Public perceptions of conflicting information surrounding COVID-19: Results from a nationally representative survey of U.S. adults.

Conflicting information surrounding COVID-19 abounds, from disagreement over the effectiveness of face masks in preventing viral transmission to competing claims about the promise of certain treatments. Despite the potential for conflicting information about COVID-19 to produce adverse public health effects, little is known about whether the U.S. public notices this information, and whether certain population subgroups are particularly likely to do so. To address these questions, we fielded a nationally representative survey of U.S. adults in late April 2020 (N = 1,007). Results showed substantial self-reported exposure to conflicting information about COVID-19, with nearly 75% of participants reporting having recently heard such information from health experts, politicians, and/or others. Participants perceived disagreement across a range of COVID-19-related issues, though from politicians more than health experts. Factors including political affiliation, information source use, and personal experience with COVID-19 were associated with perceptions of disagreement. Future research should consider potential cognitive and behavioral consequences of such perceptions.

Introduction

On February 15, 2020, World Health Organization (WHO) Director-General Tedros Adhanom Ghebreyesus cautioned, “We’re not just fighting an epidemic; we’re fighting an infodemic” [1]. Ghebreyesus was referring to the proliferation of information, and particularly misinformation, about the novel coronavirus (SARS-CoV-2) and the disease it causes, COVID-19. The rapid spread of inaccurate information and conspiracy theories about COVID-19 via social media and in other spaces poses a clear threat to public understanding and decision making. Yet while much attention to date has been directed toward documenting and combatting such misinformation [2-5], another aspect of the infodemic has been relatively overlooked: conflicting information surrounding COVID-19. Whereas misinformation refers to false information that is shared either knowingly or accidentally (e.g., wearing a mask can cause deadly rebreathing of exhaled carbon dioxide) [6], conflicting information has been defined as two or more health-related propositions that are logically inconsistent with one another (e.g., wearing a mask does versus does not help prevent viral transmission) [7]. Assessing the extent to which the public notices conflicting information about COVID-19 is critically important, given evidence that exposure to conflicting health messages can translate into adverse public health effects, including confusion about and decreased trust in health recommendations and, in turn, reduced engagement with prevention behaviors [8-10].

Since its outbreak in the United States in early 2020, discourse about COVID-19 has been characterized by substantial disagreement among politicians and health experts alike—observed across a range of issues, including who is most at risk for coronavirus infection, how dangerous such infection is, whether there is adequate access to diagnostic testing, how effective certain treatments are, and how effective personal health and policy-level strategies are in preventing the virus’s spread. Such disagreement contributes to the dissemination of conflicting information. Conflicting health information can take many forms, such as inconsistent results across research studies, distinct recommendations among professional organizations, and—perhaps most germane to the COVID-19 context—debate or disagreement about research or recommendations among key stakeholders or sources [11]. A prominent example of such disagreement has been conflicting guidance on face masks [12], but examples have emerged regarding other aspects of the pandemic as well, such as recent and related debate over the prevalence of asymptomatic transmission (i.e., asymptomatic viral spread is common versus rare) and opposing claims about whether drugs such as chloroquine and hydroxychloroquine are effective in treating COVID-19 (i.e., they are effective versus they are ineffective). Although systematic analyses of COVID-19 media coverage are not yet available in the literature, even a cursory glance at coverage to date suggests widespread reporting of conflicting information from sources including the White House [13-15]—a pattern so dominant that it has been summarized in its own right [16]—the World Health Organization [17-19], and other health experts [12, 20, 21].

In general, two major factors can give rise to conflicting health information: the very process of scientific discovery, which features incremental advances and occasional steps backward; and journalistic norms, which emphasize conflict as a core news value [11, 22]. The unique COVID-19 context, characterized by deep scientific uncertainty and severe health consequences, likely amplifies opportunities for conflicting information to arise. Scientists are pursuing questions about a new disease triggered by a novel virus, and they are doing so with great urgency. Given how much is unknown, the rapid evolution of scientific knowledge necessarily increases the likelihood of shifting evidence and, in turn, seemingly ever-changing advice [23, 24]. This picture is complicated by journalistic practice, which not only prioritizes conflict as a news value but also emphasizes novelty and, in turn, has a “default rhythm of constant piecemeal updates [that] is ill-suited to covering an event as large as the pandemic” [23]. This reporting pattern, previously described as the study du jour phenomenon [25, 26], could further underscore what the public might perceive to be frequent shifts in recommendations. The sheer volume of news attention to COVID-19—itself a function of the scale and import of the pandemic—could also multiply opportunities for exposure to conflicting information.

Although conditions are ripe for the emergence of conflicting information surrounding COVID-19, little is known about the extent to which the public notices it. Past research gives reason to believe they would: People tend to perceive conflict when such information is prevalent in the media, and this has been observed across health topics as varied as nutrition [8, 27], mammography screening [28], medications [29], and e-cigarettes [30]. Pew Research Center found initial evidence of public exposure to conflicting information about COVID-19, with 26% of Americans reporting, “I have seen mostly conflicting facts across the sources I turn to for news”; however, this generalized assessment was captured early in the outbreak (March 10–16, 2020) [31]. Observations of media coverage since then point to greater opportunities for exposure to conflict, but public perceptions of conflicting information surrounding COVID-19 need to be assessed systematically. Also unknown is whether certain population subgroups are particularly likely to notice such information. Although few studies have examined correlates of conflicting health information exposure [27, 29], factors such as sociodemographic characteristics, personal experience with COVID-19, geographic context, and use of COVID-19 information sources could shape public perceptions. For example, if someone has had personal experience with COVID-19 or lives in an area particularly hard hit by the disease, this could make it more personally salient, which, in turn, could heighten their attention to COVID-19 information and increase the likelihood that they will notice conflict and disagreement. Documenting these associations will enable us to identify subpopulations who could be particularly susceptible to conflicting information and its potential downstream consequences.

Moreover, it is important to examine whether the public perceives disagreement among health experts, politicians, or both—a necessary differentiation, given the extent to which COVID-19 has been politicized [32]. The politicization of health issues, defined as when political cues become integrated into those issues’ public presentation (e.g., when politicians’ perspectives appear in news media coverage of an issue to either endorse or highlight political conflict), has been well documented in recent years—among issues as wide-ranging as the human papillomavirus (HPV) vaccine, mammography screening, and the Affordable Care Act (ACA)) [33, 34]. In the context of COVID-19 in the U.S., political cues have been present from the outset, with rhetoric from not only the White House but also governors and other politicians; such discourse has occurred alongside COVID-19-related messaging from health experts, including federal, state, and local health department officials and scientists at academic research institutions [32]. Ultimately, then, there could be opportunities for the public to perceive disagreement among both politicians and health experts, and the level of perceived disagreement could vary across these sources. Assessing these possibilities is critical, as the source of conflicting information could influence whether people notice it and how they respond to it.

The current study addresses two overarching research questions: 1) to what extent does the U.S. public perceive conflicting information surrounding COVID-19, whether from health experts, politicians, or both; and 2) what factors are associated with these perceptions? To answer these questions, we draw on data from a nationally representative survey of U.S. adults conducted in late April 2020, in which participants reported their overall exposure to conflicting information surrounding COVID-19, as well as their perceptions of debate or disagreement among health experts and politicians across a set of specific issues. To better understand observed perceptions of disagreement, we examine several potential correlates of these perceptions, including sociodemographic characteristics, personal experience with COVID-19, geographic context, and use of COVID-19 information sources.

Materials and methods

Sample and procedure

Data reported here were collected as part of the AmeriSpeak Omnibus Survey, fielded from April 23–27, 2020 (N = 1,007). The Omnibus is a multi-client shared cost survey that is conducted bi-weekly among a nationally representative sample of ~1,000 U.S. adults aged 18 or older by NORC at the University of Chicago. Omnibus participants are drawn from NORC’s AmeriSpeak Panel, a probability-based panel of approximately 43,000 households designed to be representative of the U.S. household population. To recruit panel members, NORC randomly selects U.S. households using area probability and address-based sampling; sampled households are then contacted via mail, telephone, and face-to-face field interviews [35]. The panel provides sample coverage of approximately 97% of the U.S. household population; those excluded include those with P.O. Box only addresses, some addresses not listed in the USPS Delivery Sequence File, and some newly constructed dwellings. The Omnibus survey is administered in English in mixed mode, with approximately 85% of the interviews conducted online and 15% by phone. On average, AmeriSpeak panelists participate in 2–4 surveys per month; to minimize respondent burden, NORC limits panelist participation to 4 surveys per month.

Our team added several survey questions to the late April 2020 Omnibus instrument, a subset of which are included in the current study. Data not analyzed here come from questions that assessed participants’ perceptions of disparities in COVID-19 mortality, other COVID-19-related cognitions (e.g., self-efficacy to reduce risk of infection), patterns of information avoidance, and past mitigation behaviors (e.g., stockpiling groceries and other supplies). Data that describe levels of public awareness of disparities in COVID-19 mortality and correlates of that awareness are reported elsewhere [36]. The late April 2020 Omnibus instrument had a total average completion time of 20 minutes.

Ethics statement

The University of Minnesota Institutional Review Board approved this study (STUDY00009529), determining it to have met the criteria for exemption (Category 2). All participants had previously been consented by NORC to participate in the AmeriSpeak Omnibus Survey.

Measures

Perceptions of conflicting information surrounding COVID-19

Perceptions of conflicting information surrounding COVID-19 were assessed in two ways. To assess overall self-reported exposure to conflicting information, we adapted a previously validated measure of conflicting health information exposure to the COVID-19 context [27]. Participants were asked, “Thinking now about the past few weeks, how much conflicting or contradictory information have you heard about COVID-19 (coronavirus), whether from health experts, politicians, and/or others?” Response options included “None” (1), “A little” (2), “Some” (3), and “A lot” (4). This measure captures a generalized assessment of cross-source exposure to conflicting information about COVID-19, but it does not provide a more nuanced, source- and issue-specific examination. We therefore asked a series of questions to assess public perceptions of debate or disagreement surrounding COVID-19—the type of conflicting information most relevant to the COVID-19 context—prior to the more generalized assessment, so that participants would not be primed to think about conflicting information when responding to these more nuanced questions.

One set of questions focused on health experts, and an identical set of questions focused on politicians. For the health expert questions, we provided the following introduction, adapted from past research on public perceptions of politicized health controversies [37]: “Some health issues seem to arouse a lot of debate or disagreements among health experts, while there is more agreement on other health issues.” This was followed by defining language: “When we say health experts, we mean, for example, scientists at the Centers for Disease Control (CDC), the National Institutes of Health (NIH), such as Dr. Anthony Fauci, state or local health departments, and academic research institutions.” Then participants were asked, “Based on what you’ve read, seen or heard in the past week, how much disagreement do you think there is about the following aspects of COVID-19 (coronavirus) among health experts: i) Who is most at risk of being infected with COVID-19 (coronavirus); ii) How dangerous it is for someone to become infected with COVID-19 (coronavirus); iii) Whether there is adequate access to testing for COVID-19 (coronavirus); and iv) Whether the drugs chloroquine and hydroxychloroquine are effective in treating COVID-19 (coronavirus).” For each aspect, response options included “No disagreement” (1), “A little disagreement” (2), “Some disagreement” (3), and “A lot of disagreement” (4). Again referencing what they read, saw, or heard in the past week, participants were asked, “How much disagreement do you think there is among health experts about the effectiveness of the following strategies for preventing the spread of COVID-19 (coronavirus): i) Keeping 6 feet away from other people, except those you live with; ii) Wearing a mask or other face covering when out in public; iii) Keeping schools closed; iv) Keeping all businesses closed except those considered essential (e.g., grocery stores, pharmacies); v) Self-quarantining when sick; and vi) Washing your hands with soap several times per day.” For each strategy, response options again included “No disagreement” (1), “A little disagreement” (2), “Some disagreement” (3), and “A lot of disagreement” (4). The same sets of questions were asked about politicians, with similar introductory and defining language: “Some health issues seem to arouse a lot of debate or disagreements among politicians, while there is more agreement on other health issues. When we say politicians, we mean, for example, the president and U.S. Congress, governors and state legislators, and local mayors and city council members.” Health expert and politician question blocks appeared in random order, such that some participants saw the two health expert batteries first, and some saw the two politician batteries first. Within each battery, the order in which specific aspects or strategies appeared was randomized as well.

The four aspects and six strategies were purposively selected from the universe of COVID-19-related issues prevalent in the media during mid/late April, with the goal of including a range of issues about which perceptions of disagreement might vary. Because our primary research question was concerned with public perceptions of conflicting information surrounding COVID-19 rather than issue-specific perceptions, we averaged across items to generate four summary indices, which are the focus of our analyses: perceptions of disagreement among 1) health experts and 2) politicians about specific aspects of COVID-19, and perceptions of disagreement among 3) health experts and 4) politicians about the effectiveness of strategies for preventing the spread of COVID-19. Issue-specific perceptions are reported in an appendix (S1 and S2 Tables). For both the health expert and politician question blocks, we kept the aspects and strategies indices separate because of the conceptual distinctions between them. All six items in the strategies index assess the effectiveness of strategies for preventing the spread of COVID-19; the four items in the aspects index, though more loosely connected to one another, are concerned with perceptions of risk, testing, and treatment. The indices are not so strongly correlated so as to suggest they are capturing the same phenomena (aspects and strategies indices, health experts: r = 0.55; politicians: r = 0.55).

Correlates of perceptions of disagreement surrounding COVID-19

Potential correlates of perceived disagreement fell into several categories: sociodemographic characteristics, personal experience with COVID-19, geographic context, and use of COVID-19 information sources.

Sociodemographic characteristics. NORC provides demographic profile data as part of their standard data delivery, including gender (male, female), age (18–29, 30–44, 45–59, 60+), race/ethnicity (White, non-Hispanic; Black, non-Hispanic; Hispanic; other), education (less than high school, high school graduate or equivalent, some college, bachelor’s degree or above), and household income (<$25,000, $25,000-$49,999, $50,000-$74,999, $75,000-$99,999, $100,000+). We assessed political affiliation using a 7-point self-placement measure [38]. Participants were asked, “Generally speaking, would you call yourself…”, with response options “A strong Democrat” (1), “A Democrat” (2), “Someone who leans Democratic” (3), “An Independent” (4), “Someone who leans Republican” (5), “A Republican” (6), and “A strong Republican” (7). These options were subsequently collapsed into “Democrat,” “Independent,” and “Republican.”

Personal experience with COVID-19. Participants were asked two questions to gauge their personal experience with COVID-19. First, they were asked, “Have you been told by a doctor or other health care provider that you have COVID-19 (coronavirus)?” Response options included “No” (1), “No, but I have or have had concerning symptoms” (2), “Yes,” (3), and “I don’t know” (4). Then they were asked, “Do you personally know anyone, other than yourself, who has been told by a doctor or other health care provider that they have COVID-19 (coronavirus)?” Response options included “No” (1), “Yes,” (2), and “I don’t know” (3). The two items were combined to create a summary measure of personal experience with COVID-19 (yes to either/have or have had concerning symptoms (1), no/I don’t know (0)).

Geographic context. Geographic context was assessed in two ways. First, NORC provided participant profile data on geographic region as part of their standard data delivery (Northeast, Midwest, South, West). Second, we merged county-level COVID-19 mortality rate data from Kaiser Health News (case counts as of April 22, 2020) [39] with our survey data, which we then categorized by quartile (<1 per 100,000, 1–3 per 100,000, 3–9 per 100,000, >9 per 100,000).

Use of COVID-19 information sources. To assess participants’ COVID-19 information use patterns, we asked, “Thinking now about specific information sources, which of the following sources have you turned to for information about COVID-19 (coronavirus) in the past week: i) Fox News or its website; ii) MSNBC or its website; iii) CNN or its website; iv) NPR or its website; v) The New York Times or its website; vi) The Washington Post or its website; vii) Local television news in your area or their websites; viii) Local newspaper in your area or its website; ix) National network news (ABC World News Tonight, CBS Evening News, or NBC Nightly News) or their websites; x) White House press briefings; xi) State governor briefings; xii) Centers for Disease Control (CDC); xiii) World Health Organization (WHO); xiv) State or local health department; xv) Other people (such as family, friends, or co-workers); and xvi) Another source (specify).” Participants were asked to select all that apply; the order in which sources appeared was randomized. Participants selected 4.2 sources on average (SD = 2.7); only 22 participants selected the “another source” option without selecting at least one other source. Guided by the Pew Research Center’s public opinion work on news and information sources, both in the COVID-19 context and more generally [40, 41], specific sources were collapsed to generate broader information source categories: cable news (sources i-iii); national news (iv-vi, ix); local news (vii-viii); direct political sources (x-xi); direct health sources (xii-xiv); and interpersonal sources (xv) (yes (1), no (0) for each source category).

In addition, to gauge the extent to which participants actively looked for information, we asked, “In the past week, how often have you checked for news about COVID-19 (coronavirus) from any source?” Response options included “Every hour or more frequently” (1), “About 5–6 times a day” (2), “About 2–3 times a day” (3), “Once a day” (4), “Multiple times per week, but less than once a day” (5), “Less than once per week,” (6), and “Never” (7). These were subsequently collapsed into three categories: “Two or more times a day,” “Once a day,” and “Less than once a day.”

Analytic approach

To assess the prevalence of public perceptions of conflicting information surrounding COVID-19, frequency analyses were conducted for the generalized assessment of self-reported exposure to conflicting information, and descriptive statistics were calculated for the four summary source-specific indices of perceptions of disagreement. Multivariable linear regression models were estimated to predict each of these four summary indices: Perceptions of disagreement among 1) health experts and 2) politicians about specific aspects of COVID-19, and among 3) health experts and 4) politicians about the effectiveness of strategies for preventing the spread of COVID-19. Models included the following independent variables, as defined above: gender, age, race/ethnicity, education, household income, political affiliation, personal experience with COVID-19, region, county-level COVID-19 mortality rate, COVID-19 information sources, and frequency of checking news about COVID-19. These variables were assessed for potential collinearity using both Spearman’s correlation and variance inflation factors (VIF); there was no evidence of collinearity. As a sensitivity analysis, models were conducted to account for the potential correlation among multiple participants within county; because few participants were from a given county, and no significant differences in conclusions were observed, these models are not presented. NORC survey weights—based on national census benchmarks and balanced by gender, age, education, race/ethnicity, and region—were applied to adjust for potential biases in sampling and nonresponse to produce nationally representative estimates. Across models, P-values of < .05 were considered statistically significant. All analyses were conducted in SAS version 9.4 (Cary, NC).

Results

Participant characteristics

Just over one-third of participants (35.2%) reported having had personal experience with COVID-19. About 40% of participants (42.8%) were Democrats; 29.8% were Republicans, and 27.4% were Independents. The information sources that participants reported turning to most for information about COVID-19 were local news (55.6%), direct political sources (53.2%), cable news (51.6%), and direct health sources (46.6%). Nearly half of participants (48.9%) reported checking for news about COVID-19 two or more times a day. Additional participant characteristics are reported in Table 1.

Table 1

Sample characteristics (N = 1,007).

Variable	Weighted %a
Gender
Male	48.6
Female	51.4
Age (years)
18–29	18.1
30–44	26.7
45–59	24.5
60+	30.7
Race/ethnicity
White, non-Hispanic	62.6
Black, non-Hispanic	12.0
Hispanic	16.5
Other	8.9
Education
Less than high school	8.8
High school graduate or equivalent	27.5
Some college	28.5
Bachelor’s degree or above	35.3
Household income
<$25,000	20.5
$25,000-$49,999	25.6
$50,000-$74,999	18.5
$75,000-$99,999	12.8
$100,000+	22.5
Political affiliation
Republican	29.8
Independent	27.4
Democrat	42.8
Personal experience with COVID-19
Yes	35.2
No	64.8
Region
Northeast	17.6
Midwest	20.7
South	37.8
West	23.9
County-level COVID-19 mortality rate
<1 per 100,000	23.9
1–3 per 100,000	25.3
3–9 per 100,000	26.4
>9 per 100,000	24.4
COVID-19 information sourcesb
Cable news	51.6
National news	24.5
Local news	55.6
Direct political sources	53.2
Direct health sources	46.6
Interpersonal sources	23.3
How often check news about COVID-19
Two or more times a day	48.9
Once a day	28.5
Less than once a day	22.6

a Percentages may not sum to 100 due to missing data and rounding.

b Participants could check all that apply.

Perceptions of conflicting information surrounding COVID-19

Overall, nearly three-quarters of participants (72.3%) reported hearing some or a lot of conflicting information about COVID-19, whether from health experts, politicians, and/or others; only 3.3% reported no exposure to such information. Consistent with this pattern, participants reported perceiving debate or disagreement about specific aspects of COVID-19 among both health experts (M = 2.28, 95% CI: 2.21–2.34, range = 1–4, Cronbach’s α = 0.79) and politicians (M = 2.68, 95% CI: 2.61–2.74, range = 1–4, Cronbach’s α = 0.80), where 2 was “a little disagreement” and 3 was “some disagreement.” They also reported perceiving disagreement about the effectiveness of strategies for preventing viral spread, although to a lesser extent—again among health experts (M = 1.61, 95% CI: 1.55–1.66, range = 1–4, Cronbach’s α = 0.84) and politicians (M = 1.97, 95% CI: 1.91–2.03, range = 1–4, Cronbach’s α = 0.85). Although disagreement was observed across these two sources, participants perceived even greater disagreement among politicians, as evident in the non-overlapping confidence intervals for both the aspects and strategies indices (Tables 2 and 3, respectively; both p < .001) and issue-specific perceptions (S1 and S2 Tables).

Table 2

Multivariable linear regression models predicting perceptions of disagreement among health experts and politicians about aspects of COVID-19 (coronavirus).

	Among health experts			Among politicians
	N	M (SE)	95% CI	N	M (SE)	95% CI
Summary index (range = 1–4)	996	2.28 (0.03)	2.21, 2.34	997	2.68 (0.03)	2.61, 2.74
	Among health experts (N = 977)			Among politicians (N = 977)
Variable	b	SE	p	b	SE	p
Gender			0.528			0.696
Female (ref)	0.00			0.00
Male	0.04	0.07		-0.03	0.07
Age (years)			0.058			0.081
18–29 (ref)	0.00			0.00
30–44	-0.18	0.11		0.01	0.12
45–59	-0.26	0.12		0.05	0.12
60+	-0.06	0.12		0.21	0.12
Race/ethnicity			0.806			0.035
White, non-Hispanic (ref)	0.00			0.00
Black, non-Hispanic	0.08	0.11		-0.07	0.12
Hispanic	-0.06	0.12		-0.34	0.12
Other	0.03	0.12		-0.01	0.12
Education			0.746			0.010
Less than high school (ref)	0.00			0.00
High school graduate or equivalent	-0.07	0.18		-0.14	0.15
Some college	-0.04	0.17		-0.05	0.14
Bachelor’s degree or above	-0.11	0.17		0.14	0.14
Household income			0.247			0.822
<$25,000 (ref)	0.00			0.00
$25,000-$49,999	-0.26	0.11		-0.12	0.12
$50,000-$74,999	-0.17	0.12		-0.06	0.12
$75,000-$99,999	-0.22	0.12		-0.07	0.12
$100,000+	-0.20	0.11		-0.04	0.12
Political affiliation			0.437			0.001
Republican (ref)	0.00			0.00
Independent	-0.03	0.09		0.28	0.09
Democrat	-0.11	0.09		0.31	0.09
Personal experience with COVID-19			0.020			0.045
No (ref)	0.00			0.00
Yes	0.16	0.07		0.14	0.07
Region			0.125			0.653
Northeast (ref)	0.00			0.00
Midwest	0.10	0.11		0.12	0.10
South	0.17	0.11		0.05	0.11
West	0.27	0.12		0.09	0.11
County-level COVID-19 mortality rate			0.757			0.606
<1 per 100,000 (ref)	0.00			0.00
1–3 per 100,000	0.02	0.09		0.05	0.09
3–9 per 100,000	-0.06	0.09		0.08	0.09
>9 per 100,000	0.02	0.11		-0.04	0.10
COVID-19 information sources
Cable news			0.526			0.269
No (ref)	0.00			0.00
Yes	0.04	0.07		0.08	0.07
National news			0.061			0.043
No (ref)	0.00			0.00
Yes	-0.16	0.09		0.16	0.08
Local news			0.524			0.364
No (ref)	0.00			0.00
Yes	0.04	0.07		0.06	0.07
Direct political sources			0.090			0.086
No (ref)	0.00			0.00
Yes	-0.12	0.07		-0.12	0.07
Direct health sources			0.103			0.015
No (ref)	0.00			0.00
Yes	0.12	0.07		0.18	0.07
Interpersonal sources			0.516			0.458
No (ref)	0.00			0.00
Yes	0.05	0.08		0.06	0.07
How often check news about COVID-19			0.190			0.774
Less than once a day (ref)	0.00			0.00
Once a day	-0.07	0.10		-0.07	0.10
Two or more times a day	-0.16	0.10		-0.04	0.10

Table 3

Multivariable linear regression models predicting perceptions of disagreement among health experts and politicians about the effectiveness of strategies for preventing the spread of COVID-19 (coronavirus).

	Among health experts			Among politicians
	N	M (SE)	95% CI	N	M (SE)	95% CI
Summary index (range = 1–4)	1002	1.61 (0.03)	1.55, 1.66	1000	1.97 (0.03)	1.91, 2.03
	Among health experts (N = 981)			Among politicians (N = 980)
Variable	b	SE	p	b	SE	p
Gender			0.393			0.816
Female (ref)	0.00			0.00
Male	-0.05	0.05		-0.02	0.07
Age (years)			0.016			0.020
18–29 (ref)	0.00			0.00
30–44	-0.23	0.10		-0.31	0.12
45–59	-0.31	0.10		-0.37	0.12
60+	-0.27	0.10		-0.33	0.12
Race/ethnicity			0.928			0.003
White, non-Hispanic (ref)	0.00			0.00
Black, non-Hispanic	0.03	0.10		-0.07	0.12
Hispanic	-0.03	0.10		-0.33	0.09
Other	0.03	0.08		-0.20	0.11
Education			0.076			0.788
Less than high school (ref)	0.00			0.00
High school graduate or equivalent	0.15	0.14		0.06	0.14
Some college	-0.06	0.13		0.04	0.13
Bachelor’s degree or above	-0.03	0.13		0.10	0.13
Household income			0.461			0.253
<$25,000 (ref)	0.00			0.00
$25,000-$49,999	-0.15	0.09		-0.15	0.010
$50,000-$74,999	-0.15	0.10		0.03	0.11
$75,000-$99,999	-0.09	0.11		-0.01	0.11
$100,000+	-0.14	0.09		-0.02	0.11
Political affiliation			0.096			0.688
Republican (ref)	0.00			0.00
Independent	-0.12	0.08		0.06	0.08
Democrat	-0.16	0.07		0.07	0.09
Personal experience with COVID-19			0.090			0.030
No (ref)	0.00			0.00
Yes	0.10	0.06		0.14	0.07
Region			0.117			0.943
Northeast (ref)	0.00			0.00
Midwest	0.12	0.09		-0.01	0.10
South	0.13	0.09		-0.02	0.10
West	0.21	0.09		0.03	0.10
County-level COVID-19 mortality rate			0.329			0.791
<1 per 100,000 (ref)	0.00			0.00
1–3 per 100,000	-0.14	0.08		-0.06	0.08
3–9 per 100,000	-0.07	0.07		0.02	0.08
>9 per 100,000	-0.07	0.09		-0.04	0.09
COVID-19 information sources
Cable news			0.141			0.312
No (ref)	0.00			0.00
Yes	0.09	0.06		0.06	0.06
National news			0.058			0.021
No (ref)	0.00			0.00
Yes	-0.13	0.07		0.18	0.08
Local news			0.727			0.702
No (ref)	0.00			0.00
Yes	0.02	0.05		-0.02	0.06
Direct political sources			0.031			0.028
No (ref)	0.00			0.00
Yes	-0.13	0.06		-0.14	0.07
Direct health sources			0.200			0.012
No (ref)	0.00			0.00
Yes	0.08	0.06		0.16	0.06
Interpersonal sources			0.799			0.785
No (ref)	0.00			0.00
Yes	-0.02	0.07		-0.02	0.07
How often check news about COVID-19			0.217			0.202
Less than once a day (ref)	0.00			0.00
Once a day	-0.06	0.08		-0.12	0.09
Two or more times a day	-0.13	0.08		-0.17	0.09

Correlates of perceptions of disagreement surrounding COVID-19

Several factors were significantly associated with perceptions of disagreement about specific aspects of COVID-19 (Table 2). Participants who reported personal experience with COVID-19 perceived greater disagreement among both health experts (p = .020) and politicians (p = .045) than did those without personal experience. Political affiliation also was significantly associated with perceptions of disagreement, but only for disagreement among politicians (p = .001): both Democrats and Independents tended to perceive greater disagreement among politicians than did Republicans. Other sociodemographic characteristics associated with perceived disagreement included race/ethnicity and education. Hispanic participants tended to perceive less disagreement among politicians (p = .035), compared with non-Hispanic White participants, and participants with a Bachelor’s degree or above reported perceiving more disagreement among politicians (p = .010) than those with less than a high school education. Several information sources were associated with perceptions of disagreement as well. Specifically, participants attending to national news reported perceiving greater disagreement among politicians (p = .043) and, to some extent, less disagreement among health experts (p = .061) than those who did not use this source to obtain information about COVID-19. Those attending to direct health sources reported perceiving greater disagreement among politicians (p = .015).

There were also several factors significantly associated with perceptions of disagreement about the effectiveness of strategies for preventing the spread of COVID-19 (Table 3). Again participants with personal experience with COVID-19 reported perceiving greater disagreement, though this association was statistically significant only for disagreement among politicians (p = .030). Compared to the aspects findings, political affiliation was not significantly associated with perceptions of disagreement among either politicians (p = .688) or health experts (p = .096). Race/ethnicity was again associated with perceived disagreement, with Hispanic participants tending to perceive less disagreement among politicians (p = .003), compared with their non-White Hispanic counterparts. In addition, older participants reported perceiving less disagreement than younger participants, and this was observed among both health experts (p = .016) and politicians (p = .020). Consistent with the aspects results, those attending to national news reported perceiving greater disagreement among politicians (p = .021) and, to some extent, less disagreement among health experts (p = .058). Participants attending to direct political sources reported perceiving less disagreement among both health experts (p = .031) and politicians (p = .028), whereas those attending to direct health sources reported perceiving greater disagreement among politicians (p = .012).

Discussion

The primary goal of this study was to assess the extent to which the U.S. public perceives conflicting information surrounding COVID-19. This dimension of the COVID-19 infodemic has been not been assessed systematically, despite widespread media coverage of conflicting information about COVID-19 and the potential for such content to produce adverse affective, cognitive, and behavioral responses. To address this research question, we fielded a nationally representative survey of ~1,000 U.S. adults in late April 2020 and found substantial self-reported exposure to conflicting information about COVID-19, with nearly 75% of participants reporting that in recent weeks they had heard such information from health experts, politicians, and/or others. In addition, results showed that participants perceived disagreement across a range of COVID-19-related issues, among both health experts and politicians, suggesting that such exposure is not confined to particular issues or sources and instead may be cumulative. That said, there were systematic differences in perceived disagreement across sources—overall, participants noticed more disagreement among politicians than health experts—and mean levels of perceived disagreement tended to be higher for specific aspects of COVID-19 than the effectiveness of strategies for preventing its spread.

These descriptive patterns may be better understood by examining factors associated with perceived disagreement surrounding COVID-19. For example, there was at least some evidence that perceptions could be shaped by one’s political affiliation. For specific aspects of COVID-19, Democrats and Independents perceived greater disagreement among politicians than did Republicans. This pattern could reflect motivated reasoning [42]: If the White House is what came to mind when Democrats and Independents read “politicians,” then that cue could have prompted them to perceive and ultimately interpret disagreement in ways consistent with their political beliefs (e.g., the president routinely contradicts himself, does not tell the truth, and so forth), whereas the same cue could have triggered the opposite response among Republicans (e.g., the White House is a trustworthy source). Future research should investigate these possibilities, given scholarly concern about motivated reasoning in the COVID-19 context [32].

Additionally, where and how people obtain their information about COVID-19 might shape their perceptions of disagreement. For instance, across aspects and strategies, participants who reported turning to direct health sources (international/national health organizations like WHO and CDC, state and local health departments) tended to perceive more disagreement among politicians. In contrast, those attending to direct political sources (White House and/or governor briefings) tended to perceive less disagreement among both health experts and politicians. We can only speculate as to why, but it could be that there is more unified messaging delivered via government briefings—which, at the federal and state levels, can feature both health experts and politicians. By comparison, direct health sources, such as health department and organization officials, might more readily acknowledge scientific uncertainty and the concomitant shifts in guidance, yet the public might attribute such shifts in guidance to political sources involved in such communications (e.g., state government officials working for the health department). Further, there was evidence that participants who reported turning to national news perceived greater disagreement among politicians and, to some extent, less disagreement among health experts. A cursory review of national news headlines suggests that both politicians and health experts have issued contradictory messages surrounding COVID-19 [12, 13, 17], but the public might perceive this to be particularly true of politicians.

Results also showed that one’s personal experience with COVID-19 was associated with perceptions of conflict, whereas the broader geographic context in which one lives seemed to exert less influence. Personal experience might have made the disease more salient, perhaps encouraging participants to pay greater attention and, in turn, increase the likelihood that they notice more disagreement—especially about issues such as how dangerous it is and possible treatments. Where someone lives, and whether there was substantial COVID-19 mortality in their area, did not seem to be consequential here, at least in a multivariable model that took people’s personal experience with the disease into account. There was also evidence of associations with certain demographics beyond political affiliation, such as Hispanic ethnicity and age, though it will be important to see whether such patterns hold up in future studies.

This study has several strengths, including the use of population-based data and the adaptation of survey measures previously developed to assess self-reported exposure to conflicting health information [27] and perceptions of politicized health controversies [37]. Nonetheless, results should be considered with several limitations in mind. First, study results are based on weighted data, which should reflect the distributions in the population of noninstitutionalized, English-speaking U.S. adults aged 18 and older; that said, the NORC survey weights incorporate several variables (gender, age, education, race/ethnicity, and region) but not political affiliation. There are more Democrats in the study sample, and generalizability might be constrained with respect to political affiliation. Second, survey space constraints limited our ability to look at disagreement between health experts and politicians. Our global assessment of cross-source exposure to conflicting information surrounding COVID-19 suggests that perceptions of such disagreement likely exist, but future research should directly address this question. Third, to avoid priming conflict and activating responses, the source- and issue-specific measures of perceived disagreement preceded the global assessment measure; however, this ordering may have resulted in overreporting of cross-source exposure to conflicting information about COVID-19. Fourth, although these measures were informed by past research [27, 36], they did not undergo cognitive testing prior to study launch; lower literacy participants might have found the definitions to be challenging. Fifth, although this cross-sectional survey study enables us to identify factors that are associated with perceptions of disagreement, inferences regarding causality cannot be established. Sixth, we report perceptions observed in the U.S.; it is not clear whether similar perceptions have emerged in other countries, though some research hints at this possibility [43]. Last, the four aspects and six strategies were selected from the universe of COVID-19-related issues prevalent in the media during mid/late April; given the rapid changes in COVID-19 science and, in turn, the fluidity of media and public discourse, it is conceivable that repeating this survey later in 2020 would yield different public perceptions of disagreement about one or more issues. It is for this reason that we report and predict the summary indices of the broader phenomena that those responses instantiate, rather than the issue-specific perceptions.

Rapidly evolving science necessarily can create conditions ripe for disagreement among experts, and when that science is politicized, politicians can be an additional source of conflict. This study suggests that, overall, such disagreements are not confined to professional circles, but instead they play out in the broader public information environment, and this content does not go unnoticed by the public. Whether such perceptions of disagreement are ultimately consequential is an empirical question that should be addressed in future longitudinal survey or experimental research—after all, just because someone perceives disagreement among politicians does not necessarily mean they are insecure in their own beliefs—but existing theory and research on the effects of exposure to conflicting health information raise cause for concern [8–11, 43]. Although researchers are well prepared to make sense of evolving scientific evidence and health recommendations, the public may struggle to do so, due in part to limited literacy about scientific research [11]. Several experimental studies have found that exposure to conflicting health information produces negative emotional reactions to such content (such as frustration, annoyance, and distress) [9] and undesirable cognitive outcomes, including confusion (perceived ambiguity about the health topic in question or health research in general) [8-10], backlash (negative beliefs or attitudes toward the health topic in question or health research in general) [10, 44], and attitudinal ambivalence (positive and negative evaluations of a given object at the same time) [9, 44]. Moreover, there is some evidence that these affective and cognitive responses could translate into behavioral effects [8]. This is a pressing concern with COVID-19, as perceptions of conflict and disagreement could produce not only confusion about and decreased trust in recommendations but also reduced compliance with mitigation behaviors, including those for which there is substantial consensus (e.g., hand washing). Such undesirable effects could be particularly likely if those recommendations are perceived to be coming from politicians, given that, across issues, participants perceived politicians to be debating and disagreeing more than health experts. That said, our results suggest that such effects could vary across population subgroups and might be shaped by factors such as political affiliation, information source use, and personal experience with COVID-19.

This study’s findings have important implications for public health communication research and practice. First, results underscore that the public perceives disagreement among health experts and, to an even greater degree, among politicians. Such patterns might not be confined to the COVID-19 context, but could in fact be likely whenever a health issue becomes politicized—a troubling trend that has been documented in recent years [33]. Because the source of conflicting information could influence both whether people notice it and how they respond to it, future research should examine the myriad sources of such content. Second, we need to anticipate that many people likely interpret local, state, and federal health department and organization COVID-19 recommendations against a backdrop of seemingly ever-shifting advice, which could undermine the effectiveness of that strategic messaging. Crisis risk communication research suggests that public health messengers should anticipate such perceived conflict and respond accordingly, by exhibiting compassion and explicitly acknowledging uncertainty and shifts [45]. Some efforts already have been undertaken here; for example, Harvard T.H. Chan School of Public Health, in its “COVID-19 Path Forward” plan, underscores unified guidance and consensus while simultaneously acknowledging and normalizing evolving scientific advice [46]. Health journalists could follow similar strategies in their reporting. Just as there are growing efforts to intervene and address COVID-19-related misinformation [3, 47, 48], so, too, should attention be devoted to addressing conflicting information.

Issue-specific perceptions of disagreement among health experts and politicians about aspects of COVID-19 (coronavirus).

(DOCX)

Click here for additional data file.

Issue-specific perceptions of disagreement among health experts and politicians about the effectiveness of strategies for preventing the spread of COVID-19 (coronavirus).

(DOCX)

Click here for additional data file.

7 Aug 2020

PONE-D-20-22073

Public perceptions of conflicting information surrounding COVID-19: Results from a nationally representative survey of U.S. adults

PLOS ONE

Dear Dr. Nagler,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Sep 21 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Sze Yan Liu, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please improving statistical reporting and refer to p-values as "p<.001" instead of "p=.000". Our statistical reporting guidelines are available at https://journals.plos.org/plosone/s/submission-guidelines#loc-statistical-reporting

3. Please provide additional details regarding participant consent. In the ethics statement in the Methods and online submission information, please ensure that you have specified what type you obtained (for instance, written or verbal, and if verbal, how it was documented and witnessed). If your study included minors, state whether you obtained consent from parents or guardians. If the need for consent was waived by the ethics committee, please include this information.

4. We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thank you for the opportunity to review “Public perceptions of conflicting information surrounding COVID-19: results from a nationally representative survey of U.S. adults.” The article presents interesting and timely findings regarding whether the public notices conflicting COVID information, particularly between health experts and politicians, and correlates of perception of conflicting information.

Abstract

• Well-written

Introduction

• The introduction is well-written and presents relevant previous studies and a rationale for the current study.

Methods

• Define the acronym NORC

• Recommend cutting the sentence on line 155 starting with, “as a multi-client.” Sounds like an unnecessary endorsement of NORC.

• Can you note how long the total survey was so we can get an idea of how long the survey was and respondent burden?

• Some of the conflicting information questions seem to be at a high level of literacy. For example, the definition of health experts is long and it is unlikely the general public knows what an “academic research institution” is. Were these questions piloted with individuals with low health literacy? If not, this should be noted as a limitation.

• Why was social media left off as a source for COVID-19 info?

• Line 277: please provide more detail about the regressions, including what the dependent variables were in the models.

• Please state what level of significance was (alpha)

Results

• Line 304: since the means are between 2 and 3, can you present those response options rather than the 1 and 4 response options. Can you also comment on whether these data were skewed or kurtotic?

• Line 311: please change p=.000 to p<0.001

• There are a lot of variables included in the regression- did you check for collinearity?

• Have you thought about entering the number of sources as a variable in the model rather than listing cable news, national news, local news, etc, separately?

Discussion

• Nice discussion of motivated reasoning

• Limitations should also mention that there were more Democrats and well-educated respondents in the sample and how this may affect the results

Reviewer #2: This is a very well written article dealing with a delicate topic. The paper presents the data of nationally representative surveys and study design and statistical analyses are strong. Thus, it appeal to readers' interest and/or help to understand the impact of conflicting information surrounding COVID-19 abounds on non-pharmaceutical intervention. Thanks for your outstanding contributions.

Reviewer #3: The purpose of this study was to examine perceptions of conflicting information from both health experts and politicians about COVID-19 among a nationally representative sample of slightly over 1,000 U.S. adults. Data were collected in late April 2020. Three-quarters of respondents reported being exposed to conflicting information, and the authors report demographic and other characteristics associated with perceptions of conflicting information. The study is primarily descriptive and is not hypothesis-driven. I enjoyed reading this paper and found the information to have great implications for health communication strategies surrounding COVID-19. The introduction does a very nice job of situating this study in the existing literature on conflicting health information. Strengths of the paper include a thorough and succinct introduction, national sampling procedures, counter balancing of measures (i.e. when measures were duplicated about health experts and politicians), and a strong discussion section. The methods and analyses are largely solid. As the authors note in the abstract, it will be important to understand how these beliefs are associated with cognitive and behavioral outcomes, and it is unfortunate that measures of that type were not included in this paper. I describe minor suggestions for improving the paper.

Introduction

1. Do the findings of this study have implications beyond understanding COVID? This could be addressed more in the intro and/or discussion.

2. Page 5, line 97: “what seem like daily swings in recommendations” sounds subjective and I think different terminology might be used to emphasize the frequency without this language.

3. I had difficulty understanding the definition of the politicization of health issues (pg 6, lines 122-123). An example might help to clarify.

Materials and Methods

4. For transparency, could the authors describe the other measures that were included in the survey but that are not analyzed here?

5. More of a description/explanation about NORC would be helpful. For example, what exactly does it mean that the panel “provides sample coverage of approximately 97% of the U.S. household population”? Could this be stated in more plain language? Also, how big is the panel, and do panel members complete multiple surveys?

6. Four summary indices of perceptions of conflicting information are reported. What is the justification for keeping these separate rather than combining all the items referring to health experts and all of those referring to politicians (for a total of two scales)? I would like to see the alphas of the scales and correlations among the scales.

7. Additional justification is needed for how the specific news sources were collapsed into the categories of cable news, national news, etc. I would think that the sources categorized as cable news tend to be more conservative than those classified as national news, which is a potential limitation/confound of the categories reported. I am not sure how meaningful the categories are as reported. I would imagine that relying on different news sources differs by political affiliation in meaningful ways, and the authors might consider accounting for this in their analyses.

Discussion

8. I was unclear what the sentence starting with “By comparison…” (page 23, lines 392-395) meant.

9. Please add citations to the statement that theory and research on the effects of exposure to conflicting health information raise cause for concern (page 25, lines 439-440).

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

28 Aug 2020

RESPONSE TO REVIEWERS

MANUSCRIPT: Public perceptions of conflicting information surrounding COVID-19: Results

from a nationally representative survey of U.S. adults (PONE-D-20-22073)

DATE: August 28, 2020

Thank you very much for the expeditious review of our manuscript, “Public perceptions of conflicting information surrounding COVID-19: Results from a nationally representative survey of U.S. adults.” Based on the reviewers’ helpful suggestions, we have made several improvements to the manuscript. Below we outline the concerns raised and our responses. In the revised manuscript, we indicate all revisions using track changes.

Reviewer #1

1) In the sample and procedure section, define the acronym NORC; consider cutting the sentence on line 155, starting with “As a multi-client shared cost survey,” which sounds like an unnecessary endorsement of NORC; and note the total length of the survey.

We have added information on the total survey length to the sample and procedure section (pp. 7-8). We appreciate the reviewer’s point about the sentence on line 155. We agree that we should be mindful not to inadvertently suggest a dependent relationship, so we deleted the part that sounded like an endorsement. We do think it is important for readers to know that these data come from a multi-client shared cost survey so they can appreciate that the survey included questions that were not designed by our team (and thus are not part of the current project). For this reason, we retained that content. And strange though it may seem, NORC is not an acronym; rather, it is the organization’s formal name, “as with organizations such as IBM, AT&T, RAND, and GEICO” (https://www.norc.org/about/Pages/about-our-name.aspx).

2) Some of the conflicting information questions seem to be at a high level of literacy. For example, the definition of health experts is long, and it is unlikely the general public knows what an “academic research institution” is. Were these questions piloted with individuals with low health literacy? If not, this should be noted as a limitation.

The reviewer raises a good point. Although our measurement approach was informed by past

research on self-reported exposure to conflicting health information and perceptions of politicized

health controversies, the specific measures used in this study were not piloted prior to study launch

due to time constraints (i.e., we prioritized timely data collection given the rapidly evolving crisis).

In developing our health experts definition, we did look to public opinion surveys that asked about

trust in specific health expert sources (e.g., https://www.kff.org/coronavirus-covid-19/report/kff-health-tracking-poll-early-april-2020/), which might allay some concerns about differential understanding across literacy levels. We nonetheless now note this as a limitation (p. 26).

3) Why was social media left off as a source for COVID-19 information?

This survey question, which asked participants about the “specific information sources” they turned to for information about COVID-19, focused on substantive source content rather than the precise vehicle of information delivery. For example, if a participant selected “NPR or its website,” “National network news or their websites,” and “World Health Organization (WHO),” that tells us something about the types of substantive information sources they sought out, but in each case they might have engaged with that content via a different vehicle. In other words, they might have listened to NPR on the radio, visited the NPR website, and/or followed NPR on Twitter. Regardless of the vehicle of delivery, the type of content (which, in the case of NPR, would be national news content) is the same. Although it is true that there could be some variation in information engagement patterns depending on the vehicle of information delivery (e.g., a participant might engage more deeply with NPR content they seek out and read online versus hear in passing on the car radio or in a Tweet they see on a timeline), survey space constraints precluded this more nuanced parsing.

4) In the analytic approach section, please provide more detail about the regressions, including what the dependent variables were in the models, and state what the significance level was (alpha). Also, did you check for collinearity? There are a lot of variables in the regression models.

We added more details about the regressions and stated the significance level; we also specify that there was no evidence of collinearity in the regression models (pp. 13-14). Most Spearman correlation coefficients among independent variables were less than 0.1, and none were greater than 0.3; additionally, most variance inflation factors were below 3, and all were well below 10.

5) In the results section, since the means for the summary indices are between 2 and 3, present those response options instead (line 304), and change the p-value on line 311 to <0.001. For the summary indices, can you comment on whether those data were skewed or kurtotic?

These changes have been made (pp. 16-17). Regarding the question of skewness and kurtosis, the indices do not deviate substantially from normality (statistics are listed below). The aspects indices are symmetric, though lighter tailed; the strategies indices are moderately to highly skewed, and the health experts distribution is heavier tailed.

Aspects index, health experts: skewness = 0.1, kurtosis = -0.9

Aspects index, politicians: skewness = -0.4, kurtosis = -0.5

Strategies index, health experts: skewness = 1.5, kurtosis = 2.4

Strategies index, politicians: skewness = 0.7, kurtosis = 0.2

6) Have you thought about entering the number of sources as a variable in the model, rather than listing cable news, national news, local news, etc. separately?

We appreciate the reviewer’s question, as we agree that the volume of information exposure also could prove important. To consider this possibility, we asked participants to report the frequency with which they actively checked for news about COVID-19 from any source (p. 13). Although tallying the number of information sources could be an alternative approach to assessing the volume of exposure, conceivably someone might report using only two sources but actually use them quite heavily, while someone else might report using five or more sources but use them sporadically at best. We believe the self-reported assessment of frequency of checking for news provides a useful complement to the source-specific measures of information use.

7) Mention as a limitation that there were more Democrats and well-educated respondents in the sample, noting how this might affect the results.

We thank the reviewer for raising this issue. Study results are based on weighted data, which should reflect distributions in the population. This allays many, though not all, of the concerns about generalizability. We now mention this in the limitations section (p. 25).

Reviewer #2

This is a very well written article dealing with a delicate topic. The paper presents data from a nationally representative survey, and study design and statistical analyses are strong. Thus, it appeals to readers' interest and/or helps to understand the impact of conflicting information surrounding COVID-19. Thanks for your outstanding contributions.

Thank you very much; we appreciate your taking the time to review this work.

Reviewer #3

1) Do the findings of this study have implications beyond understanding COVID? This could be addressed more in the introduction and/or discussion.

We thank the reviewer for asking this question, as we do believe our findings have implications beyond the COVID-19 context. We therefore added text to the implications section of the discussion (pp. 27-28), where we describe the importance of the public perceiving disagreement not only among health experts but also (and to an even greater degree) among politicians. We note how such source patterns might be likely whenever a health issue becomes politicized.

2) On line 97, “what seem like daily swings in recommendations” sounds subjective; different terminology might be used to emphasize the frequency.

We have reworded this clause to read: “what the public might perceive to be frequent shifts in recommendations” (p. 5).

3) On lines 122-123, it was difficult to understand the definition of the politicization of health issues; consider adding an example to help clarify.

We have added an example to help clarify. The text now reads: “The politicization of health issues, defined as when political cues become integrated into those issues’ public presentation (e.g., when politicians’ perspectives appear in news media coverage of an issue to either endorse or highlight political conflict), has been well documented in recent years—among issues as wide-ranging as the human papillomavirus (HPV) vaccine, mammography screening, and the Affordable Care Act (ACA)) [33, 34]” (p. 6).

4) For transparency, describe the other measures that were included in the survey that are not analyzed here.

This information has been added on p. 8: “Data not analyzed here come from questions that assessed participants’ perceptions of disparities in COVID-19 mortality, other COVID-19-related cognitions (e.g., self-efficacy to reduce risk of infection), patterns of information avoidance, and past mitigation behaviors (e.g., stockpiling groceries and other supplies).” Some of these data are reported in a separate manuscript. If the editorial team prefers it, we would be glad to include in an appendix the full set of survey questions that we added to NORC’s AmeriSpeak Omnibus instrument.

5) More of a description/explanation about NORC would be helpful. For example, what exactly does it mean that the panel “provides sample coverage of approximately 97% of the U.S. household population”; how big is the panel; and do panel members complete multiple surveys?

We have included additional information about NORC in the sample and procedure section (pp. 7-8). Specifically, we have clarified which types of households are excluded from the panel, noted the panel’s current size, and described panelists’ typical survey completion patterns.

6) Four summary indices of perceptions of conflicting information are reported. What is the justification for keeping these separate rather than combining all the items referring to health experts and all of those referring to politicians (for a total of two scales)? I would like to see the alphas of the scales and correlations among the scales.

For both the health expert and politician question blocks, we kept the indices separate because of the conceptual distinctions between them. All six items in the strategies index assess the effectiveness of strategies for preventing the spread of COVID-19. In contrast, the four items in the aspects index, though more loosely connected to one another, do not focus on the effectiveness of prevention strategies; rather, they are concerned with perceptions of risk, testing, and treatment.

As requested, we are providing the alphas and correlations below; the alphas also have been added to the manuscript on p. 16. The aspects and strategies indices are not so strongly correlated so as to suggest they are capturing the same phenomena.

Aspects index, health experts: Cronbach’s alpha = 0.79

Aspects index, politicians: Cronbach’s alpha = 0.80

Strategies index, health experts: Cronbach’s alpha = 0.84

Strategies index, politicians: Cronbach’s alpha = 0.85

Correlation between aspects and strategies indices, health experts = 0.55

Correlation between aspects and strategies indices, politicians = 0.55

7) Provide additional justification for how the specific news sources were collapsed into the categories of cable news, national news, etc. I would think that the sources categorized as cable news tend to be more conservative than those classified as national news, which is a potential limitation/confound of the categories reported. I also would imagine that relying on different news sources differs by political affiliation in meaningful ways; the authors might consider accounting for this in their analyses.

We appreciate the reviewer’s point and agree that we should have provided such justification. Our measurement decisions were informed by the Pew Research Center’s public opinion work on news and information sources, both in the COVID-19 context and more generally (see, for example, https://www.journalism.org/2020/03/25/americans-who-primarily-get-news-through-social-media-are-least-likely-to-follow-covid-19-coverage-most-likely-to-report-seeing-made-up-news/ and https://www.pewresearch.org/fact-tank/2018/01/05/fewer-americans-rely-on-tv-news-what-type-they-watch-varies-by-who-they-are/). We have added this information to the text (p. 12), as well as to the reference list.

As these examples illustrate, categorization in terms of local, national/network, and cable news is typical, with ideological variation within each category—perhaps particularly in the case of cable news (see, for example, https://www.journalism.org/2020/04/01/cable-tv-and-covid-19-how-americans-perceive-the-outbreak-and-view-media-coverage-differ-by-main-news-source/). Similarly, in our categorization, cable news encompasses more conservative (Fox News), more liberal (MSNBC), and more centrist (CNN) sources. Given this distribution, we would not say that the cable news category would necessarily be more conservative than those sources classified as national news.

We agree that relying on different news sources may vary by political affiliation, particularly for certain types of sources. We had considered including source by affiliation interaction terms in our models, but decided against this approach for two reasons. First, this was not an a priori research question (i.e., we had no clear rationale to believe that these factors, in combination, would be associated with varying perceptions of disagreement). Second, as Reviewer 1 notes, many independent variables are already included in the models, given our research question on correlates; thus probing interaction terms, especially without a priori justification, would likely be problematic from a collinearity and power perspective. Moreover, one might imagine that other interactions (e.g., with age or education) would be worth investigation; absent clear subgroup-related research questions or hypotheses at the outset, it would have been hard to know where to draw the line. Ultimately, we agree that future research should consider the interplay between specific sources and political affiliation (e.g., as has been conducted in the context of COVID-19 misinformation and right-leaning media; https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7251254/).

8) In the discussion, clarify the sentence that starts with “By comparison…” (on lines 392-395); and add citations to the statement that theory and research on the effects of exposure to conflicting health information raises cause for concern (on lines 439-440).

We have made these changes on p. 24 and p. 27.

Thank you again for the opportunity to revise and improve this manuscript. We are hopeful that this paper is now acceptable for publication in PLOS ONE.

Submitted filename: COVID conflict_PLOS ONE response to reviewers R1 .docx

Click here for additional data file.

29 Sep 2020

PONE-D-20-22073R1

Public perceptions of conflicting information surrounding COVID-19: Results from a nationally representative survey of U.S. adults

PLOS ONE

Dear Dr. Nagler,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

As noted by the reviewers' the authors did an excellent job responding to earlier concerns.  There are two minor suggestions that we would like to authors to add to the manuscript to further strengthen it.

Please submit your revised manuscript by Nov 13 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Sze Yan Liu, PhD

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #3: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #3: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #3: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #3: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors did an excellent job responding to reviewer comments. I look forward to seeing this published.

Reviewer #3: The authors were very responsive to the reviews and have improved the manuscript. I just have two minor additional comments related to issues that were addressed in the response to the reviewers but that I believe could be addressed more directly in the manuscript.

First, the authors state that some of the data not reported here are reported in a separate manuscript (Reviewer #3, Point 4). For transparency, I think the authors should cite the other paper (even if still in preparation or under review) and briefly describe the content so that reviewers/readers can be confident that the papers do not overlap.

Second, the authors describe the conceptual distinction between aspects and strategies related to COVID-19 in the letter (Reviewer #3, Point 6), but I did not see this described in the manuscript. I would recommend adding the justification from the letter to the manuscript, and perhaps including the correlations between the scales as well, in order to justify this decision to the reader.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

1 Oct 2020

RESPONSE TO REVIEWERS

MANUSCRIPT: Public perceptions of conflicting information surrounding COVID-19: Results from a nationally representative survey of U.S. adults (PONE-D-20-22073R1)

DATE: September 29, 2020

Thank you very much for the review of our manuscript, “Public perceptions of conflicting information surrounding COVID-19: Results from a nationally representative survey of U.S. adults.” Based on the reviewers’ helpful suggestions, we have made two additional improvements to the manuscript. Below we outline the concerns raised and our responses. In the revised manuscript, we indicate all revisions using track changes.

Reviewer #1

The authors did an excellent job responding to reviewer comments. I look forward to seeing this published.

Thank you very much; we appreciate your taking the time to review this work.

Reviewer #3

1) The authors were very responsive to the reviews and have improved the manuscript. I just have two minor additional comments related to issues that were addressed in the response to the reviewers but that I believe could be addressed more directly in the manuscript.

First, the authors state that some of the data not reported here are reported in a separate manuscript (Reviewer #3, Point 4). For transparency, I think the authors should cite the other paper (even if still in preparation or under review) and briefly describe the content so that reviewers/readers can be confident that the papers do not overlap.

We now cite this paper, which was recently accepted for publication, and the following text has been added to p. 8: “Data that describe levels of public awareness of disparities in COVID-19 mortality and correlates of that awareness are reported elsewhere.[36]”

2) Second, the authors describe the conceptual distinction between aspects and strategies related to COVID-19 in the letter (Reviewer #3, Point 6), but I did not see this described in the manuscript. I would recommend adding the justification from the letter to the manuscript, and perhaps including the correlations between the scales as well, in order to justify this decision to the reader.

We thank the reviewer for this suggestion. The following text has been added to pp. 10-11: “For both the health expert and politician question blocks, we kept the aspects and strategies indices separate because of the conceptual distinctions between them. All six items in the strategies index assess the effectiveness of strategies for preventing the spread of COVID-19; the four items in the aspects index, though more loosely connected to one another, are concerned with perceptions of risk, testing, and treatment. The indices are not so strongly correlated so as to suggest they are capturing the same phenomena (aspects and strategies indices, health experts: r = 0.55; politicians: r = 0.55).”

Thank you again for the opportunity to revise and improve this manuscript. We are hopeful that this paper is now acceptable for publication in PLOS ONE.

Submitted filename: COVID conflict_PLOS ONE response to reviewers R2.docx

Click here for additional data file.

5 Oct 2020

Public perceptions of conflicting information surrounding COVID-19: Results from a nationally representative survey of U.S. adults

PONE-D-20-22073R2

Dear Dr. Nagler,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Sze Yan Liu, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

13 Oct 2020

PONE-D-20-22073R2

Public perceptions of conflicting information surrounding COVID-19:Results from a nationally representative survey of U.S. adults

Dear Dr. Nagler:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Sze Yan Liu

Academic Editor

PLOS ONE