they have been introduced. Sci-
entists are attempting to find better technology that could marginally reduce the privacy
risks or the number of false positives and false negatives. We argue that the reasons

23
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

behind this failure lie elsewhere, and marginal technological improvements alone cannot
turn the table around. CTAs touch fundamental individual rights, and their success ul-
timately depends on the fact that people recognize their importance and accept to use
them. Consequently, the way forward is to look at behavioral sciences.
We have shown that in-app notifications based on the insights of behavioral sciences
can lead to more prudent behavior. But behavioral sciences can inform many aspects
of CTAs development, as even small differences in how COVID-19 related information is
presented can have a big impact on how people react [53]. Consequently, we suggest that
the CTAs should be abandoned in favor of Behavioral Feedback Apps. To begin with,
behavioral Feedback App is a less threatening name that does not immediately evoke
monitoring by the government. Moreover, it would be more appropriate to describe the
functioning of the apps as we envisage them, as contact tracing, besides being entirely
optional, would be only one of the functions performed by apps. Future studies could
test whether this re-branding of the app could increase the number of downloads, and
how often people using a BFA would activate a tracing function.
Additionally, it is possible to devise mechanisms of carrots and sticks to incentivize
users to activate the tracing function. For example, as a carrot the governments could
partner with businesses to provide incentives and discounts to customers that activated
contact tracing when they visit stores. As a stick, stores and restaurants could be given
the possibility to refuse entrance to customers that do not have the contact tracing
function activated. This can be achieved via a Bluetooth Handshake that has already
been introduced in some CTAs [19]. While this solution would push people to install
the app, it is likely to be perceived very differently from a government imposition. In
fact, if the adoption came from businesses it could be interpreted by many people as the
sign that business owners care for the safety of their employees and customers and could
contribute to create a climate of trust among people, and between people and the app.
On the contrary, a similar rule introduced by the government would be seen in many
countries as an unacceptable violation of freedom.

24
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

9. Tables

25
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 3: Frequency Table for Demographic Variables: Number, Percentage and Cumulative Percentage of
respondents for the following variables: Age, Education, Income, Political orientation, Gender. Column
1 presents the statistics for the control group, Column 2 for those participants seeing only the pros about
the app, Column 3 only for participants who see pros and cons about the app and Column 4 for the
entire sample.

(l1em)2-10 Control Group Pros Only Info Pros and Cons Total
No. Col Cum No. Col Cum No. Col Cum No. Col Cum
% % % % % % % %
Age
18-25 years old 111 21.9 21.94 129 25.6 25.60 101 20.0 19.96 341 22.5 22.49
26-35 years old 162 32.0 53.95 152 30.2 55.75 163 32.2 52.17 477 31.5 53.96
36-45 years old 137 27.1 81.03 129 25.6 81.35 130 25.7 77.87 396 26.1 80.08
46-55 years old 46 9.1 90.12 41 8.1 89.48 50 9.9 87.75 137 9.0 89.12
56-65 years old 23 4.5 94.66 23 4.6 94.05 36 7.1 94.86 82 5.4 94.53
66-75 years old 16 3.2 97.83 18 3.6 97.62 15 3.0 97.83 49 3.2 97.76
>75 years old 11 2.2 100.00 12 2.4 100.00 11 2.2 100.00 34 2.2 100.00
Total 506 100.0 504 100.0 506 100.0 1516 100.0
Education
Less than high school degree 2 0.4 0.40 4 0.8 0.80 2 0.4 0.40 8 0.5 0.54
High school graduate or equivalent 47 9.5 9.90 34 6.8 7.65 49 9.8 10.20 130 8.7 9.25
Some college but no degree 107 21.6 31.52 105 21.1 28.77 100 20.0 30.20 312 20.9 30.16
Associate degree in college (2-year) 29 5.9 37.37 24 4.8 33.60 34 6.8 37.00 87 5.8 35.99
Bachelor’s degree in college 173 34.9 72.32 179 36.0 69.62 158 31.6 68.60 510 34.2 70.17
Master’s or Professional Degree 120 24.2 96.57 134 27.0 96.58 137 27.4 96.00 391 26.2 96.38
Doctoral degree 17 3.4 100.00 17 3.4 100.00 20 4.0 100.00 54 3.6 100.00
Total 495 100.0 497 100.0 500 100.0 1492 100.0
Income
Less than $10,000 30 6.1 6.13 29 5.8 5.85 23 4.7 4.68 82 5.6 5.56
$10,000 - $19,999 31 6.3 12.47 33 6.7 12.50 33 6.7 11.41 97 6.6 12.13
$20,000 - $29,999 39 8.0 20.45 47 9.5 21.98 28 5.7 17.11 114 7.7 19.85
$30,000 - $39,999 47 9.6 30.06 41 8.3 30.24 57 11.6 28.72 145 9.8 29.67
$40,000 - $49,999 36 7.4 37.42 43 8.7 38.91 46 9.4 38.09 125 8.5 38.14
$50,000 - $59,999 39 8.0 45.40 50 10.1 48.99 42 8.6 46.64 131 8.9 47.02
$60,000 - $69,999 36 7.4 52.76 30 6.0 55.04 38 7.7 54.38 104 7.0 54.07
$70,000 - $79,999 34 7.0 59.71 37 7.5 62.50 40 8.1 62.53 111 7.5 61.59
$80,000 - $89,999 16 3.3 62.99 20 4.0 66.53 22 4.5 67.01 58 3.9 65.51
$80,000 - $89,999 33 6.7 69.73 28 5.6 72.18 27 5.5 72.51 88 6.0 71.48
$90,000 - $99,999 100 20.4 90.18 105 21.2 93.35 91 18.5 91.04 296 20.1 91.53
$150,000 or more 48 9.8 100.00 33 6.7 100.00 44 9.0 100.00 125 8.5 100.00
Total 489 100.0 496 100.0 491 100.0 1476 100.0
Political Orientation 26
Other 28 5.5 5.53 32 6.3 6.35 26 5.1 5.14 86 5.7 5.67
Democrat 229 45.3 50.79 229 45.4 51.79 228 45.1 50.20 686 45.3 50.92
Republican 119 23.5 74.31 108 21.4 73.21 118 23.3 73.52 345 22.8 73.68
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 4: Summary statistics (mean and standard deviation (SD)) for the variables considered in the
regression tables. These include: Worry About Covid-19, stay at home whenever possible, attendance of
small meetings, attendance of large meetings, seeing people at risk, likelihood to wear a mask whenever
going out in the next week, gender, being an essential worker, education, economic situation, age,
republican, democrat, and all the measures of trust: in people known by the respondent, people not
known to the respondent, in the media, in insurance companies, in the healthcare system, in the US
government and in the CDC. Column 1 presents the statistics for the control group, Column 2 for those
participants seeing only the pros about the app, Column 3 only for participants who see pros and cons
about the app and Column 4 for the entire sample.

(All Sample) (Pros Only) (Pros and Cons) (Control)

mean sd mean sd mean sd mean sd

Worry About Covid-19 6.48 2.90 6.33 2.93 6.52 2.84 6.60 2.93
Stay at home 7.97 2.55 7.93 2.59 7.90 2.59 8.09 2.46
Attend Small Meetings 3.44 3.53 3.41 3.59 3.72 3.58 3.19 3.42
Attend Large Meetings 1.66 2.95 1.62 2.96 1.82 3.04 1.54 2.83
See People at Risk 2.76 3.38 2.81 3.47 2.91 3.47 2.56 3.19
Likelihood to Wear Masks Next Week 8.99 2.06 9.15 1.80 8.89 2.20 8.92 2.16
Male 0.54 0.50 0.53 0.50 0.54 0.50 0.54 0.50
Essential worker 0.30 0.46 0.27 0.45 0.32 0.47 0.30 0.46
Education 4.58 1.41 4.63 1.38 4.58 1.44 4.52 1.41
Economic situation 2.72 0.90 2.73 0.89 2.73 0.89 2.71 0.91
Age 36.04 12.60 35.40 12.71 36.91 12.82 35.82 12.25
Republican 0.23 0.42 0.21 0.41 0.23 0.42 0.24 0.42
Democrat 0.45 0.50 0.45 0.50 0.45 0.50 0.45 0.50
Trust: People known 8.05 1.76 8.16 1.63 8.09 1.83 7.91 1.82
Trust: People not known 3.54 2.63 3.58 2.64 3.64 2.73 3.40 2.52
Trust: Media 4.14 2.83 4.14 2.83 4.23 2.87 4.04 2.80
Trust: Insurance 3.99 2.90 3.90 2.91 4.18 2.90 3.88 2.89
Trust: Healthcare 5.59 2.73 5.60 2.65 5.85 2.74 5.32 2.78
Trust: US government 3.41 3.06 3.36 3.08 3.52 3.09 3.34 3.03
Trust: CDC 6.77 2.51 6.97 2.36 6.74 2.60 6.59 2.55
Observations 1516 504 506 506

27
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 5: Determinants of worry: combined controls, ordered logit estimates, robust standard errors

(1) (2) (3) (4)

Worry About Covid-19 Worry About Covid-19 Worry About Covid-19 Worry About Covid-19
Worry About Covid-19
Treatment 1: Only Positive Info -0.418∗∗∗ -0.378∗∗∗ -0.418∗∗∗ -0.378∗∗∗
(0.001) (0.004) (0.001) (0.004)
Treatment 2: Positive and Negative Info -0.230∗ -0.188 -0.230∗ -0.188
(0.083) (0.167) (0.083) (0.167)
essential 0.577∗∗∗ 0.584∗∗∗ 0.577∗∗∗ 0.584∗∗∗
(0.000) (0.000) (0.000) (0.000)
econsituation -0.348∗∗∗ -0.330∗∗∗ -0.348∗∗∗ -0.330∗∗∗
(0.000) (0.000) (0.000) (0.000)
Male -0.184 -0.179 -0.184 -0.179
(0.102) (0.125) (0.102) (0.125)
Education 0.0512 0.0332 0.0512 0.0332
(0.252) (0.488) (0.252) (0.488)
Age 0.0120∗∗∗ 0.0128∗∗∗ 0.0120∗∗∗ 0.0128∗∗∗
(0.008) (0.007) (0.008) (0.007)
Income -0.0134 -0.0222 -0.0134 -0.0222
(0.435) (0.229) (0.435) (0.229)
stateresponse -0.0759∗ -0.103∗∗ -0.0759∗ -0.103∗∗
(0.093) (0.030) (0.093) (0.030)
USresponse -0.191∗∗∗ -0.168∗∗∗ -0.191∗∗∗ -0.168∗∗∗
(0.002) (0.007) (0.002) (0.007)
likelydownload 0.261∗∗∗ 0.266∗∗∗ 0.261∗∗∗ 0.266∗∗∗
(0.000) (0.000) (0.000) (0.000)
importance 0.249∗∗∗ 0.225∗∗∗ 0.249∗∗∗ 0.225∗∗∗
(0.000) (0.001) (0.000) (0.001)
Usdownloads 0.0135∗∗∗ 0.0104∗∗∗ 0.0135∗∗∗ 0.0104∗∗∗
(0.000) (0.002) (0.000) (0.002)
trustUSgov -0.0160 -0.0383 -0.0160 -0.0383
(0.646) (0.296) (0.646) (0.296)
trustinsurance 0.0362 0.0113 0.0362 0.0113
(0.297) (0.765) (0.297) (0.765)
trustCDC 0.0837∗∗∗ 0.0495 0.0837∗∗∗ 0.0495
(0.003) (0.120) (0.003) (0.120)
Republican 0.307∗ 0.307∗
(0.081) (0.081)
Democrat 0.270∗∗ 0.270∗∗
(0.033) (0.033)
householdsize 0.0714 0.0714
(0.133) (0.133)
trustmedia 0.0822∗∗∗ 0.0822∗∗∗
(0.007) (0.007)
trusthealthcare 0.0169 0.0169
(0.579) (0.579)
/

(0.000) (0.000) (0.000) (0.000)

Observations 1158 1126 1158 1126

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

28
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 6: Determinants of download: combined controls, ordered logit estimates, robust standard errors

(1) (2) (3) (4) (5) (6)

likelydownload likelydownload likelydownload likelydownload likelydownload likelydownload
likelydownload
Treatment 1: Only Positive Info 0.247∗ 0.217 0.213 0.242∗ 0.185 0.178
(0.076) (0.124) (0.133) (0.085) (0.201) (0.220)
Treatment 2: Positive and Negative Info 0.288∗∗ 0.260∗ 0.277∗ 0.312∗∗ 0.256∗ 0.265∗
(0.038) (0.067) (0.052) (0.026) (0.078) (0.068)
Worry About Covid-19 0.140∗∗∗ 0.136∗∗∗ 0.123∗∗∗ 0.135∗∗∗ 0.130∗∗∗ 0.120∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
essential -0.193 -0.190 -0.209 -0.175 -0.168 -0.183
(0.179) (0.194) (0.164) (0.222) (0.260) (0.232)
econsituation -0.0662 -0.0688 -0.0761 -0.0959 -0.0932 -0.0946
(0.440) (0.425) (0.379) (0.267) (0.289) (0.286)
Male 0.113 0.152 0.123 0.0391 0.0573 0.0449
(0.340) (0.208) (0.321) (0.750) (0.652) (0.727)
Education 0.0529 0.0571 0.0485 0.0352 0.0354 0.0314
(0.228) (0.205) (0.290) (0.439) (0.457) (0.518)
Age -0.0155∗∗∗ -0.0146∗∗∗ -0.0141∗∗∗ -0.0154∗∗∗ -0.0148∗∗∗ -0.0143∗∗
(0.002) (0.006) (0.010) (0.003) (0.007) (0.011)
Income 0.0355∗ 0.0284 0.0217 0.0322∗ 0.0257 0.0206
(0.060) (0.154) (0.286) (0.096) (0.213) (0.328)
stateresponse 0.0157 0.00816 0.00310 0.00774 -0.0182 -0.0228
(0.749) (0.870) (0.951) (0.876) (0.723) (0.661)
USresponse 0.0498 0.0701 0.0720 -0.0654 -0.0653 -0.0552
(0.387) (0.276) (0.263) (0.399) (0.420) (0.492)
importance 1.114∗∗∗ 1.111∗∗∗ 1.093∗∗∗ 1.074∗∗∗ 1.096∗∗∗ 1.087∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Usdownloads 0.0299∗∗∗ 0.0303∗∗∗ 0.0294∗∗∗ 0.0289∗∗∗ 0.0293∗∗∗ 0.0285∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Republican -0.00770 -0.0550 -0.0339 -0.0784
(0.967) (0.772) (0.861) (0.694)
Democrat 0.201 0.193 0.0983 0.0950
(0.132) (0.149) (0.477) (0.492)
householdsize 0.0510 0.0401 0.0351 0.0244
(0.261) (0.387) (0.441) (0.601)
smallmeet 0.00635 0.00324
(0.806) (0.903)
largemeet 0.0418 0.0367
(0.274) (0.346)
seeatrisk 0.0302 0.0283
(0.227) (0.271)
Likelihood to Wear Masks Next Week 0.0753∗ 0.0640
(0.058) (0.123)
stayhome 0.00795 0.00208
(0.761) (0.939)
trustUSgov 0.0503 0.0610 0.0562
(0.206) (0.138) (0.182)
trustinsurance 0.0565 0.0412 0.0390
(0.106) (0.292) (0.324)
trustCDC 0.0493∗ 0.0289 0.0273
(0.088) (0.376) (0.415)
trustmedia 0.0209 0.0171
(0.524) (0.605)
trusthealthcare 0.0321 0.0312
(0.304) (0.325)
/

Observations 1169 1150 1144 1158 1126 1121

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

29
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 7: Determinants of wearing a mask: combined controls, robust standard errors

(1) (2) (3) (4) (5)

LWmask LWmask LWmask LWmask LWmask
LWmask
prosonly 0.446∗∗∗ 0.465∗∗∗ 0.421∗∗ 0.443∗∗ 0.153
(0.009) (0.007) (0.014) (0.011) (0.468)
prosandcons 0.370∗∗ 0.347∗∗ 0.387∗∗ 0.357∗∗ 0.386∗
(0.031) (0.044) (0.025) (0.039) (0.077)
worry 0.207∗∗∗ 0.202∗∗∗ 0.196∗∗∗ 0.194∗∗∗ 0.124∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.001)
essential -0.362∗∗ -0.338∗∗ -0.351∗∗ -0.325∗∗ -0.231
(0.022) (0.035) (0.027) (0.044) (0.268)
econsituation 0.0375 0.0225 0.0609 0.0547 0.0404
(0.694) (0.818) (0.526) (0.581) (0.739)
male -0.304∗∗ -0.251∗ -0.289∗∗ -0.186 -0.188
(0.033) (0.086) (0.047) (0.220) (0.314)
edu -0.00442 -0.0161 0.00706 -0.00250 -0.00235
(0.934) (0.772) (0.896) (0.965) (0.973)
age 0.00668 0.00613 0.00770 0.00660 0.00254
(0.233) (0.295) (0.184) (0.279) (0.751)
income -0.00221 -0.000787 0.0000929 -0.00350 -0.0261
(0.918) (0.973) (0.997) (0.882) (0.356)
stateresponse -0.0293 -0.0309 -0.0555 -0.0515 -0.0443
(0.603) (0.589) (0.334) (0.377) (0.539)
USresponse -0.189∗∗∗ -0.172∗∗ -0.0149 -0.0269 -0.0182
(0.002) (0.017) (0.856) (0.759) (0.871)
likelydownload 0.0391 0.0242 0.0597 0.0460 0.0569
(0.609) (0.755) (0.452) (0.574) (0.565)
importance 0.162∗∗ 0.165∗∗ 0.134∗ 0.148∗ 0.239∗∗
(0.027) (0.025) (0.072) (0.052) (0.022)
Usdownloads -0.00325 -0.00362 -0.00220 -0.00220 -0.00100
(0.430) (0.392) (0.606) (0.619) (0.855)
republican 0.317 0.441∗ 0.370
(0.166) (0.064) (0.223)
democrat 0.465∗∗∗ 0.487∗∗∗ 0.481∗∗
(0.003) (0.003) (0.016)
householdsize -0.0175 -0.0163 0.0220
(0.762) (0.776) (0.753)
trustUSgov -0.0893∗∗ -0.0966∗∗ -0.0841
30
(0.040) (0.036) (0.140)
trustinsurance -0.0377 -0.0387 0.0420
(0.365) (0.384) (0.422)
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 8: Determinants of large gatherings: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

NWlargemeet NWlargemeet NWlargemeet NWlargemeet NWlargemeet NWlargemeet NWlargemeet NWlargemeet NWlargemeet
NWlargemeet
worse -0.276 -0.230 -0.145 -0.140 -0.279 -0.228 -0.143 -0.133 -0.361
(0.216) (0.313) (0.523) (0.548) (0.214) (0.321) (0.531) (0.570) (0.224)
better -0.234 -0.246 -0.204 -0.237 -0.235 -0.247 -0.205 -0.239 -0.397
(0.313) (0.305) (0.379) (0.330) (0.311) (0.304) (0.377) (0.326) (0.202)
super -0.544∗∗ -0.502∗∗ -0.516∗∗ -0.493∗ -0.544∗∗ -0.500∗∗ -0.516∗∗ -0.494∗ -0.592∗
(0.029) (0.049) (0.039) (0.054) (0.029) (0.050) (0.039) (0.054) (0.053)
worry -0.0978∗∗∗ -0.0946∗∗∗ -0.0735∗∗ -0.0790∗∗ -0.0992∗∗∗ -0.0964∗∗∗ -0.0745∗∗ -0.0800∗∗ -0.0133
(0.002) (0.004) (0.020) (0.017) (0.002) (0.004) (0.018) (0.015) (0.762)
essential 0.582∗∗∗ 0.568∗∗∗ 0.583∗∗∗ 0.575∗∗∗ 0.581∗∗∗ 0.570∗∗∗ 0.584∗∗∗ 0.579∗∗∗ 0.366
(0.002) (0.003) (0.003) (0.004) (0.002) (0.003) (0.002) (0.003) (0.165)
econsituation 0.0515 0.0792 0.0176 0.0384 0.0536 0.0818 0.0190 0.0392 0.213
(0.680) (0.531) (0.888) (0.764) (0.667) (0.518) (0.880) (0.761) (0.156)
male 0.353∗∗ 0.399∗∗ 0.355∗∗ 0.360∗ 0.352∗∗ 0.395∗∗ 0.353∗∗ 0.358∗ 0.406∗
(0.043) (0.025) (0.048) (0.052) (0.044) (0.026) (0.049) (0.053) (0.090)
edu 0.142∗∗ 0.202∗∗∗ 0.135∗∗ 0.185∗∗∗ 0.144∗∗ 0.203∗∗∗ 0.135∗∗ 0.184∗∗∗ 0.178∗∗
(0.022) (0.002) (0.032) (0.006) (0.021) (0.002) (0.032) (0.007) (0.032)
age -0.0347∗∗∗ -0.0294∗∗∗ -0.0393∗∗∗ -0.0334∗∗∗ -0.0347∗∗∗ -0.0292∗∗∗ -0.0392∗∗∗ -0.0331∗∗∗ -0.0288∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.009)
income 0.0667∗∗ 0.0422 0.0677∗∗ 0.0474∗ 0.0656∗∗ 0.0409 0.0670∗∗ 0.0465 0.0538
(0.010) (0.127) (0.010) (0.094) (0.012) (0.141) (0.012) (0.103) (0.141)
stateresponse -0.0644 -0.0693 -0.0208 -0.0415 -0.0651 -0.0706 -0.0208 -0.0418 -0.0809
(0.383) (0.369) (0.790) (0.617) (0.377) (0.360) (0.790) (0.614) (0.437)
USresponse 0.524∗∗∗ 0.444∗∗∗ 0.312∗∗∗ 0.278∗∗∗ 0.525∗∗∗ 0.444∗∗∗ 0.311∗∗∗ 0.276∗∗∗ 0.275∗∗
(0.000) (0.000) (0.001) (0.006) (0.000) (0.000) (0.001) (0.006) (0.030)
likelydownload 0.105 0.0953 0.0803 0.0717 0.107 0.0980 0.0825 0.0738 0.187
(0.271) (0.326) (0.417) (0.479) (0.264) (0.314) (0.408) (0.470) (0.204)
importance -0.292∗∗∗ -0.303∗∗∗ -0.244∗∗∗ -0.291∗∗∗ -0.290∗∗∗ -0.301∗∗∗ -0.243∗∗∗ -0.291∗∗∗ -0.357∗∗
(0.001) (0.001) (0.009) (0.004) (0.001) (0.001) (0.009) (0.004) (0.024)
Usdownloads 0.0112∗∗∗ 0.0114∗∗∗ 0.00896∗∗ 0.00869∗ 0.0112∗∗∗ 0.0114∗∗∗ 0.00894∗∗ 0.00860∗ 0.00579
(0.009) (0.010) (0.047) (0.064) (0.009) (0.010) (0.048) (0.067) (0.374)
republican 0.335 0.233 0.333 0.233 0.321
(0.189) (0.366) (0.194) (0.367) (0.340)
democrat 0.0322 0.0212 0.0296 0.0237 0.0448
(0.875) (0.921) (0.886) (0.912) (0.854)
householdsize 0.164∗∗ 0.163∗∗ 0.166∗∗ 0.166∗∗ 0.126
(0.020) (0.026) (0.019) (0.023) (0.184)
trustUSgov 0.0886∗ 0.0840 0.0890∗ 0.0848 0.0500
(0.089) (0.122) (0.089) (0.121) (0.502)
trustinsurance 0.0974∗ 0.0941 0.0975∗ 0.0949 0.0163
(0.058) (0.102) (0.059) (0.101) (0.832)
trustCDC -0.169∗∗∗ -0.156∗∗∗ -0.170∗∗∗ -0.159∗∗∗ -0.0682
(0.000) (0.000) (0.000) (0.000) (0.321)
trustmedia 0.0786∗ 0.0780 0.0402
(0.100) (0.103) (0.521)
trusthealthcare -0.0619 -0.0601 -0.0272
(0.179) (0.192) (0.684)
prosonly -0.0837 -0.0965 -0.0476 -0.0438 0.162
(0.684) (0.643) (0.818) (0.834) (0.528)
prosandcons -0.0133 -0.0944 -0.0607 -0.124 -0.0109
(0.948) (0.649) (0.772) (0.561) (0.968)
applength -0.293∗∗∗
(0.000)
lesslikelytogetcovid 0.00379
(0.947)
Appmeanssafe 0.103∗∗
(0.041)
applessinfected -0.0302
(0.612)
Applessmasks 0.213∗∗∗
(0.000)
/

Observations 1017 1002 1006 982 1017 1002 1006 982 699

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

31
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 9: Determinants of small gatherings: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

NWsmallmeet NWsmallmeet NWsmallmeet NWsmallmeet NWsmallmeet NWsmallmeet NWsmallmeet NWsmallmeet NWsmallmeet
NWsmallmeet
worse -0.173 -0.156 -0.112 -0.112 -0.179 -0.161 -0.117 -0.115 -0.228
(0.291) (0.353) (0.507) (0.521) (0.274) (0.339) (0.491) (0.511) (0.286)
better -0.153 -0.157 -0.141 -0.136 -0.153 -0.157 -0.139 -0.135 -0.273
(0.364) (0.382) (0.430) (0.462) (0.382) (0.383) (0.436) (0.464) (0.238)
super -0.619∗∗∗ -0.623∗∗∗ -0.603∗∗∗ -0.637∗∗∗ -0.623∗∗∗ -0.627∗∗∗ -0.606∗∗∗ -0.640∗∗∗ -0.714∗∗∗
(0.000) (0.001) (0.001) (0.001) (0.000) (0.001) (0.001) (0.001) (0.001)
worry -0.0858∗∗∗ -0.0834∗∗∗ -0.0826∗∗∗ -0.0829∗∗∗ -0.0841∗∗∗ -0.0813∗∗∗ -0.0807∗∗∗ -0.0809∗∗∗ -0.0462
(0.000) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.002) (0.152)
essential 0.306∗∗ 0.324∗∗ 0.321∗∗ 0.325∗∗ 0.302∗∗ 0.322∗∗ 0.318∗∗ 0.325∗∗ 0.426∗∗
(0.034) (0.036) (0.036) (0.039) (0.047) (0.039) (0.038) (0.040) (0.029)
econsituation 0.000901 -0.0244 -0.0186 -0.0423 -0.000196 -0.0249 -0.0197 -0.0431 0.00348
(0.991) (0.769) (0.821) (0.613) (0.998) (0.763) (0.810) (0.607) (0.973)
male 0.122 0.144 0.105 0.101 0.127 0.150 0.110 0.107 0.0170
(0.317) (0.257) (0.413) (0.437) (0.314) (0.241) (0.390) (0.413) (0.915)
edu 0.0776∗ 0.0807 0.0678 0.0588 0.0763∗ 0.0789 0.0660 0.0570 0.0161
(0.092) (0.101) (0.151) (0.248) (0.100) (0.109) (0.162) (0.263) (0.798)
age -0.0297∗∗∗ -0.0311∗∗∗ -0.0310∗∗∗ -0.0335∗∗∗ -0.0300∗∗∗ -0.0312∗∗∗ -0.0312∗∗∗ -0.0335∗∗∗ -0.0281∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
income 0.0906∗∗∗ 0.0894∗∗∗ 0.0905∗∗∗ 0.0919∗∗∗ 0.0920∗∗∗ 0.0908∗∗∗ 0.0919∗∗∗ 0.0931∗∗∗ 0.0950∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
stateresponse 0.0815 0.0952∗ 0.0978∗ 0.104∗ 0.0811 0.0952∗ 0.0975∗ 0.104∗ 0.107
(0.101) (0.074) (0.061) (0.062) (0.110) (0.074) (0.062) (0.060) (0.129)
USresponse 0.211∗∗∗ 0.145∗∗ 0.0930 0.0508 0.213∗∗∗ 0.145∗∗ 0.0944 0.0494 0.107
(0.000) (0.014) (0.178) (0.486) (0.000) (0.015) (0.175) (0.502) (0.199)
likelydownload 0.106 0.122∗ 0.0892 0.0950 0.101 0.117∗ 0.0837 0.0915 0.153∗
(0.113) (0.075) (0.194) (0.172) (0.140) (0.087) (0.225) (0.190) (0.084)
importance -0.113∗ -0.112 -0.0940 -0.0896 -0.113 -0.114 -0.0950 -0.0925 -0.0256
(0.094) (0.108) (0.177) (0.210) (0.103) (0.106) (0.176) (0.201) (0.792)
Usdownloads 0.00355 0.00282 0.00269 0.00230 0.00360 0.00290 0.00276 0.00238 -0.000249
(0.292) (0.408) (0.419) (0.504) (0.274) (0.395) (0.406) (0.490) (0.952)
republican 0.484∗∗ 0.403∗∗ 0.489∗∗ 0.408∗∗ 0.619∗∗
(0.015) (0.048) (0.014) (0.046) (0.015)
democrat -0.0313 -0.0592 -0.0292 -0.0577 0.0323
(0.829) (0.689) (0.840) (0.696) (0.854)
householdsize -0.0261 -0.0315 -0.0272 -0.0320 -0.0624
(0.607) (0.538) (0.592) (0.531) (0.319)
trustUSgov 0.0296 0.0231 0.0296 0.0235 0.00157
(0.428) (0.555) (0.426) (0.547) (0.974)
trustinsurance 0.0725∗∗ 0.0538 0.0720∗∗ 0.0547 0.0390
(0.029) (0.157) (0.031) (0.150) (0.419)
trustCDC -0.0616∗∗ -0.0733∗∗ -0.0611∗∗ -0.0725∗∗ 0.00123
(0.033) (0.021) (0.035) (0.023) (0.975)
trustmedia 0.0150 0.0149 -0.0145
(0.659) (0.662) (0.721)
trusthealthcare 0.0272 0.0255 0.0113
(0.385) (0.417) (0.779)
prosonly 0.116 0.127 0.125 0.124 0.253
(0.434) (0.395) (0.400) (0.411) (0.161)
prosandcons 0.167 0.134 0.154 0.102 0.157
(0.258) (0.369) (0.302) (0.500) (0.406)
applength -0.276∗∗∗
(0.000)
lesslikelytogetcovid -0.0315
(0.373)
Appmeanssafe 0.0708∗
(0.076)
applessinfected -0.00509
(0.887)
Applessmasks 0.0921∗∗
(0.016)
/

Observations 1018 1003 1008 984 1018 1003 1008 984 701

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

32
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 10: Determinants of seeing people at risk: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

NWseeatrisk NWseeatrisk NWseeatrisk NWseeatrisk NWseeatrisk NWseeatrisk NWseeatrisk NWseeatrisk NWseeatrisk
NWseeatrisk
worse -0.128 -0.168 -0.0751 -0.142 -0.129 -0.167 -0.0751 -0.140 0.0400
(0.470) (0.354) (0.677) (0.442) (0.469) (0.357) (0.677) (0.448) (0.869)
better 0.175 0.127 0.188 0.131 0.175 0.127 0.188 0.132 0.354
(0.329) (0.489) (0.298) (0.477) (0.329) (0.489) (0.298) (0.476) (0.139)
super -0.516∗∗∗ -0.555∗∗∗ -0.491∗∗ -0.561∗∗∗ -0.516∗∗∗ -0.555∗∗∗ -0.492∗∗ -0.561∗∗∗ -0.350
(0.007) (0.004) (0.010) (0.004) (0.007) (0.004) (0.011) (0.004) (0.164)
worry -0.0616∗∗ -0.0583∗∗ -0.0548∗∗ -0.0623∗∗ -0.0613∗∗ -0.0579∗∗ -0.0544∗∗ -0.0616∗∗ -0.0421
(0.017) (0.028) (0.032) (0.019) (0.020) (0.033) (0.037) (0.023) (0.249)
essential 0.497∗∗∗ 0.530∗∗∗ 0.504∗∗∗ 0.538∗∗∗ 0.497∗∗∗ 0.532∗∗∗ 0.504∗∗∗ 0.542∗∗∗ 0.603∗∗∗
(0.001) (0.000) (0.001) (0.000) (0.001) (0.000) (0.001) (0.000) (0.002)
econsituation 0.184∗∗ 0.177∗ 0.174∗ 0.157∗ 0.184∗∗ 0.177∗ 0.174∗ 0.156 0.256∗∗
(0.048) (0.057) (0.063) (0.097) (0.048) (0.058) (0.063) (0.100) (0.034)
male 0.424∗∗∗ 0.432∗∗∗ 0.416∗∗∗ 0.397∗∗∗ 0.424∗∗∗ 0.433∗∗∗ 0.417∗∗∗ 0.398∗∗∗ 0.292
(0.002) (0.002) (0.002) (0.005) (0.002) (0.002) (0.002) (0.005) (0.103)
edu 0.0788 0.100∗ 0.0713 0.0728 0.0787 0.0999∗ 0.0709 0.0719 0.0946
(0.108) (0.055) (0.150) (0.170) (0.109) (0.057) (0.152) (0.176) (0.149)
age -0.0197∗∗∗ -0.0177∗∗∗ -0.0205∗∗∗ -0.0196∗∗∗ -0.0197∗∗∗ -0.0176∗∗∗ -0.0205∗∗∗ -0.0194∗∗∗ -0.0147∗∗
(0.000) (0.001) (0.000) (0.000) (0.000) (0.001) (0.000) (0.000) (0.027)
income 0.0332∗ 0.0165 0.0327 0.0168 0.0334∗ 0.0164 0.0329 0.0168 0.00675
(0.100) (0.448) (0.107) (0.442) (0.099) (0.452) (0.107) (0.445) (0.807)
stateresponse 0.0432 0.0553 0.0701 0.0645 0.0433 0.0554 0.0701 0.0647 0.0527
(0.444) (0.341) (0.228) (0.280) (0.443) (0.341) (0.228) (0.278) (0.499)
USresponse 0.230∗∗∗ 0.147∗∗ 0.106 0.0650 0.230∗∗∗ 0.146∗∗ 0.105 0.0615 0.0548
(0.000) (0.014) (0.160) (0.405) (0.000) (0.014) (0.163) (0.433) (0.573)
likelydownload 0.120∗ 0.129∗ 0.108 0.108 0.120∗ 0.129∗ 0.108 0.108 0.223∗∗
(0.092) (0.074) (0.133) (0.138) (0.095) (0.075) (0.137) (0.140) (0.022)
importance -0.181∗∗ -0.178∗∗ -0.161∗∗ -0.187∗∗ -0.181∗∗ -0.179∗∗ -0.161∗∗ -0.189∗∗ -0.101
(0.013) (0.015) (0.029) (0.015) (0.013) (0.015) (0.029) (0.014) (0.342)
Usdownloads 0.00219 0.00122 0.00169 0.0000195 0.00220 0.00122 0.00170 -0.0000160 -0.000650
(0.536) (0.736) (0.644) (0.996) (0.535) (0.737) (0.643) (0.997) (0.897)
republican 0.438∗∗ 0.393∗ 0.439∗∗ 0.394∗ 0.533∗
(0.026) (0.051) (0.025) (0.051) (0.050)
democrat -0.0499 -0.130 -0.0503 -0.130 0.0161
(0.740) (0.406) (0.738) (0.407) (0.930)
householdsize 0.0819 0.0785 0.0830 0.0802 0.140∗∗
(0.113) (0.133) (0.111) (0.127) (0.029)
trustUSgov 0.0571 0.0346 0.0572 0.0358 0.0138
(0.150) (0.405) (0.152) (0.393) (0.794)
trustinsurance 0.0347 0.00654 0.0348 0.00682 -0.00149
(0.356) (0.876) (0.354) (0.870) (0.978)
trustCDC -0.0615∗∗ -0.0781∗∗ -0.0617∗∗ -0.0794∗∗ -0.0225
(0.049) (0.029) (0.049) (0.028) (0.657)
trustmedia 0.0875∗∗ 0.0875∗∗ 0.0318
(0.013) (0.013) (0.479)
trusthealthcare 0.0159 0.0167 0.0450
(0.660) (0.646) (0.326)
prosonly 0.0122 0.0165 0.0180 0.0369 0.0247
(0.940) (0.920) (0.912) (0.823) (0.901)
prosandcons 0.0133 -0.0301 0.00547 -0.0462 -0.130
(0.931) (0.846) (0.972) (0.769) (0.515)
applength -0.0748
(0.270)
lesslikelytogetcovid -0.0931∗∗
(0.020)
Appmeanssafe 0.105∗∗∗
(0.005)
applessinfected -0.0350
(0.415)
Applessmasks 0.136∗∗∗
(0.001)
/

Observations 1017 1002 1006 982 1017 1002 1006 982 701

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

33
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 11: Determinants of staying home: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

NWstayhome NWstayhome NWstayhome NWstayhome NWstayhome NWstayhome NWstayhome NWstayhome NWstayhome
NWstayhome
worse 0.219 0.235 0.175 0.233 0.214 0.230 0.168 0.228 0.269
(0.199) (0.172) (0.310) (0.182) (0.211) (0.183) (0.331) (0.195) (0.203)
better 0.234 0.248 0.242 0.295∗ 0.235 0.250 0.245 0.300∗ 0.360∗
(0.171) (0.146) (0.158) (0.088) (0.170) (0.146) (0.155) (0.085) (0.087)
super 0.442∗∗∗ 0.460∗∗∗ 0.475∗∗∗ 0.522∗∗∗ 0.442∗∗∗ 0.459∗∗∗ 0.475∗∗∗ 0.523∗∗∗ 0.494∗∗
(0.008) (0.006) (0.005) (0.002) (0.008) (0.007) (0.005) (0.002) (0.017)
worry 0.129∗∗∗ 0.124∗∗∗ 0.122∗∗∗ 0.118∗∗∗ 0.127∗∗∗ 0.122∗∗∗ 0.121∗∗∗ 0.116∗∗∗ 0.0785∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.020)
essential -0.634∗∗∗ -0.645∗∗∗ -0.626∗∗∗ -0.635∗∗∗ -0.640∗∗∗ -0.652∗∗∗ -0.633∗∗∗ -0.645∗∗∗ -0.515∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.005)
econsituation -0.0707 -0.0536 -0.0645 -0.0312 -0.0699 -0.0519 -0.0629 -0.0277 -0.0611
(0.425) (0.551) (0.470) (0.731) (0.432) (0.566) (0.483) (0.762) (0.572)
male -0.0550 -0.0249 -0.0825 -0.0310 -0.0573 -0.0290 -0.0852 -0.0363 -0.199
(0.653) (0.841) (0.509) (0.809) (0.640) (0.816) (0.497) (0.778) (0.205)
edu 0.00540 0.00246 -0.00123 -0.00225 0.00893 0.00617 0.00315 0.00309 0.00209
(0.909) (0.960) (0.980) (0.965) (0.852) (0.900) (0.948) (0.952) (0.973)
age 0.0160∗∗∗ 0.0172∗∗∗ 0.0175∗∗∗ 0.0188∗∗∗ 0.0155∗∗∗ 0.0165∗∗∗ 0.0170∗∗∗ 0.0181∗∗∗ 0.0167∗∗
(0.001) (0.001) (0.000) (0.000) (0.002) (0.001) (0.001) (0.000) (0.011)
income -0.0549∗∗∗ -0.0565∗∗∗ -0.0573∗∗∗ -0.0638∗∗∗ -0.0559∗∗∗ -0.0571∗∗∗ -0.0585∗∗∗ -0.0645∗∗∗ -0.0572∗∗
(0.004) (0.006) (0.003) (0.003) (0.003) (0.006) (0.002) (0.002) (0.028)
stateresponse -0.0694 -0.0794 -0.0958∗ -0.100∗ -0.0705 -0.0812 -0.0975∗ -0.103∗ -0.0613
(0.157) (0.114) (0.061) (0.057) (0.153) (0.108) (0.058) (0.053) (0.351)
USresponse -0.159∗∗∗ -0.149∗∗ -0.120 -0.134∗ -0.157∗∗∗ -0.145∗∗ -0.114 -0.126 -0.0501
(0.003) (0.013) (0.108) (0.088) (0.003) (0.015) (0.128) (0.109) (0.586)
likelydownload -0.0361 -0.0433 -0.0492 -0.0509 -0.0362 -0.0437 -0.0493 -0.0513 -0.0482
(0.594) (0.527) (0.479) (0.472) (0.595) (0.525) (0.480) (0.471) (0.583)
importance 0.166∗∗ 0.164∗∗ 0.116 0.112 0.170∗∗ 0.169∗∗ 0.120 0.117 0.130
(0.020) (0.025) (0.111) (0.137) (0.018) (0.021) (0.102) (0.120) (0.199)
Usdownloads 0.00578 0.00631∗ 0.00613 0.00709∗ 0.00573 0.00628∗ 0.00616 0.00717∗ 0.00658
(0.113) (0.087) (0.104) (0.067) (0.116) (0.088) (0.103) (0.063) (0.160)
republican -0.00236 0.0514 -0.00801 0.0461 -0.0966
(0.991) (0.797) (0.968) (0.817) (0.707)
democrat 0.131 0.0979 0.128 0.0939 0.118
(0.347) (0.503) (0.358) (0.522) (0.510)
householdsize 0.0328 0.0359 0.0298 0.0323 0.0252
(0.531) (0.491) (0.569) (0.535) (0.684)
trustUSgov -0.00213 -0.000495 -0.00291 -0.00141 0.0178
(0.957) (0.990) (0.942) (0.973) (0.721)
trustinsurance -0.00976 -0.00936 -0.0126 -0.0115 0.0289
(0.800) (0.829) (0.744) (0.792) (0.589)
trustCDC 0.0915∗∗∗ 0.0941∗∗∗ 0.0943∗∗∗ 0.0977∗∗∗ 0.0494
(0.003) (0.004) (0.002) (0.003) (0.272)
trustmedia -0.0161 -0.0169 -0.0440
(0.627) (0.612) (0.286)
trusthealthcare 0.0154 0.0140 0.0112
(0.631) (0.663) (0.776)
prosonly -0.103 -0.119 -0.125 -0.146 -0.218
(0.480) (0.420) (0.396) (0.328) (0.233)
prosandcons 0.0492 0.0566 0.0617 0.0799 0.0106
(0.741) (0.707) (0.678) (0.594) (0.953)
applength 0.0242
(0.705)
lesslikelytogetcovid 0.0336
(0.365)
Appmeanssafe -0.0161
(0.680)
applessinfected 0.0253
(0.531)
Applessmasks -0.130∗∗∗
(0.002)
/

Observations 1016 1001 1005 981 1016 1001 1005 981 701

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

34
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 12: Determinants of wearing a mask: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

NWmask NWmask NWmask NWmask NWmask NWmask NWmask NWmask NWmask
NWmask
worse 0.293 0.256 0.185 0.197 0.284 0.246 0.172 0.181 0.491∗
(0.162) (0.224) (0.390) (0.362) (0.177) (0.244) (0.424) (0.401) (0.064)
better 0.341 0.363∗ 0.321 0.390∗ 0.340 0.364∗ 0.324 0.398∗ 0.509∗
(0.100) (0.084) (0.127) (0.070) (0.102) (0.084) (0.124) (0.065) (0.078)
super 0.321 0.362∗ 0.333 0.408∗ 0.313 0.355 0.328 0.404∗ 0.702∗∗
(0.128) (0.091) (0.117) (0.058) (0.141) (0.100) (0.126) (0.062) (0.012)
worry 0.219∗∗∗ 0.214∗∗∗ 0.209∗∗∗ 0.208∗∗∗ 0.225∗∗∗ 0.221∗∗∗ 0.215∗∗∗ 0.215∗∗∗ 0.156∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
essential -0.338∗∗ -0.319∗ -0.324∗ -0.298∗ -0.342∗∗ -0.323∗ -0.327∗ -0.302∗ -0.181
(0.045) (0.062) (0.054) (0.078) (0.043) (0.058) (0.052) (0.075) (0.440)
econsituation -0.00222 -0.0141 0.0193 0.0176 -0.0133 -0.0272 0.00887 0.00501 -0.121
(0.983) (0.893) (0.856) (0.871) (0.899) (0.799) (0.934) (0.963) (0.365)
male -0.500∗∗∗ -0.484∗∗∗ -0.529∗∗∗ -0.464∗∗∗ -0.500∗∗∗ -0.480∗∗∗ -0.528∗∗∗ -0.459∗∗∗ -0.397∗
(0.002) (0.003) (0.001) (0.006) (0.002) (0.003) (0.001) (0.007) (0.079)
edu -0.0733 -0.0889 -0.0772 -0.0945 -0.0761 -0.0937 -0.0779 -0.0969 -0.153∗
(0.233) (0.164) (0.214) (0.147) (0.218) (0.145) (0.213) (0.140) (0.081)
age 0.00474 0.00419 0.00789 0.00642 0.00428 0.00370 0.00716 0.00566 0.00149
(0.438) (0.509) (0.216) (0.333) (0.488) (0.564) (0.265) (0.397) (0.877)
income 0.00414 0.00647 0.00743 0.00182 0.00916 0.0120 0.0122 0.00740 -0.00245
(0.857) (0.787) (0.751) (0.942) (0.697) (0.623) (0.612) (0.770) (0.938)
stateresponse -0.0154 -0.0174 -0.0632 -0.0600 -0.00639 -0.00659 -0.0569 -0.0510 -0.0453
(0.807) (0.785) (0.329) (0.363) (0.920) (0.918) (0.381) (0.443) (0.599)
USresponse -0.240∗∗∗ -0.193∗∗ -0.0666 -0.0587 -0.244∗∗∗ -0.198∗∗∗ -0.0648 -0.0598 0.0440
(0.000) (0.011) (0.455) (0.523) (0.000) (0.010) (0.476) (0.525) (0.725)
likelydownload -0.00478 -0.00974 -0.0153 -0.0161 -0.0140 -0.0176 -0.0262 -0.0258 -0.0826
(0.955) (0.910) (0.863) (0.861) (0.871) (0.840) (0.773) (0.783) (0.489)
importance 0.280∗∗∗ 0.271∗∗∗ 0.212∗∗ 0.214∗∗ 0.273∗∗∗ 0.263∗∗∗ 0.207∗∗ 0.208∗∗ 0.230∗
(0.001) (0.001) (0.014) (0.016) (0.001) (0.002) (0.016) (0.019) (0.057)
Usdownloads -0.00200 -0.00241 0.000192 -0.000236 -0.00185 -0.00235 0.000402 -0.000114 0.00260
(0.653) (0.598) (0.967) (0.961) (0.678) (0.609) (0.931) (0.981) (0.681)
republican 0.113 0.255 0.137 0.285 0.125
(0.635) (0.300) (0.567) (0.250) (0.708)
democrat 0.413∗∗ 0.373∗∗ 0.435∗∗ 0.394∗∗ 0.322
(0.017) (0.041) (0.012) (0.032) (0.153)
householdsize -0.00999 -0.00730 -0.0128 -0.0116 0.0158
(0.875) (0.907) (0.840) (0.854) (0.842)
trustUSgov -0.0560 -0.0620 -0.0581 -0.0650 -0.0946
(0.258) (0.228) (0.243) (0.210) (0.194)
trustinsurance -0.0692 -0.100∗∗ -0.0703 -0.0983∗∗ 0.00430
(0.131) (0.037) (0.128) (0.042) (0.948)
trustCDC 0.165∗∗∗ 0.141∗∗∗ 0.167∗∗∗ 0.146∗∗∗ 0.0899∗
(0.000) (0.000) (0.000) (0.000) (0.092)
trustmedia -0.0399 -0.0400 -0.0402
(0.348) (0.354) (0.499)
trusthealthcare 0.0989∗∗ 0.0949∗∗ 0.114∗∗
(0.019) (0.025) (0.045)
prosonly 0.333∗ 0.362∗ 0.281 0.320∗ -0.126
(0.075) (0.057) (0.138) (0.098) (0.609)
prosandcons 0.325∗ 0.341∗ 0.385∗∗ 0.388∗∗ 0.377
(0.077) (0.066) (0.039) (0.039) (0.126)
applength 0.139∗
(0.090)
lesslikelytogetcovid 0.0385
(0.427)
Appmeanssafe 0.0573
(0.291)
applessinfected 0.0381

35 (0.464)
Applessmasks -0.313∗∗∗
(0.000)
/
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 13: Determinants of length of keeping the app: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

applength applength applength applength applength applength applength applength applength
applength
prosonly -0.0640 -0.0320 -0.0656 -0.0236 -0.0640 -0.0320 -0.0656 -0.0236 -0.0422
(0.710) (0.853) (0.707) (0.894) (0.710) (0.853) (0.707) (0.894) (0.816)
prosandcons -0.180 -0.173 -0.182 -0.148 -0.180 -0.173 -0.182 -0.148 -0.145
(0.315) (0.336) (0.316) (0.424) (0.315) (0.336) (0.316) (0.424) (0.441)
worse 0.210 0.235 0.208 0.247 0.210 0.235 0.208 0.247 0.150
(0.314) (0.263) (0.341) (0.263) (0.314) (0.263) (0.341) (0.263) (0.496)
better 0.371∗ 0.401∗∗ 0.359∗ 0.400∗ 0.371∗ 0.401∗∗ 0.359∗ 0.400∗ 0.262
(0.059) (0.044) (0.072) (0.052) (0.059) (0.044) (0.072) (0.052) (0.207)
super 0.292 0.321∗ 0.263 0.295 0.292 0.321∗ 0.263 0.295 0.140
(0.131) (0.099) (0.177) (0.136) (0.131) (0.099) (0.177) (0.136) (0.503)
worry 0.0416 0.0419 0.0332 0.0321 0.0416 0.0419 0.0332 0.0321 0.0280
(0.188) (0.178) (0.299) (0.315) (0.188) (0.178) (0.299) (0.315) (0.395)
essential -0.0288 -0.0270 -0.00934 -0.00874 -0.0288 -0.0270 -0.00934 -0.00874 0.0752
(0.862) (0.871) (0.956) (0.960) (0.862) (0.871) (0.956) (0.960) (0.678)
econsituation 0.137 0.164 0.120 0.143 0.137 0.164 0.120 0.143 0.163
(0.181) (0.112) (0.250) (0.180) (0.181) (0.112) (0.250) (0.180) (0.126)
male -0.143 -0.104 -0.134 -0.0642 -0.143 -0.104 -0.134 -0.0642 -0.0641
(0.321) (0.469) (0.359) (0.662) (0.321) (0.469) (0.359) (0.662) (0.677)
edu 0.0220 0.0440 0.0201 0.0535 0.0220 0.0440 0.0201 0.0535 0.0827
(0.678) (0.428) (0.714) (0.357) (0.678) (0.428) (0.714) (0.357) (0.165)
age -0.00333 -0.000486 -0.00534 -0.000303 -0.00333 -0.000486 -0.00534 -0.000303 -0.00219
(0.548) (0.932) (0.339) (0.958) (0.548) (0.932) (0.339) (0.958) (0.726)
income -0.0200 -0.0347 -0.0167 -0.0328 -0.0200 -0.0347 -0.0167 -0.0328 -0.0270
(0.365) (0.129) (0.458) (0.168) (0.365) (0.129) (0.458) (0.168) (0.260)
stateresponse -0.0342 -0.0415 -0.0291 -0.0324 -0.0342 -0.0415 -0.0291 -0.0324 -0.0539
(0.516) (0.435) (0.588) (0.555) (0.516) (0.435) (0.588) (0.555) (0.335)
USresponse -0.0467 -0.131 0.0156 -0.0793 -0.0467 -0.131 0.0156 -0.0793 -0.0614
(0.534) (0.126) (0.870) (0.441) (0.534) (0.126) (0.870) (0.441) (0.543)
likelydownload 0.296∗∗∗ 0.316∗∗∗ 0.297∗∗∗ 0.317∗∗∗ 0.296∗∗∗ 0.316∗∗∗ 0.297∗∗∗ 0.317∗∗∗ 0.318∗∗∗
(0.001) (0.000) (0.001) (0.000) (0.001) (0.000) (0.001) (0.000) (0.001)
importance 0.293∗∗∗ 0.276∗∗∗ 0.301∗∗∗ 0.296∗∗∗ 0.293∗∗∗ 0.276∗∗∗ 0.301∗∗∗ 0.296∗∗∗ 0.238∗∗
(0.001) (0.003) (0.001) (0.002) (0.001) (0.003) (0.001) (0.002) (0.013)
Usdownloads 0.00666 0.00707 0.00709 0.00830∗ 0.00666 0.00707 0.00709 0.00830∗ 0.0102∗∗
(0.127) (0.105) (0.118) (0.075) (0.127) (0.105) (0.118) (0.075) (0.037)
republican 0.528∗ 0.599∗∗ 0.528∗ 0.599∗∗ 0.591∗∗
(0.061) (0.038) (0.061) (0.038) (0.046)
democrat 0.0928 0.146 0.0928 0.146 0.0856
(0.544) (0.377) (0.544) (0.377) (0.613)
householdsize 0.0670 0.0675 0.0670 0.0675 0.0971∗
(0.221) (0.226) (0.221) (0.226) (0.099)
trustUSgov -0.0653 -0.0642 -0.0653 -0.0642 -0.0451
(0.158) (0.167) (0.158) (0.167) (0.337)
trustinsurance 0.0201 0.0302 0.0201 0.0302 0.0346
(0.600) (0.479) (0.600) (0.479) (0.441)
trustCDC 0.00410 0.0273 0.00410 0.0273 -0.00547
(0.911) (0.488) (0.911) (0.488) (0.894)
trustmedia -0.0412 -0.0412 -0.0517
(0.263) (0.263) (0.175)
trusthealthcare -0.0120 -0.0120 -0.00819
(0.729) (0.729) (0.823)
lesslikelytogetcovid 0.0368
(0.261)
Appmeanssafe -0.0770∗
(0.066)
applessinfected 0.164∗∗∗
(0.000)
Applessmasks -0.111∗∗
(0.025)
/
(0.435) (0.891) heightObservations 736 728 729 716 736 728
729 716 701

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

36
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 14: Determinants of saying you’re less likely to get covid: combined controls, robust standard
errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

lesslikelytogetcovid lesslikelytogetcovid lesslikelytogetcovid lesslikelytogetcovid lesslikelytogetcovid lesslikelytogetcovid lesslikelytogetcovid lesslikelytogetcovid lesslikelytogetcovid
lesslikelytogetcovid
prosonly -0.199 -0.189 -0.169 -0.121 -0.199 -0.189 -0.169 -0.121 0.00832
(0.152) (0.179) (0.226) (0.392) (0.152) (0.179) (0.226) (0.392) (0.961)
prosandcons -0.317∗∗ -0.345∗∗ -0.294∗∗ -0.295∗∗ -0.317∗∗ -0.345∗∗ -0.294∗∗ -0.295∗∗ -0.210
(0.028) (0.018) (0.041) (0.044) (0.028) (0.018) (0.041) (0.044) (0.211)
worse 0.276∗ 0.280∗ 0.235 0.243 0.276∗ 0.280∗ 0.235 0.243 0.263
(0.076) (0.078) (0.145) (0.138) (0.076) (0.078) (0.145) (0.138) (0.165)
better 0.328∗∗ 0.325∗ 0.308∗ 0.314∗ 0.328∗∗ 0.325∗ 0.308∗ 0.314∗ 0.289
(0.048) (0.053) (0.068) (0.066) (0.048) (0.053) (0.068) (0.066) (0.131)
super 0.263 0.275∗ 0.285∗ 0.278 0.263 0.275∗ 0.285∗ 0.278 0.293
(0.111) (0.095) (0.090) (0.101) (0.111) (0.095) (0.090) (0.101) (0.113)
worry 0.0870∗∗∗ 0.0776∗∗∗ 0.0754∗∗∗ 0.0622∗∗ 0.0870∗∗∗ 0.0776∗∗∗ 0.0754∗∗∗ 0.0622∗∗ 0.0210
(0.000) (0.002) (0.002) (0.015) (0.000) (0.002) (0.002) (0.015) (0.500)
essential -0.262∗ -0.249∗ -0.196 -0.189 -0.262∗ -0.249∗ -0.196 -0.189 -0.377∗∗
(0.070) (0.085) (0.178) (0.197) (0.070) (0.085) (0.178) (0.197) (0.021)
econsituation 0.0752 0.0849 0.0539 0.0427 0.0752 0.0849 0.0539 0.0427 0.0445
(0.330) (0.285) (0.482) (0.592) (0.330) (0.285) (0.482) (0.592) (0.611)
male 0.510∗∗∗ 0.541∗∗∗ 0.471∗∗∗ 0.468∗∗∗ 0.510∗∗∗ 0.541∗∗∗ 0.471∗∗∗ 0.468∗∗∗ 0.280∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.045)
edu 0.0489 0.0519 0.0201 0.00968 0.0489 0.0519 0.0201 0.00968 -0.0855
(0.286) (0.284) (0.658) (0.845) (0.286) (0.284) (0.658) (0.845) (0.125)
age -0.0290∗∗∗ -0.0270∗∗∗ -0.0293∗∗∗ -0.0276∗∗∗ -0.0290∗∗∗ -0.0270∗∗∗ -0.0293∗∗∗ -0.0276∗∗∗ -0.00888
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.215)
income -0.0193 -0.0245 -0.0235 -0.0253 -0.0193 -0.0245 -0.0235 -0.0253 -0.0451∗
(0.313) (0.238) (0.224) (0.224) (0.313) (0.238) (0.224) (0.224) (0.075)
stateresponse 0.0543 0.0429 0.0295 0.0213 0.0543 0.0429 0.0295 0.0213 -0.0637
(0.281) (0.398) (0.556) (0.675) (0.281) (0.398) (0.556) (0.675) (0.303)
USresponse 0.310∗∗∗ 0.382∗∗∗ 0.256∗∗∗ 0.322∗∗∗ 0.310∗∗∗ 0.382∗∗∗ 0.256∗∗∗ 0.322∗∗∗ 0.402∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
likelydownload 0.312∗∗∗ 0.295∗∗∗ 0.266∗∗∗ 0.255∗∗∗ 0.312∗∗∗ 0.295∗∗∗ 0.266∗∗∗ 0.255∗∗∗ 0.141∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.063)
importance 0.375∗∗∗ 0.339∗∗∗ 0.333∗∗∗ 0.300∗∗∗ 0.375∗∗∗ 0.339∗∗∗ 0.333∗∗∗ 0.300∗∗∗ 0.0173
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.828)
Usdownloads 0.00440 0.00497 0.00206 0.00182 0.00440 0.00497 0.00206 0.00182 0.000522
(0.235) (0.186) (0.582) (0.644) (0.235) (0.186) (0.582) (0.644) (0.902)
republican -0.303 -0.259 -0.303 -0.259 -0.165
(0.111) (0.180) (0.111) (0.180) (0.457)
democrat 0.358∗∗∗ 0.265∗ 0.358∗∗∗ 0.265∗ 0.364∗∗
(0.008) (0.057) (0.008) (0.057) (0.023)
householdsize 0.0610 0.0592 0.0610 0.0592 0.0618
(0.212) (0.227) (0.212) (0.227) (0.272)
trustUSgov 0.00864 0.0189 0.00864 0.0189 0.0113
(0.823) (0.639) (0.823) (0.639) (0.796)
trustinsurance 0.0853∗∗ 0.0754∗ 0.0853∗∗ 0.0754∗ -0.0358
(0.018) (0.061) (0.018) (0.061) (0.439)
trustCDC 0.101∗∗∗ 0.0877∗∗∗ 0.101∗∗∗ 0.0877∗∗∗ 0.0623
(0.000) (0.007) (0.000) (0.007) (0.147)
trustmedia 0.0620 0.0620 0.0125
(0.102) (0.102) (0.775)
trusthealthcare -0.0374 -0.0374 -0.0483
(0.281) (0.281) (0.256)
prosonly 0
(.)
prosandcons 0
(.)
prosonly 0
(.)
prosandcons 0
(.)
prosonly 0
(.)
prosandcons 0
(.)
prosonly 0
(.)
prosandcons 0
(.)
prosonly 0
(.)
prosandcons 0
(.)
applength 0.0676
(0.167)
Appmeanssafe 0.261∗∗∗
(0.000)
applessinfected 0.423∗∗∗
(0.000)
Applessmasks -0.00958
(0.804)
/

Observations 1002 989 993 971 1002 989 993 971 701

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

37
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 15: Determinants of saying app makes us safer: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Appmeanssafe Appmeanssafe Appmeanssafe Appmeanssafe Appmeanssafe Appmeanssafe Appmeanssafe Appmeanssafe Appmeanssafe
Appmeanssafe
prosonly -0.0542 -0.0739 -0.00217 -0.00975 -0.0542 -0.0739 -0.00217 -0.00975 0.0824
(0.713) (0.618) (0.988) (0.947) (0.713) (0.618) (0.988) (0.947) (0.632)
prosandcons -0.250∗ -0.296∗ -0.276∗ -0.332∗∗ -0.250∗ -0.296∗ -0.276∗ -0.332∗∗ -0.0939
(0.093) (0.051) (0.067) (0.029) (0.093) (0.051) (0.067) (0.029) (0.614)
worse -0.373∗∗ -0.354∗∗ -0.318∗ -0.304∗ -0.373∗∗ -0.354∗∗ -0.318∗ -0.304∗ -0.622∗∗∗
(0.020) (0.032) (0.053) (0.071) (0.020) (0.032) (0.053) (0.071) (0.002)
better -0.233 -0.241 -0.215 -0.223 -0.233 -0.241 -0.215 -0.223 -0.529∗∗
(0.180) (0.171) (0.217) (0.203) (0.180) (0.171) (0.217) (0.203) (0.013)
super -0.355∗∗ -0.337∗∗ -0.286∗ -0.311∗ -0.355∗∗ -0.337∗∗ -0.286∗ -0.311∗ -0.554∗∗∗
(0.032) (0.044) (0.085) (0.065) (0.032) (0.044) (0.085) (0.065) (0.004)
worry 0.00783 0.00306 0.0108 -0.00194 0.00783 0.00306 0.0108 -0.00194 -0.0524
(0.749) (0.902) (0.666) (0.940) (0.749) (0.902) (0.666) (0.940) (0.114)
essential 0.318∗∗ 0.324∗∗ 0.367∗∗ 0.378∗∗ 0.318∗∗ 0.324∗∗ 0.367∗∗ 0.378∗∗ 0.460∗∗
(0.029) (0.028) (0.013) (0.014) (0.029) (0.028) (0.013) (0.014) (0.010)
econsituation 0.0908 0.0933 0.0478 0.0310 0.0908 0.0933 0.0478 0.0310 -0.0535
(0.255) (0.244) (0.541) (0.693) (0.255) (0.244) (0.541) (0.693) (0.539)
male 0.396∗∗∗ 0.412∗∗∗ 0.285∗∗ 0.229∗ 0.396∗∗∗ 0.412∗∗∗ 0.285∗∗ 0.229∗ -0.0419
(0.001) (0.001) (0.025) (0.079) (0.001) (0.001) (0.025) (0.079) (0.779)
edu 0.107∗∗ 0.134∗∗∗ 0.0875∗ 0.0914∗ 0.107∗∗ 0.134∗∗∗ 0.0875∗ 0.0914∗ 0.0443
(0.026) (0.009) (0.077) (0.085) (0.026) (0.009) (0.077) (0.085) (0.480)
age -0.0369∗∗∗ -0.0353∗∗∗ -0.0392∗∗∗ -0.0398∗∗∗ -0.0369∗∗∗ -0.0353∗∗∗ -0.0392∗∗∗ -0.0398∗∗∗ -0.0265∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
income 0.0494∗∗ 0.0376∗ 0.0394∗∗ 0.0308 0.0494∗∗ 0.0376∗ 0.0394∗∗ 0.0308 0.0254
(0.011) (0.076) (0.048) (0.155) (0.011) (0.076) (0.048) (0.155) (0.315)
stateresponse 0.102∗∗ 0.0965∗ 0.0890∗ 0.0690 0.102∗∗ 0.0965∗ 0.0890∗ 0.0690 0.0470
(0.046) (0.063) (0.084) (0.184) (0.046) (0.063) (0.084) (0.184) (0.425)
USresponse 0.364∗∗∗ 0.374∗∗∗ 0.115 0.165∗∗ 0.364∗∗∗ 0.374∗∗∗ 0.115 0.165∗∗ 0.000132
(0.000) (0.000) (0.116) (0.046) (0.000) (0.000) (0.116) (0.046) (0.999)
likelydownload 0.257∗∗∗ 0.243∗∗∗ 0.206∗∗∗ 0.180∗∗ 0.257∗∗∗ 0.243∗∗∗ 0.206∗∗∗ 0.180∗∗ 0.0522
(0.000) (0.001) (0.004) (0.013) (0.000) (0.001) (0.004) (0.013) (0.572)
importance 0.145∗∗ 0.136∗ 0.126∗ 0.103 0.145∗∗ 0.136∗ 0.126∗ 0.103 -0.115
(0.035) (0.056) (0.077) (0.172) (0.035) (0.056) (0.077) (0.172) (0.236)
Usdownloads 0.0113∗∗∗ 0.0113∗∗∗ 0.00789∗∗ 0.00640∗ 0.0113∗∗∗ 0.0113∗∗∗ 0.00789∗∗ 0.00640∗ 0.00613
(0.001) (0.001) (0.022) (0.075) (0.001) (0.001) (0.022) (0.075) (0.141)
republican -0.168 -0.227 -0.168 -0.227 -0.275
(0.405) (0.267) (0.405) (0.267) (0.283)
democrat 0.0312 -0.107 0.0312 -0.107 -0.207
(0.823) (0.468) (0.823) (0.468) (0.212)
householdsize 0.107∗∗ 0.102∗ 0.107∗∗ 0.102∗ 0.113∗
(0.039) (0.053) (0.039) (0.053) (0.078)
trustUSgov 0.0899∗∗ 0.0852∗∗ 0.0899∗∗ 0.0852∗∗ 0.0752
(0.018) (0.029) (0.018) (0.029) (0.106)
trustinsurance 0.166∗∗∗ 0.123∗∗∗ 0.166∗∗∗ 0.123∗∗∗ 0.0321
(0.000) (0.002) (0.000) (0.002) (0.488)
trustCDC -0.0325 -0.0682∗∗ -0.0325 -0.0682∗∗ -0.0696
(0.259) (0.031) (0.259) (0.031) (0.117)
trustmedia 0.119∗∗∗ 0.119∗∗∗ 0.0647
(0.001) (0.001) (0.115)
trusthealthcare 0.0162 0.0162 0.0600
(0.614) (0.614) (0.126)
prosonly 0
(.)
prosandcons 0
(.)
applength -0.125∗∗
(0.035)
lesslikelytogetcovid 0.269∗∗∗
(0.000)
applessinfected 0.0682∗
(0.062)
Applessmasks 0.404∗∗∗
(0.000)
/

Observations 1004 991 993 971 1004 991 993 971 701

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

38
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 16: Determinants of saying app means less infected: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

applessinfected applessinfected applessinfected applessinfected applessinfected applessinfected applessinfected applessinfected applessinfected
applessinfected
prosonly -0.101 -0.105 -0.0969 -0.0812 -0.101 -0.105 -0.0969 -0.0812 -0.0791
(0.462) (0.445) (0.481) (0.559) (0.462) (0.445) (0.481) (0.559) (0.627)
prosandcons -0.168 -0.161 -0.171 -0.180 -0.168 -0.161 -0.171 -0.180 0.123
(0.225) (0.249) (0.219) (0.204) (0.225) (0.249) (0.219) (0.204) (0.464)
worse 0.180 0.206 0.178 0.182 0.180 0.206 0.178 0.182 0.0938
(0.260) (0.202) (0.271) (0.266) (0.260) (0.202) (0.271) (0.266) (0.633)
better 0.346∗∗ 0.365∗∗ 0.347∗∗ 0.391∗∗ 0.346∗∗ 0.365∗∗ 0.347∗∗ 0.391∗∗ 0.196
(0.034) (0.027) (0.036) (0.020) (0.034) (0.027) (0.036) (0.020) (0.323)
super 0.170 0.198 0.241 0.238 0.170 0.198 0.241 0.238 0.0859
(0.289) (0.218) (0.138) (0.150) (0.289) (0.218) (0.138) (0.150) (0.665)
worry 0.0522∗∗ 0.0462∗ 0.0451∗ 0.0312 0.0522∗∗ 0.0462∗ 0.0451∗ 0.0312 -0.00717
(0.024) (0.052) (0.053) (0.206) (0.024) (0.052) (0.053) (0.206) (0.811)
essential -0.0769 -0.0843 -0.0301 -0.0235 -0.0769 -0.0843 -0.0301 -0.0235 0.117
(0.579) (0.543) (0.830) (0.866) (0.579) (0.543) (0.830) (0.866) (0.483)
econsituation 0.0271 0.0259 0.00425 -0.0287 0.0271 0.0259 0.00425 -0.0287 -0.170∗
(0.712) (0.728) (0.955) (0.708) (0.712) (0.728) (0.955) (0.708) (0.055)
male 0.511∗∗∗ 0.509∗∗∗ 0.450∗∗∗ 0.409∗∗∗ 0.511∗∗∗ 0.509∗∗∗ 0.450∗∗∗ 0.409∗∗∗ 0.131
(0.000) (0.000) (0.000) (0.001) (0.000) (0.000) (0.000) (0.001) (0.356)
edu 0.103∗∗ 0.0961∗∗ 0.0805∗ 0.0370 0.103∗∗ 0.0961∗∗ 0.0805∗ 0.0370 0.00775
(0.019) (0.037) (0.068) (0.438) (0.019) (0.037) (0.068) (0.438) (0.888)
age -0.0143∗∗∗ -0.0144∗∗∗ -0.0133∗∗∗ -0.0149∗∗∗ -0.0143∗∗∗ -0.0144∗∗∗ -0.0133∗∗∗ -0.0149∗∗∗ 0.0115
(0.004) (0.007) (0.009) (0.006) (0.004) (0.007) (0.009) (0.006) (0.100)
income 0.00482 0.0126 0.00255 0.0157 0.00482 0.0126 0.00255 0.0157 0.0212
(0.794) (0.518) (0.891) (0.427) (0.794) (0.518) (0.891) (0.427) (0.368)
stateresponse 0.174∗∗∗ 0.158∗∗∗ 0.151∗∗∗ 0.126∗∗ 0.174∗∗∗ 0.158∗∗∗ 0.151∗∗∗ 0.126∗∗ 0.114∗
(0.000) (0.001) (0.002) (0.013) (0.000) (0.001) (0.002) (0.013) (0.063)
USresponse -0.0256 0.0264 -0.0883 -0.0162 -0.0256 0.0264 -0.0883 -0.0162 -0.0915
(0.649) (0.685) (0.236) (0.851) (0.649) (0.685) (0.236) (0.851) (0.284)
likelydownload 0.241∗∗∗ 0.223∗∗∗ 0.191∗∗∗ 0.160∗∗ 0.241∗∗∗ 0.223∗∗∗ 0.191∗∗∗ 0.160∗∗ -0.0556
(0.001) (0.002) (0.008) (0.034) (0.001) (0.002) (0.008) (0.034) (0.502)
importance 0.545∗∗∗ 0.535∗∗∗ 0.488∗∗∗ 0.492∗∗∗ 0.545∗∗∗ 0.535∗∗∗ 0.488∗∗∗ 0.492∗∗∗ 0.303∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Usdownloads 0.00323 0.00404 0.00226 0.00164 0.00323 0.00404 0.00226 0.00164 0.000229
(0.384) (0.288) (0.537) (0.666) (0.384) (0.288) (0.537) (0.666) (0.955)
republican -0.140 -0.100 -0.140 -0.100 -0.0202
(0.459) (0.606) (0.459) (0.606) (0.924)
democrat 0.213 0.0587 0.213 0.0587 -0.0264
(0.121) (0.672) (0.121) (0.672) (0.863)
householdsize -0.0299 -0.0422 -0.0299 -0.0422 -0.115∗∗
(0.517) (0.357) (0.517) (0.357) (0.028)
trustUSgov 0.0217 0.0191 0.0217 0.0191 -0.0302
(0.575) (0.637) (0.575) (0.637) (0.503)
trustinsurance 0.0703∗∗ 0.0378 0.0703∗∗ 0.0378 0.0182
(0.039) (0.309) (0.039) (0.309) (0.699)
trustCDC 0.101∗∗∗ 0.0739∗∗ 0.101∗∗∗ 0.0739∗∗ 0.0761∗
(0.000) (0.023) (0.000) (0.023) (0.085)
trustmedia 0.109∗∗∗ 0.109∗∗∗ 0.0649
(0.003) (0.003) (0.109)
trusthealthcare -0.00119 -0.00119 0.00517
(0.971) (0.971) (0.898)
applength 0.248∗∗∗
(0.000)
lesslikelytogetcovid 0.382∗∗∗
(0.000)
Appmeanssafe 0.0818∗∗
(0.026)
Applessmasks 0.0346
(0.291)
/

Observations 1001 988 992 970 1001 988 992 970 701

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

39
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 17: Determinants of saying app means masks are less important: combined controls, robust
standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

Applessmasks Applessmasks Applessmasks Applessmasks Applessmasks Applessmasks Applessmasks Applessmasks Applessmasks
Applessmasks
prosonly -0.159 -0.163 -0.134 -0.119 -0.159 -0.163 -0.134 -0.119 0.201
(0.312) (0.304) (0.398) (0.459) (0.312) (0.304) (0.398) (0.459) (0.327)
prosandcons -0.272∗ -0.268 -0.319∗ -0.304∗ -0.272∗ -0.268 -0.319∗ -0.304∗ 0.0373
(0.097) (0.105) (0.057) (0.072) (0.097) (0.105) (0.057) (0.072) (0.867)
worse -0.155 -0.139 -0.0664 -0.0760 -0.155 -0.139 -0.0664 -0.0760 -0.0534
(0.398) (0.451) (0.723) (0.691) (0.398) (0.451) (0.723) (0.691) (0.827)
better -0.0478 -0.0536 -0.0210 -0.0231 -0.0478 -0.0536 -0.0210 -0.0231 -0.0210
(0.795) (0.772) (0.908) (0.903) (0.795) (0.772) (0.908) (0.903) (0.932)
super -0.180 -0.198 -0.111 -0.144 -0.180 -0.198 -0.111 -0.144 -0.0746
(0.311) (0.269) (0.535) (0.428) (0.311) (0.269) (0.535) (0.428) (0.737)
worry -0.0665∗∗ -0.0676∗∗∗ -0.0579∗∗ -0.0647∗∗ -0.0665∗∗ -0.0676∗∗∗ -0.0579∗∗ -0.0647∗∗ -0.0349
(0.010) (0.010) (0.028) (0.016) (0.010) (0.010) (0.028) (0.016) (0.334)
essential 0.0938 0.0881 0.0936 0.0966 0.0938 0.0881 0.0936 0.0966 -0.0200
(0.548) (0.576) (0.554) (0.549) (0.548) (0.576) (0.554) (0.549) (0.918)
econsituation 0.0540 0.0446 0.0277 -0.00305 0.0540 0.0446 0.0277 -0.00305 0.207∗
(0.528) (0.607) (0.744) (0.972) (0.528) (0.607) (0.744) (0.972) (0.052)
male 0.343∗∗ 0.331∗∗ 0.260∗ 0.196 0.343∗∗ 0.331∗∗ 0.260∗ 0.196 0.0152
(0.012) (0.016) (0.062) (0.167) (0.012) (0.016) (0.062) (0.167) (0.931)
edu 0.179∗∗∗ 0.183∗∗∗ 0.173∗∗∗ 0.154∗∗∗ 0.179∗∗∗ 0.183∗∗∗ 0.173∗∗∗ 0.154∗∗∗ 0.150∗∗
(0.000) (0.001) (0.001) (0.004) (0.000) (0.001) (0.001) (0.004) (0.034)
age -0.0227∗∗∗ -0.0239∗∗∗ -0.0248∗∗∗ -0.0279∗∗∗ -0.0227∗∗∗ -0.0239∗∗∗ -0.0248∗∗∗ -0.0279∗∗∗ -0.00874
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.240)
income 0.0345∗ 0.0353 0.0337 0.0446∗∗ 0.0345∗ 0.0353 0.0337 0.0446∗∗ 0.0460
(0.099) (0.111) (0.112) (0.048) (0.099) (0.111) (0.112) (0.048) (0.104)
stateresponse -0.000335 0.00136 0.0233 0.00609 -0.000335 0.00136 0.0233 0.00609 -0.0867
(0.995) (0.980) (0.668) (0.915) (0.995) (0.980) (0.668) (0.915) (0.221)
USresponse 0.486∗∗∗ 0.461∗∗∗ 0.237∗∗∗ 0.287∗∗∗ 0.486∗∗∗ 0.461∗∗∗ 0.237∗∗∗ 0.287∗∗∗ 0.130
(0.000) (0.000) (0.002) (0.000) (0.000) (0.000) (0.002) (0.000) (0.211)
likelydownload 0.176∗∗ 0.187∗∗ 0.136∗ 0.128∗ 0.176∗∗ 0.187∗∗ 0.136∗ 0.128∗ 0.137
(0.015) (0.010) (0.064) (0.086) (0.015) (0.010) (0.064) (0.086) (0.137)
importance 0.115 0.115 0.129∗ 0.113 0.115 0.115 0.129∗ 0.113 -0.0300
(0.125) (0.128) (0.098) (0.164) (0.125) (0.128) (0.098) (0.164) (0.787)
Usdownloads 0.00622∗ 0.00609∗ 0.00277 0.00112 0.00622∗ 0.00609∗ 0.00277 0.00112 -0.00712
(0.079) (0.089) (0.447) (0.768) (0.079) (0.089) (0.447) (0.768) (0.164)
republican 0.106 0.0192 0.106 0.0192 -0.0180
(0.625) (0.934) (0.625) (0.934) (0.949)
democrat -0.104 -0.183 -0.104 -0.183 -0.190
(0.488) (0.238) (0.488) (0.238) (0.317)
householdsize 0.0151 -0.0195 0.0151 -0.0195 -0.0596
(0.760) (0.701) (0.760) (0.701) (0.370)
trustUSgov 0.0846∗∗ 0.0695∗ 0.0846∗∗ 0.0695∗ 0.0569
(0.036) (0.097) (0.036) (0.097) (0.304)
trustinsurance 0.152∗∗∗ 0.146∗∗∗ 0.152∗∗∗ 0.146∗∗∗ 0.141∗∗
(0.000) (0.000) (0.000) (0.000) (0.010)
trustCDC -0.109∗∗∗ -0.118∗∗∗ -0.109∗∗∗ -0.118∗∗∗ -0.168∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.001)
trustmedia 0.120∗∗∗ 0.120∗∗∗ 0.109∗∗
(0.002) (0.002) (0.024)
trusthealthcare -0.0560 -0.0560 -0.0631
(0.108) (0.108) (0.215)
applength -0.205∗∗∗
(0.004)
lesslikelytogetcovid -0.00223
(0.957)
Appmeanssafe 0.386∗∗∗
(0.000)
applessinfected 0.0673
(0.119)
/

Observations 1005 992 994 972 1005 992 994 972 701

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

40
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 18: Determinants of sneezing in elbow: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

sneeze sneeze sneeze sneeze sneeze sneeze sneeze sneeze sneeze
sneeze
worse 0.106 0.0926 0.0453 0.105 0.112 0.0965 0.0441 0.108 0.133
(0.619) (0.666) (0.835) (0.634) (0.598) (0.652) (0.839) (0.624) (0.639)
better 0.0530 0.0431 0.0183 0.0598 0.0559 0.0437 0.0199 0.0584 -0.0716
(0.805) (0.843) (0.933) (0.787) (0.794) (0.840) (0.927) (0.791) (0.796)
super -0.0206 -0.0151 0.0118 0.0826 -0.0155 -0.0122 0.0154 0.0845 -0.101
(0.921) (0.943) (0.955) (0.701) (0.940) (0.954) (0.942) (0.693) (0.703)
worry 0.0964∗∗∗ 0.0959∗∗∗ 0.0880∗∗∗ 0.0930∗∗∗ 0.0927∗∗∗ 0.0923∗∗∗ 0.0833∗∗∗ 0.0880∗∗∗ 0.0872∗∗
(0.001) (0.001) (0.002) (0.002) (0.001) (0.002) (0.004) (0.004) (0.022)
essential -0.0651 -0.0627 -0.0316 -0.0355 -0.0668 -0.0626 -0.0379 -0.0394 0.101
(0.699) (0.715) (0.853) (0.835) (0.694) (0.717) (0.825) (0.819) (0.645)
econsituation -0.102 -0.119 -0.101 -0.101 -0.0979 -0.115 -0.0958 -0.0959 -0.175
(0.295) (0.233) (0.313) (0.323) (0.316) (0.248) (0.337) (0.345) (0.153)
male -0.543∗∗∗ -0.509∗∗∗ -0.566∗∗∗ -0.482∗∗∗ -0.547∗∗∗ -0.514∗∗∗ -0.573∗∗∗ -0.491∗∗∗ -0.605∗∗∗
(0.000) (0.001) (0.000) (0.002) (0.000) (0.001) (0.000) (0.002) (0.002)
edu -0.105∗ -0.117∗∗ -0.124∗∗ -0.129∗∗ -0.101∗ -0.114∗∗ -0.120∗∗ -0.126∗∗ -0.129
(0.055) (0.041) (0.030) (0.034) (0.066) (0.047) (0.038) (0.039) (0.101)
age 0.0229∗∗∗ 0.0221∗∗∗ 0.0257∗∗∗ 0.0254∗∗∗ 0.0229∗∗∗ 0.0221∗∗∗ 0.0256∗∗∗ 0.0253∗∗∗ 0.0267∗∗∗
(0.000) (0.001) (0.000) (0.000) (0.000) (0.001) (0.000) (0.000) (0.002)
income -0.00982 -0.00862 -0.0107 -0.0198 -0.0133 -0.0115 -0.0141 -0.0232 -0.0496
(0.647) (0.711) (0.625) (0.414) (0.534) (0.619) (0.519) (0.336) (0.106)
stateresponse 0.00820 0.0153 -0.0225 -0.00902 0.00562 0.0124 -0.0258 -0.0143 -0.0305
(0.894) (0.805) (0.724) (0.890) (0.927) (0.841) (0.687) (0.828) (0.708)
USresponse -0.160∗∗∗ -0.164∗∗ -0.0830 -0.147∗ -0.159∗∗∗ -0.162∗∗ -0.0766 -0.140 -0.0638
(0.009) (0.018) (0.307) (0.084) (0.008) (0.019) (0.348) (0.103) (0.553)
likelydownload -0.0463 -0.0392 -0.0620 -0.0601 -0.0407 -0.0352 -0.0565 -0.0554 -0.134
(0.572) (0.634) (0.467) (0.494) (0.616) (0.667) (0.504) (0.525) (0.243)
importance 0.218∗∗∗ 0.206∗∗∗ 0.169∗∗ 0.173∗∗ 0.224∗∗∗ 0.212∗∗∗ 0.174∗∗ 0.180∗∗ 0.201∗
(0.004) (0.008) (0.030) (0.034) (0.004) (0.006) (0.025) (0.028) (0.074)
Usdownloads 0.00413 0.00421 0.00550 0.00646 0.00401 0.00411 0.00552 0.00648 0.00641
(0.331) (0.325) (0.207) (0.147) (0.345) (0.337) (0.206) (0.147) (0.281)
republican 0.118 0.207 0.111 0.194 0.0205
(0.616) (0.396) (0.637) (0.426) (0.947)
democrat 0.124 0.151 0.121 0.147 -0.0388
(0.439) (0.373) (0.451) (0.386) (0.850)
householdsize -0.0410 -0.0323 -0.0410 -0.0326 -0.00131
(0.476) (0.585) (0.479) (0.582) (0.985)
trustUSgov -0.00831 0.00262 -0.00962 0.00124 -0.00656
(0.857) (0.957) (0.836) (0.980) (0.904)
trustinsurance -0.0390 -0.0612 -0.0424 -0.0657 -0.00753
(0.396) (0.232) (0.360) (0.200) (0.898)
trustCDC 0.125∗∗∗ 0.116∗∗∗ 0.130∗∗∗ 0.118∗∗∗ 0.0980∗
(0.000) (0.001) (0.000) (0.001) (0.059)
trustmedia -0.0844∗∗ -0.0855∗∗ -0.0800
(0.040) (0.037) (0.114)
trusthealthcare 0.101∗∗ 0.107∗∗∗ 0.121∗∗
(0.011) (0.008) (0.031)
prosonly -0.269 -0.238 -0.299∗ -0.307∗ -0.456∗∗
(0.123) (0.175) (0.089) (0.082) (0.036)
prosandcons -0.177 -0.161 -0.129 -0.155 -0.251
(0.313) (0.363) (0.471) (0.385) (0.249)
applength 0.135∗
(0.067)
lesslikelytogetcovid -0.0345
(0.429)
Appmeanssafe -0.0348
(0.416)
applessinfected 0.0844∗
41 (0.054)
Applessmasks -0.155∗∗∗
(0.000)
/
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 19: Determinants of social distancing after warning: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

socialdistance socialdistance socialdistance socialdistance socialdistance socialdistance socialdistance socialdistance socialdistance
socialdistance
worse 0.185 0.192 0.126 0.189 0.190 0.199 0.130 0.202 0.251
(0.330) (0.315) (0.519) (0.338) (0.315) (0.297) (0.504) (0.307) (0.307)
better 0.122 0.139 0.102 0.167 0.124 0.142 0.103 0.168 -0.0272
(0.510) (0.456) (0.584) (0.379) (0.503) (0.447) (0.581) (0.376) (0.912)
super 0.130 0.154 0.144 0.249 0.137 0.161 0.151 0.258 0.319
(0.477) (0.402) (0.437) (0.185) (0.456) (0.381) (0.416) (0.169) (0.177)
worry 0.152∗∗∗ 0.155∗∗∗ 0.146∗∗∗ 0.152∗∗∗ 0.150∗∗∗ 0.153∗∗∗ 0.143∗∗∗ 0.149∗∗∗ 0.112∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.002)
essential -0.500∗∗∗ -0.495∗∗∗ -0.499∗∗∗ -0.501∗∗∗ -0.500∗∗∗ -0.494∗∗∗ -0.501∗∗∗ -0.501∗∗∗ -0.276
(0.001) (0.002) (0.002) (0.001) (0.002) (0.002) (0.002) (0.001) (0.167)
econsituation -0.0374 -0.0536 -0.0503 -0.0532 -0.0333 -0.0493 -0.0451 -0.0473 -0.158
(0.689) (0.571) (0.595) (0.576) (0.723) (0.603) (0.635) (0.620) (0.179)
male -0.147 -0.152 -0.181 -0.154 -0.150 -0.156 -0.185 -0.161 -0.150
(0.273) (0.264) (0.181) (0.269) (0.267) (0.254) (0.174) (0.250) (0.390)
edu -0.0672 -0.0850 -0.0836 -0.106∗ -0.0663 -0.0835 -0.0823 -0.105∗ -0.129∗
(0.192) (0.110) (0.109) (0.056) (0.200) (0.117) (0.117) (0.059) (0.064)
age 0.0317∗∗∗ 0.0305∗∗∗ 0.0339∗∗∗ 0.0317∗∗∗ 0.0319∗∗∗ 0.0307∗∗∗ 0.0339∗∗∗ 0.0320∗∗∗ 0.0373∗∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
income -0.0435∗∗ -0.0365∗ -0.0394∗ -0.0435∗∗ -0.0455∗∗ -0.0390∗ -0.0418∗∗ -0.0469∗∗ -0.0522∗
(0.035) (0.097) (0.056) (0.050) (0.028) (0.076) (0.044) (0.034) (0.061)
stateresponse 0.0219 0.0200 -0.00596 -0.0258 0.0189 0.0162 -0.00960 -0.0324 -0.0353
(0.689) (0.718) (0.916) (0.658) (0.731) (0.771) (0.866) (0.581) (0.652)
USresponse -0.213∗∗∗ -0.158∗∗ -0.167∗∗ -0.135 -0.211∗∗∗ -0.154∗∗ -0.162∗∗ -0.127 -0.0655
(0.000) (0.017) (0.035) (0.108) (0.000) (0.021) (0.045) (0.136) (0.548)
likelydownload 0.0277 0.00951 0.0149 -0.0133 0.0325 0.0138 0.0204 -0.00801 -0.0970
(0.709) (0.900) (0.846) (0.869) (0.661) (0.854) (0.790) (0.920) (0.344)
importance 0.243∗∗∗ 0.263∗∗∗ 0.193∗∗∗ 0.205∗∗∗ 0.246∗∗∗ 0.267∗∗∗ 0.196∗∗∗ 0.211∗∗∗ 0.144
(0.001) (0.000) (0.008) (0.008) (0.001) (0.000) (0.008) (0.007) (0.193)
Usdownloads 0.00458 0.00429 0.00425 0.00442 0.00447 0.00416 0.00414 0.00424 0.00496
(0.228) (0.265) (0.271) (0.263) (0.239) (0.281) (0.283) (0.285) (0.341)
republican -0.257 -0.217 -0.270 -0.237 -0.0168
(0.208) (0.300) (0.187) (0.258) (0.949)
democrat -0.0144 -0.0208 -0.0194 -0.0294 -0.102
(0.923) (0.895) (0.897) (0.852) (0.605)
householdsize -0.0395 -0.0396 -0.0379 -0.0377 0.00495
(0.459) (0.460) (0.479) (0.486) (0.940)
trustUSgov -0.0219 -0.0307 -0.0227 -0.0316 -0.0561
(0.583) (0.465) (0.571) (0.453) (0.286)
trustinsurance 0.0124 -0.0246 0.0117 -0.0265 0.0525
(0.757) (0.580) (0.771) (0.553) (0.333)
trustCDC 0.0870∗∗∗ 0.0549 0.0882∗∗∗ 0.0538 -0.0217
(0.007) (0.114) (0.006) (0.123) (0.663)
trustmedia -0.0272 -0.0274 -0.0136
(0.463) (0.459) (0.769)
trusthealthcare 0.130∗∗∗ 0.135∗∗∗ 0.137∗∗∗
(0.000) (0.000) (0.004)
prosonly -0.165 -0.189 -0.196 -0.260 -0.412∗∗
(0.287) (0.225) (0.210) (0.105) (0.036)
prosandcons -0.155 -0.184 -0.157 -0.242 -0.234
(0.347) (0.268) (0.344) (0.150) (0.264)
applength 0.182∗∗∗
(0.005)
lesslikelytogetcovid 0.0858∗∗
(0.030)
Appmeanssafe -0.0181
(0.671)
applessinfected 0.00394
(0.920)
Applessmasks -0.210∗∗∗
(0.000)
/

Observations 1017 1002 1006 982 1017 1002 1006 982 700

p-values in parentheses
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01

42
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Table 20: Determinants of washing hands: combined controls, robust standard errors

(1) (2) (3) (4) (5) (6) (7) (8) (9)

washhands washhands washhands washhands washhands washhands washhands washhands washhands
washhands
worse 0.264 0.225 0.220 0.235 0.289 0.250 0.242 0.264 0.321
(0.193) (0.270) (0.281) (0.253) (0.154) (0.222) (0.237) (0.201) (0.236)
better 0.517∗∗ 0.507∗∗ 0.491∗∗ 0.519∗∗ 0.530∗∗ 0.517∗∗ 0.499∗∗ 0.526∗∗ 0.662∗∗
(0.016) (0.020) (0.024) (0.019) (0.014) (0.017) (0.022) (0.017) (0.031)
super -0.0114 0.00158 0.00945 0.0712 -0.00140 0.00970 0.0155 0.0756 -0.0911
(0.955) (0.994) (0.963) (0.729) (0.994) (0.962) (0.939) (0.713) (0.716)
worry 0.130∗∗∗ 0.129∗∗∗ 0.123∗∗∗ 0.130∗∗∗ 0.127∗∗∗ 0.126∗∗∗ 0.119∗∗∗ 0.126∗∗∗ 0.0851∗∗
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.020)
essential -0.335∗∗ -0.329∗ -0.309∗ -0.320∗ -0.327∗ -0.316∗ -0.306∗ -0.311∗ -0.373∗
(0.048) (0.055) (0.071) (0.061) (0.056) (0.069) (0.075) (0.072) (0.093)
econsituation -0.0756 -0.0813 -0.0643 -0.0497 -0.0675 -0.0742 -0.0567 -0.0432 -0.114
(0.479) (0.449) (0.551) (0.647) (0.526) (0.488) (0.599) (0.689) (0.382)
male -0.328∗∗ -0.333∗∗ -0.327∗∗ -0.277∗ -0.338∗∗ -0.344∗∗ -0.340∗∗ -0.290∗ -0.406∗∗
(0.032) (0.030) (0.037) (0.080) (0.028) (0.026) (0.032) (0.070) (0.044)
edu -0.0520 -0.0537 -0.0698 -0.0562 -0.0510 -0.0519 -0.0685 -0.0543 -0.0267
(0.378) (0.387) (0.251) (0.386) (0.387) (0.402) (0.259) (0.404) (0.754)
age 0.0152∗∗ 0.0166∗∗ 0.0181∗∗∗ 0.0197∗∗∗ 0.0157∗∗ 0.0174∗∗ 0.0185∗∗∗ 0.0204∗∗∗ 0.0217∗∗
(0.019) (0.015) (0.008) (0.007) (0.016) (0.012) (0.007) (0.006) (0.035)
income -0.0204 -0.0251 -0.0219 -0.0366 -0.0239 -0.0295 -0.0253 -0.0411∗ -0.0441
(0.347) (0.287) (0.323) (0.135) (0.270) (0.212) (0.251) (0.093) (0.153)
stateresponse -0.0218 -0.0203 -0.0633 -0.0501 -0.0261 -0.0245 -0.0664 -0.0549 -0.0327
(0.725) (0.748) (0.329) (0.452) (0.674) (0.697) (0.305) (0.412) (0.701)
USresponse -0.146∗∗ -0.132∗ -0.0173 -0.0626 -0.147∗∗ -0.134∗ -0.0198 -0.0644 -0.0118
(0.024) (0.073) (0.845) (0.502) (0.022) (0.068) (0.823) (0.489) (0.926)
likelydownload 0.114 0.111 0.104 0.109 0.126 0.122 0.116 0.119 0.0804
(0.216) (0.233) (0.274) (0.267) (0.170) (0.191) (0.219) (0.218) (0.538)
importance 0.106 0.0962 0.0492 0.0572 0.105 0.0964 0.0494 0.0592 0.0429
(0.201) (0.255) (0.567) (0.528) (0.211) (0.261) (0.570) (0.520) (0.745)
Usdownloads 0.00244 0.00202 0.00353 0.00441 0.00229 0.00183 0.00335 0.00416 0.00644
(0.559) (0.631) (0.407) (0.309) (0.582) (0.664) (0.433) (0.339) (0.286)
republican 0.00151 0.106 -0.00414 0.0884 -0.242
(0.995) (0.655) (0.986) (0.712) (0.438)
democrat 0.140 0.183 0.134 0.177 0.0351
(0.408) (0.303) (0.428) (0.321) (0.871)
householdsize 0.0174 0.0345 0.0250 0.0400 0.0447
(0.775) (0.583) (0.685) (0.528) (0.579)
trustUSgov -0.0754 -0.0670 -0.0752 -0.0666 -0.0574
(0.113) (0.172) (0.114) (0.174) (0.354)
trustinsurance 0.0103 -0.00140 0.0120 -0.00227 0.0261
(0.814) (0.977) (0.787) (0.963) (0.686)
trustCDC 0.119∗∗∗ 0.113∗∗∗ 0.118∗∗∗ 0.109∗∗∗ 0.0513
(0.001) (0.003) (0.001) (0.005) (0.372)
trustmedia -0.0988∗∗ -0.101∗∗ -0.108∗∗
(0.023) (0.019) (0.041)
trusthealthcare 0.0848∗∗ 0.0921∗∗ 0.142∗∗∗
(0.021) (0.014) (0.007)
prosonly -0.312∗ -0.309∗ -0.333∗ -0.354∗ -0.481∗∗
(0.085) (0.092) (0.069) (0.056) (0.032)
prosandcons -0.435∗∗ -0.466∗∗ -0.385∗∗ -0.447∗∗ -0.492∗∗
(0.016) (0.010) (0.036) (0.016) (0.043)
applength 0.142∗
(0.058)
lesslikelytogetcovid 0.0865∗
(0.053)
Appmeanssafe -0.0387
(0.400)
applessinfected -0.0346
(0.482)
Applessmasks -0.237∗∗∗
(0.000)
/

Observations 1019 1004 1008 984 1019 1004 1008 984 701

p-values in parentheses
43
∗
p < 0.10, ∗∗
p < 0.05, ∗∗∗
p < 0.01
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

Acknowledgments

We are very grateful to Pranjal Drall for the invaluable research assistance.

44
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

References

[1] World Health Organization Contact tracing in the context of COVID-19. 2020. Available at
https://www.who.int/publications/i/item/contact-tracing-in-the-context-of-covid-19
[2] Buonanno, P., and others Estimating the Severity of COVID-19: Evidence From the Italian Epi-
center. 2020. Available at https://dx.doi.org/10.2139/ssrn.3567093
[3] Hellewell J. et al. Feasibility of controlling COVID-19 outbreaks by isolation of cases and contacts.
The Lancet Global Health. 2020.
[4] Ferretti L. et al. Quantifying SARS-CoV-2 transmission suggests epidemic control with digital
contact tracing Science. 368.6491. 2020.
[5] O’Neill PH A flood of coronavirus apps are tracking us. Now it’s time
to keep track of them. MIT Technology Review 2020. available at
https://www.technologyreview.com/2020/05/07/1000961/launching-mittr-covid-tracing-tracker/
[6] Xia S and Lee G How to return to normalcy: Fast and comprehensive contact tracing of COVID-19
through proximity sensing using mobile devices. arXiv preprint arXiv:2004.12576. 2020. Available
at https://arxiv.org/abs/2004.12576
[7] McLachlan S. et al. The fundamental limitations of COVID-19 contact tracing methods and how
to resolve them with a Bayesian network approach. Researchgate.2020.
[8] Sapiezynski P et al. CThe Fallibility of Contact-Tracing Apps. arXiv preprint arXiv:2005.11297.
2020. available at https://arxiv.org/abs/2005.11297
[9] Chong C About 1 million people have downloaded TraceTogether app, but more need
to do so for it to be effective: Lawrence Wong The Straits Times. 2020. avail-
able at https://www.straitstimes.com/singapore/about-one-million-people-have-downloaded-the-
tracetogether-app-but-more-need-to-do-so-for
[10] Olsen SJ Smittestopp på iPhone fungerer bare skikkelig når appen er åpen Tek.no 2020. available
at https://www.tek.no/nyheter/nyhet/i/awVAkd/smittestopp-appen-er-oppdatert
[11] Bradshaw E et al. Information safety assurances affect intentions to use COVID-19 contact trac-
ing applications, regardless of autonomy-supportive or controlling message framing. osf.io. 2020.
available at https://osf.io/5wap8
[12] Jee C 8 million people, 14 alerts: why some covid-19 apps are staying silent. MIT Technology Re-
view 2020. available at https://www.technologyreview.com/2020/07/10/1005027/8-million-people-
14-alerts-why-some-covid-19-apps-are-staying-silent/
[13] Maccari L and Cagno V Do we need a Contact Tracing App? arXiv preprint arXiv:2005.10187.
2020. available at https://arxiv.org/abs/2005.10187
[14] Ahmed N et al. A survey of covid-19 contact tracing apps. IEEE Access. 8. 134577-134601. 2020.
[15] Peltzman S. The effects of automobile safety regulation. Journal of political Economy.83. 677-725,
1975.

45
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

[16] Cialdini RB and others Social norms and energy conservation. Social influence. 1, 3–15. 2006.
[17] Croson R and Shang JY The impact of downward social information on contribution decisions.
arXiv:2006.06822 . available at https://arxiv.org/abs/2006.06822
[18] Frey B and Meier S Social comparisons and pro-social behavior: Testing” conditional cooperation”
in a field experiment. American Economic Review. 94, 1717–1722. 2004.
[19] Nuffield Department of Medicine Latest version of COVID-19 contact tracing app
ready for testing on the Isle of Wight and Newham, London. 2020. available
at https://www.ndmrb.ox.ac.uk/about/news/latest-version-of-covid-19-contact-tracing-app-ready-
for-testing-on-the-isle-of-wight-and-newham-london
[20] TechGovSingapore Two reasons why Singapore is sticking with TraceTogether’s protocol. Tech-
Gov Singapore. 2020. available at https://www.tech.gov.sg/media/technews/two-reasons-why-
singapore-sticking-with-tracetogether-protocol
[21] EN24News the French authorities call again to install the StopCovid application. 2020.
available at https://www.en24news.com/en/2020/07/the-french-authorities-call-again-to-install-
the-stopcovid-application.html
[22] DieBundesregierung Unterstützt uns im Kampf gegen Corona. 2020. available at
https://www.bundesregierung.de/breg-de/themen/corona-warn-app/unterstuetzt-uns-im-kampf-
gegen-corona-1754756
[23] DieBundesregierung Was bringt die Corona-Warn-App und wie funktioniert sie? 2020. available
at https://www.bundesregierung.de/breg-de/themen/coronavirus/faq-corona-warn-app-1759988
[24] ANSA Conte, da lunedı̀ l’app Immuni in tutta Italia 2020. available at
[25] Agi Conte, da lunedı̀ l’app Immuni in tutta Italia 2020. available at
[26] Agi 4,6 milioni di download per Immuni, Pisano: ”Finora ha contenuto 2 focolai” 2020. available
at https://www.agi.it/cronaca/news/2020-08-05/immuni-download-pisano-9339133
[27] Farronato C at al. How to Get People to Actually Use Contact-Tracing Apps Harvard Business Re-
view. 2020. available at https://hbr.org/2020/07/how-to-get-people-to-actually-use-contact-tracing-
apps
[28] McCammon S. Virginia Unveils App To Aid Contact Tracing NPR. 2020. avail-
able at https://www.npr.org/sections/coronavirus-live-updates/2020/08/05/899414953/virginia-
unveils-app-to-aid-contact-tracing?t=1596804125091
[29] Hall Z Which U.S. states are using Apple’s Exposure Notification API for COVID-19 contact
tracing? https://9to5mac.com/2020/08/05/covid-19-exposure-notification-api-states/
[30] Bowden J RI Gov. Raimondo unveils new contact tracing mobile app ’Crush COVID RI’
https://thehill.com/homenews/state-watch/498548-ri-gov-raimondo-unveils-new-contact-tracing-
mobile-app-crush-covid-ri
[31] Kissik C. et al. What Ever Happened to Digital Contact Tracing? LawFare Blog 2020. available

46
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

at https://www.lawfareblog.com/what-ever-happened-digital-contact-tracing
[32] Yasaka T et al. Peer-to-Peer contact tracing: development of a privacy-preserving smartphone app
JMIR mHealth and uHealth. 8. e18936. 2020.
[33] Cho H. et al. Contact tracing mobile apps for COVID-19: Privacy considerations and related
trade-offs. arXiv preprint arXiv:2003.11511. 2020. available at https://arxiv.org/abs/2003.11511
[34] Raskar R et al. Apps gone rogue: Maintaining personal privacy in an epidemic
https://arxiv.org/abs/2003.08567 2020. available at https://osf.io/5wap8
[35] Abeler J and others Support in the UK for app-based contact tracing of COVID-19 files.osf.io 2020.
[36] Frimpong J and others Financial Incentives for Downloading COVID–19 Digital Contact Tracing
Apps files.osf.io 2020.
[37] Cho H. et al. How good is good enough for COVID19 apps? The influence of benefits, accu-
racy, and privacy on willingness to adopt. arXiv preprint arXiv:2005.04343. 2020. available at
https://arxiv.org/abs/2005.04343
[38] Holt M. and Murphy D. Individual versus community-level risk compensation following preexposure
prophylaxis of HIV. American journal of public health. 1568-1571. 2020.
[39] Holt M. and others. Community-level changes in condom use and uptake of HIV pre-exposure
prophylaxis by gay and bisexual men in Melbourne and Sydney, Australia: results of repeated
behavioural surveillance in 2013–17. The lancet HIV. e448-e456. 2018.
[40] Cartaud A et al. Beware of virus! Wearing a face mask against COVID-19 results in a reduction of
social distancing https://psyarxiv.com/ubzea. 2020.
[41] Yan Y. et al. Do Face Masks Create a False Sense of Security? A COVID-19 Dilemma. medrxiv
preprint 2020. Available at https://www.medrxiv.org/content/10.1101/2020.05.23.20111302v2.
[42] Ayres I. et al. Evidence from two large field experiments that peer comparison feedback can reduce
residential energy usage. The Journal of Law, Economics, and Organization. 29, 992–1022. 2013.
[43] Allcott H. Social norms and energy conservation. Journal of public Economics. 95, 1082–1095. 2011.
[44] Ferraro PJ, Price MK. Using nonpecuniary strategies to influence behavior: evidence from a large-
scale field experiment. Review of Economics and Statistics. 95, 64–73. 2013.
[45] Miller D. et al. Full genome viral sequences inform patterns of SARS-
CoV-2 spread into and within Israel. medrxiv. 2020. available at
https://www.medrxiv.org/content/10.1101/2020.05.21.20104521v1
[46] Patel NV What’s a coronavirus superspreader?. MIT Technology Review 2020. available at
https://www.technologyreview.com/2020/06/15/1003576/whats-a-coronavirus-superspreader/
[47] Centers for Disease Control and Prevention How to Protect Yourself and Others CDC Website. 2020.
Available at https://www.cdc.gov/coronavirus/2019-ncov/prevent-getting-sick/prevention.html
[48] Wiertz C. et al. Predicted Adoption Rates of Contact Tracing App Configurations-Insights from a
choice-based conjoint study with a representative sample of the UK population. Available at SSRN

47
medRxiv preprint doi: https://doi.org/10.1101/2020.09.09.20191320.this version posted September 11, 2020. The copyright holder for this
preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in
perpetuity.
It is made available under a CC-BY-ND 4.0 International license .

3589199. 2013.
[49] Leins et al. Tracking, tracing, trust: contemplating mitigating the impact of COVID-
19 through technological interventions The Medical Journal of Australia. 2020.
https://www.mja.com.au/journal/2020/tracking-tracing-trust-contemplating-mitigating-impact-
covid-19-through-technological
[50] Kang J. et al. Where not to eat? Improving public policy by predicting hygiene inspections
using online reviews Proceedings of the 2013 conference on empirical methods in natural language
processing. 95, 330–354. 2013.
[51] Pfefferbaum B and North CS Mental health and the Covid-19 pandemic New England Journal of
Medicine. 2020.
[52] Rajkumar RP COVID-19 and mental health: A review of the existing literature Asian journal of
psychiatry. 102066. 2020.
[53] Romano A et al. Covid-19 data: The logarithmic scale misinforms the public and affects policy
preferences. Health Economics. 2020.

48