Skip to main content
NCBI home page
PLOS One logoLink to PLOS One
. 2020 Dec 18;15(12):e0243599. doi: 10.1371/journal.pone.0243599

Disparate impact pandemic framing decreases public concern for health consequences

Ugur Yildirim 1,*
Editor: Jim P Stimpson2
PMCID: PMC7748138  PMID: 33338035

Abstract

It is known that the new coronavirus (COVID-19) is disproportionately affecting the elderly, those with underlying medical conditions, and the poor. What is the effect of informing the public about these inequalities on people’s perceptions of threat and their sensitivity to the outbreak’s human toll? This study answers this question using a novel survey experiment and finds that emphasis on the unequal aspect of the pandemic, especially as it relates to the elderly and those with medical conditions, could be causing the public to become less concerned about the outbreak and its human toll. Discussion situates this finding in the literature on scientific communication and persuasion and explains why language that emphasizes the impact of the virus on all of us—rather than singling out certain groups—could be more effective in increasing caution among the general public and make them take the situation more seriously.

Introduction

Within a few months after its first emergence in Wuhan, China in December 2019, the novel coronavirus (COVID-19) has spread to almost every country on earth, including the US [1]. As of September 2020, the human toll of the disease worldwide is more than 30 million confirmed cases and nearly one million deaths [2]. Very few disease outbreaks in history have had such a fast and widespread impact on humanity, with the closest example being the 1918 flu pandemic [3].

Despite the global nature of the outbreak that has impacted peoples of all sexes, races, and cultural backgrounds, it is known that the disease is not affecting everyone in the same way. In particular, the elderly and those with underlying medical conditions are at higher risk of severe illness due to the virus [4]. Similarly, more infections and deaths are reported in poor and low-income communities compared to wealthier ones [5]. Neither of these patterns are surprising given what we know about health disparities [6–10] and the unequal impact of epidemics on certain groups [11, 12].

While the outbreak is far from having a uniform impact on different groups, the way the media and the scientific community is talking about the outbreak does not always touch upon this unequal aspect of the pandemic. Oftentimes, the account instead emphasizes the equalizing aspect of the pandemic, whereby the virus threatens all of us—all Americans or the entirety of humanity—regardless of our background [13]. Other times, the discussion revolves specifically around how the pandemic has been especially hard on certain groups, such as the elderly and the sick [14].

How do these different framings of the pandemic affect the public opinion? In particular, is one framing more or less effective than the other in terms of how it influences whether or not the public sees the outbreak as a serious threat or not and whether it is more important to save lives or to save the economy as the outbreak unfolds? This study answers this question using a novel survey experiment and finds that emphasis on the unequal aspect of the pandemic, especially as it relates to the elderly and those with medical conditions, could be causing the public to become less concerned about the outbreak and its human toll. Discussion situates this finding in the literature on scientific communication and persuasion and explains why language that emphasizes the impact of the virus on all of us—rather than singling out certain groups—could be more effective in increasing caution among the general public and make them take the situation more seriously.

Materials and methods

The project has IRB approval from University of California-Berkeley (protocol type: Soc-Behav-Ed Exempt; protocol number: 2020-04-13247; protocol title: Perceptions of inequality during the coronavirus outbreak). Written consent was obtained from respondents at the start of the survey.

Experimental design

The study is designed as a between-subjects survey experiment. It randomized each respondent into one of three conditions corresponding to three possible framings of the pandemic: (1) the “equal pandemic” framing, which does not say anything about the disparate impact of the pandemic on different groups but instead emphasizes how the outbreak has been affecting everyone regardless of their background; (2) the “elderly and medical conditions inequality” framing, which specifically emphasizes the unequal aspect of the pandemic in that it has been especially hard on the elderly and those with medical conditions; and (3) the “class inequality” framing, which specifically emphasizes the unequal aspect of the pandemic in that it has been especially hard on the poor and low-income communities. These conditions are chosen to reflect the ways that the pandemic is discussed in public discourse.

The experiment flows as follows. First, respondents are recruited into the study and asked to give their consent. (At this stage, respondents are told that the goal of the survey is to “understand the public’s opinions regarding important societal and economic trends in the US.” This general wording is chosen over using specific words such as coronavirus and inequality in an attempt to make sure respondents are not primed to think about these issues from the start.) Second, they are asked to watch a short clip with subtitles and told that the purpose of showing this video is to assess their comprehension skills; the content of the clips depends on the experimental condition respondents are in. Third, right after watching the video, they are asked to briefly describe the content of the video using their own words. Fourth, they answer a series of general questions related to their attitudes towards inequality as well as their socio-demographic characteristics such age, gender, race, and income.

Finally, respondents answer questions that are specifically related to the coronavirus outbreak. These questions include: (1) whether the respondent thinks the coronavirus is a serious threat to the American people or not; (2) whether the respondent thinks it is more important to save lives or to save the economy during this outbreak; how satisfied the respondent is with the way (3) their city, (4) their state, and (5) the federal government has been handling the coronavirus situation; (6) how the respondent has been affected by the coronavirus outbreak; and (7) how many times the respondent went outside in the past seven days.

Answers given to questions (1) and (2) constitute the main dependent variables in the study. Both variables take values between 1 and 5 with higher values denoting higher threat perceptions in the case of the first variable (1 = not a threat at all, 2 = a small threat, 3 = a threat, 4 = a serious threat, 5 = a very serious threat) and attaching more importance to saving the economy over saving lives in the case of the second variable (1 = saving lives must be the priority even if it means the economy will suffer, 2, 3, 4, 5 = saving the economy must be the priority even if it means lives will be lost). Answers given to questions (3), (4), and (5) are similarly coded to take values between 1 and 5 with higher values denoting more satisfaction.

Multiple binary variables have been generated based on question (6), including whether the respondent or someone in the respondent’s family (i) is at risk, (ii) has contracted the virus, (iii) lost their job due to the outbreak, or (iv) experienced a significant decrease in income due to the outbreak. The “at risk” variable is particularly important here because given that the current crisis is caused by a disease outbreak, those who are at risk of severe illness and death will likely view and respond to the crisis very differently compared to those who are not at risk. The variable based on Question (7) takes values between 0 and 7. (See S1 Appendix for the experimental texts, images, videos, manipulation check question, survey questions, and other related project content including additional variables and conditions. The study design is pre-registered, while the specific hypotheses tested in this paper are not.)

Implementation and subject recruitment

The survey experiment is implemented using Qualtrics. The videos presented to respondents as part of the experiment are prepared using iMovie and subsequently uploaded to a YouTube channel created by the researcher (videos are “unlisted”, have comments disabled, and show subtitles by default). All videos showed an Adobe Stock licensed image in the background related to the content of the narrated text. The experimental texts themselves are written by the researcher after a careful reading of relevant news articles and scientific communications.

The texts narrated to respondents in the videos are recorded by a young female in her 20’s speaking Standard American English. Female voice is chosen over male voice due to evidence that shows that people tend to find the female voice to be more credible [15]. The narrated text is also displayed as actual text under the video in case the respondent experiences a problem watching the video or chooses not to watch. (As discussed later under Results, the researcher confirmed that most respondents watched and understood the videos.)

Data collection took place on Lucid Theorem. This platform gives researchers access to cheap, fast (thousands of responses within hours), and high quality data that is also nationally representative based on age, gender, ethnicity, and region. A recent scholarly work also validated the quality of Lucid samples [16]. (All code, materials, and de-identified data will be made public once the study is over.)

Sample characteristics and data structure

The survey experiment is run on a total of 2,617 respondents with approximately 870 respondents in each condition. The three conditions appear to be balanced on the demographic covariates, which gives us confidence that randomization worked as expected. All analyses are conducted on a dataset with the following simple structure: one row per respondent and as many columns as there are variables. Respondents are required to be US residents and 18 or older. (See S2 Appendix for information on sample size calculations, exact sample sizes by condition, and summary demographics by condition.)

Overview of statistical models used

Linear regression models are fit to data with the experimental condition as the independent variable. The equal pandemic condition is used as the reference category to be able to get estimates for the elderly and medical conditions inequality and class inequality conditions. (Note that the choice of reference category is somewhat arbitrary as it can be reasonably argued that equal pandemic is actually the distinct frame here. Accordingly, additional models were fit to data—see S3 Appendix—that treat the inequality conditions as the reference category to estimate an equal pandemic effect. These additional models do not change our substantive conclusions at all but allow us to see the story from the opposite angle.)

Since the inclusion of socio-demographic covariates does not change our conclusions—this is not surprising as the independent variable is randomly assigned to respondents—the main text only discusses models without these covariates. (S3 Appendix presents results both with and without socio-demographic covariates for the sake of transparency in line with recent scholarly work [17]. Models with additional outcomes as well as results based on ordinal logistic regression models—which do not change the substantive conclusions discussed in the text—are also presented.)

Results

Manipulation checks

Manipulation checks are used in experimental research to determine whether the subjects actually received the treatments the researcher intended them to receive. The researcher confirmed that most respondents actually watched the videos by checking the number of YouTube “views” of each video. Most respondents also passed the manipulation check question, that is, clearly understood the text being communicated to them. (The researcher used a custom script to look for certain keywords such as “coronavirus” or “elderly” to make sure that respondents’ description of the video was correct.) Furthermore, conclusions presented here remain unchanged regardless of whether or not we restrict the sample to only those respondents who passed the manipulation check.

Main findings

The experiment had a significant impact on respondents’ opinions regarding whether coronavirus is a serious threat or not and whether the priority should be saving lives or saving the economy. As far as opinions regarding whether coronavirus is a serious threat or not are concerned, respondents who saw the elderly and medical conditions inequality condition (which emphasizes how the pandemic has been especially hard on the elderly and those with medical conditions) reported significantly lower levels of threat perception compared to respondents who saw the equal pandemic condition (coefficient estimate = -0.166, p-value = 0.001, see left panel of Fig 1). Regarding opinions as to whether the priority should be saving lives or saving the economy, respondents who saw the elderly and medical conditions inequality condition reported significantly more support towards saving the economy over saving lives compared to equal pandemic (coefficient estimate = 0.201, p-value = 0.001, see right panel of Fig 1).

Fig 1. The effect of the informational treatment on outcomes.

Fig 1

Open in a new tab

The point estimates are predicted means. The bars denote 95% confidence intervals. N = 2,617.

Digging deeper into these patterns revealed an interesting treatment-effect heterogeneity. Both of the effects discussed in the previous paragraph are mainly driven by respondents who are neither at risk themselves nor have family members who are at risk. Significant treatment effects are observed only in this not-at-risk group, while the treatment effect decreases in magnitude by more than half and loses statistical significance among respondents who are at risk or have at risk family members (see Fig 2). Effect heterogeneity is demonstrated by fitting separate models for at-risk and not-at-risk sub-groups (i.e., fitting two separate regressions of the outcome on the experimental conditions, one on the at-risk sample and the other on the not-at-risk sample). The elderly and medical conditions inequality coefficient estimate for the outcome “coronavirus serious threat” is -0.061 (p-value = 0.454) for respondents at risk and -0.184 (p-value = 0.004) for respondents not at risk. Similarly, the elderly and medical conditions inequality coefficient estimate for the outcome “economy must be saved” is 0.101 (p-value = 0.348) for respondents at risk and 0.226 (p-value = 0.003) for respondents not at risk.

Fig 2. Effect heterogeneity based on being at risk.

Fig 2

Open in a new tab

The point estimates are predicted means. The bars denote 95% confidence intervals. N = 2,617.

In addition to the procedure described here to investigate effect heterogeneity, the researcher fitted additional, pooled models that explicitly modeled the outcome as a function of the experimental conditions, the at-risk variable, and interactions between the two. The interactions from these models are insignificant, which is not surprising because the experiment was not powered to be able to detect interaction effects. That said, the at-risk main effects are significant and both the at-risk main effects and interactions are in the expected direction (i.e., opposite of the treatment effects), which explains why the treatment effects are drastically smaller—two to three times—in the at-risk sub-sample. (See S3 Appendix for results based on the models with interactions.)

While the elderly and medical conditions inequality condition led to significant changes in both outcomes, the class inequality condition was weaker in its effects. Despite the effect being in the same direction as elderly and medical conditions inequality, class inequality led to significant changes only in the “economy must be saved” outcome. The class inequality coefficient estimates are -0.067 (p-value = 0.199) for the “coronavirus serious threat” outcome, which is less than half the magnitude of the elderly and medical conditions inequality effect, and 0.138 (p-value = 0.027) for the “economy must be saved” outcome, which is about only two-thirds of the elderly and medical conditions inequality effect. See Table 1 for a compact presentation of the estimated coefficients associated with the experimental conditions for both outcomes. (The statistically significant class inequality effect disappears when we control for the socio-demographic covariates.) On the other hand, data show that the class inequality condition had a nearly significant negative effect of -0.111 (p-value = 0.058) on satisfaction with state’s handling of the coronavirus situation; no significant effects are observed for elderly and medical conditions inequality or for the other two satisfaction outcomes (city and federal government).

Table 1. Treatment effect estimates.

Coronavirus serious threat Economy must be saved
Elderly and medical conditions inequality -0.166 (0.052)** 0.201 (0.062)**
Class inequality -0.067 (0.052) 0.138 (0.062)*
Open in a new tab

The numbers inside the parentheses are standard errors. Estimates are based on models without any demographic covariates. Stars denote p-values: ˙p<0.1

* p<0.05

** p<0.01

*** p<0.001.

Discussion

The information the public receives regarding the coronavirus outbreak influences their threat perceptions and whether they think saving the economy or saving lives should be the priority. Results from this study show that being informed about the disproportionate negative impact of the pandemic on the elderly and those with underlying medical conditions make people less likely to see coronavirus as a threat and more likely to prioritize saving the economy as opposed to saving lives, particularly among those who do not need to worry about themselves or someone in their family being at risk of severe illness.

These findings suggest that the dissemination of scientific information regarding the unequal impact of the pandemic on certain groups could actually be causing the general public to become less concerned about the outbreak and its human toll. The fact that the effect is primarily observed among people not at risk further indicate that when those people are sensitized to the situation of the weak they feel more secure about their own situation as not being at risk, which likely leads to increased optimism bias [18] and underestimation of their risk of infection [19]. These results give more support to mechanisms of deliberation and callousness as opposed to sympathy [20–22].

While information regarding the disproportionate negative impact of the pandemic on the elderly and those with underlying medical conditions had a significant impact on coronavirus threat perceptions and preferences regarding whether saving lives or saving the economy should be the priority, information regarding the disproportionate negative impact of the pandemic on the poor did not have as big of an impact on the outcomes and generally failed to achieve statistical significance. One possible explanation for this null effect is that issues around class are highly politicized in the US, and so it is more difficult to move people’s opinions on these topics compared to a more neutral and directly health-related topic such as the elderly and those with medical conditions.

The findings also have important policy implications. If the policy goal is to increase caution among the general public and make them take the situation more seriously, then information that emphasizes solidarity—“we are all in this together”—is likely to be much more effective, especially when it comes from a credible source [23, 24]. This solidarity framework should be employed even when informing the public about the unequal impact of the pandemic on certain groups, so that the general public is not left with the impression that the outbreak concerns only some—not all—of us.

Limitations

One of the limitations of the study is that the ‘elderly and medical conditions inequality’ and ‘class inequality’ conditions are completely separate from one another by design. This is justified because the study is primarily concerned with how people understand the impact of the pandemic, not about the actual facts. That said, it is certainly the case that the poor are more likely to have medical conditions as a matter of science, and the current study does not look at this issue that concerns how ‘elderly and medical conditions inequality’ and ‘class inequality’ angles intersect. Another, related limitation is that the experimental text used for the ‘class inequality’ condition mentions minorities when discussing the impact of the pandemic on the poor and low-income communities, which means that the condition refers to not only class but also race disadvantage. Once again, while this choice is justified by virtue of the fact that the framing is in line with the usual way the topic is discussed in public discourse—see, e.g., the recent United States Joint Economic Committee report on coronavirus [25]—the literature on group cues [26] tells us that whether the information is interpreted primarily in terms of class or race will likely influence the way respondents answer survey questions. Therefore, investigating how class and race axes intersect would be a fruitful area of future work. A final limitation is that the custom script used for parsing the manipulation check question is developed by the researcher alone and was not independently verified prior to data collection.

Supporting information

S1 Appendix. Experimental texts, images, videos, and other related content.

(PDF)

Click here for additional data file. (1,011.3KB, pdf)
S2 Appendix. Sample size calculations and sample characteristics.

(PDF)

Click here for additional data file. (181.6KB, pdf)
S3 Appendix. Regression results.

(PDF)

Click here for additional data file. (140.6KB, pdf)
S1 File

(ZIP)

Click here for additional data file. (18.9KB, zip)
S1 Data

(ZIP)

Click here for additional data file. (366.5KB, zip)

Acknowledgments

The author thanks David Harding, Dennis Feehan, Daniel Schneider, Gabriel Lenz, Don Moore, and Xinyi Zhang for all of their helpful comments and Anam Ahmed for kindly taking the time to record the experimental texts used in the study.

Data Availability

All relevant data are within the manuscript and its Supporting Information files.

Funding Statement

UY received a small grant from the Experimental Social Science Laboratory (XLAB) at University of California Berkeley (https://xlab.berkeley.edu/) to run this study. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

  • 1.Van Bavel J. et al. 2020. “Using social and behavioural science to support COVID-19 pandemic response.” Nature Human Behavior 4: 460–471. [DOI] [PubMed] [Google Scholar]
  • 2.“COVID-19 Dashboard by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU).” ArcGIS. Johns Hopkins University. Retrieved September 18, 2020.
  • 3.Scott S. and Duncan C. 2001. Biology of Plagues: Evidence from Historical Populations. Cambridge University Press. [Google Scholar]
  • 4.Zhou F. et al. 2020. “Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study.” The Lancet 395(10229): 1054–1062. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Von Braun J., Zamagni S., and Sorondo M. 2020. “The moment to see the poor.” Science 368(6488): 214 10.1126/science.abc2255 [DOI] [PubMed] [Google Scholar]
  • 6.Murray C., Kulkarni S., and Ezzati M. 2005. “Eight Americas: New Perspectives on U.S. Health Disparities.” American Journal of Preventive Medicine 29(5S1): 4–10. [DOI] [PubMed] [Google Scholar]
  • 7.Adler N. and Rehkopf D. 2008. “U.S. Disparities in Health: Descriptions, Causes, and Mechanisms.” Annual Review of Public Health 29: 235–252. 10.1146/annurev.publhealth.29.020907.090852 [DOI] [PubMed] [Google Scholar]
  • 8.Braveman P., Cubbin C., Egerter S., Williams D., and Pamuk E. 2010. “Socioeconomic Disparities in Health in the United States: What the Patterns Tell Us.” American Journal of Public Health 100(S1): 1375–1378. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Marmot M. 2015. “The health gap: the challenge of an unequal world.” The Lancet 386(10011): 2442–2444. [DOI] [PubMed] [Google Scholar]
  • 10.Boen C., Keister L., and Aronson B. 2020. “Beyond Net Worth: Racial Differences in Wealth Portfolios and Black-White Health Inequality across the Life Course.” Journal of Health and Social Behavior 61(1): 1–17. 10.1177/0022146520906104 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Luk J., Gross P., and Thompson W. 2001. “Observations on Mortality during the 1918 Influenza Pandemic.” Clinical Infectious Diseases 33(8): 1375–1378. 10.1086/322662 [DOI] [PubMed] [Google Scholar]
  • 12.Quinn S. and Kumar S. 2014. “Health Inequalities and Infectious Disease Epidemics: A Challenge for Global Health Security.” Biosecurity and Bioterrorism: Biodefense Strategy, Practice, and Science 12(5): 263–273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.McNeil D. 2020. “The Coronavirus in America: The Year Ahead.” The New York Times. Accessed May 18, 2020. [Google Scholar]
  • 14.“People Who Are at Higher Risk for Severe Illness.” Centers for Disease Control and Prevention. Accessed May 18, 2020. [Google Scholar]
  • 15.Siegel M., Breazeal C., and Norton M. 2009. “Persuasive Robotics: The influence of robot gender on human behavior.” IEEE/RSJ International Conference on Intelligent Robots and Systems October 11–15: 2563–2568. [Google Scholar]
  • 16.Coppock A. and McClellan O. 2019. “Validating the demographic, political, psychological, and experimental results obtained from a new source of online survey respondents.” Research and Politics January-March: 1–14. [Google Scholar]
  • 17.Lenz G. and Sahn A. 2020. “Achieving Statistical Significance with Control Variables and without Transparency.” Working Paper. [Google Scholar]
  • 18.Sharot T. 2011. “The optimism bias.” Current Biology 21(23): R941–R945. 10.1016/j.cub.2011.10.030 [DOI] [PubMed] [Google Scholar]
  • 19.Wise T., Zbozinek T., Michelini G., Hagan C., and Mobbs D. “Changes in risk perception and protective behavior during the first week of the COVID-19 pandemic in the United States.” Preprint at PsyArXiv. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Loewenstein G. and Small D. 2007. “The Scarecrow and the Tin Man: The Vicissitudes of Human Sympathy and Caring.” Review of General Psychology 11(2): 112–126. [Google Scholar]
  • 21.Small D., Loewenstein G., and Slovic P. 2007. “Sympathy and callousness: The impact of deliberative thought on donations to identifiable and statistical victims.” Organizational Behavior and Human Decision Processes 102(2): 143–153. [Google Scholar]
  • 22.Martin M. 2001. “Responsibility for Health and Blaming Victims.” Journal of Medical Humanities 22: 95–114. [Google Scholar]
  • 23.Haslam S., Reicher S., and Platow M. 2011. The New Psychology of Leadership: Identity, Influence, and Power. Routledge. [Google Scholar]
  • 24.Brinol P. and Petty R. 2009. “Source factors in persuasion: a self-validation approach.” Eur. Rev. Soc. Psychol. 20: 49–96. [Google Scholar]
  • 25.Beyer Don. “The Impact of Coronavirus on the Working Poor and People of Color.” United States Joint Economic Committee. Retrieved September 1, 2020. [Google Scholar]
  • 26.Nelson T. and Kinder D. 1996. “Issue Frames and Group-Centrism in American Public Opinion.” The Journal of Politics 58(4): 1055–1078. [Google Scholar]

Decision Letter 0

Jim P Stimpson

13 Aug 2020

PONE-D-20-17174

Disparate Impact Pandemic Framing Decreases Public Concern For Health Consequences

PLOS ONE

Dear Dr. Yildirim,

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 27 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,

Jim P Stimpson, PhD

Academic Editor

PLOS ONE

Additional Editor Comments:

The reviewers carefully reviewed this paper and made several excellent suggestions for improvement. The author should respond to the analytical issues raised by reviewers including the choice of linear regression and the choice to have an interaction effect with treatment group. How were class inequality or race included in the inequality frame and randomization? If not included, why not and provide a discussion and include limitations? If so, make this much clearer and include results and discussion. Related, please clarify the frames, particularly “natural inequality.” There should be a more robust description of methodological limitations. Finally, improve the presentation of results so that the paper is easier to read (e.g. first outcome and second outcome language) and the tables should stand on their own without reference to the text.

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 include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[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: Partly

Reviewer #3: Yes

Reviewer #4: Yes

**********

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

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

Reviewer #4: 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

Reviewer #4: 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

Reviewer #4: 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: There are parts where there is some confusion:

- For instance, when I was reading I thought it would be good to remind the reader of what "natural inequality" means when discussing the findings (term mentioned on page 4 but not again until the end of page 7).

- Page 8 “main findings”—first paragraph is how the variables were labeled/defined, why is this in the findings section and not in the methods section?

- Also on page 8—I would mention what “manipulation check” is for readers who may not know exactly what this means.

- Page 9, second paragraph, I think it is informal when using “the first outcome”—I suggest that the outcome be defined and explicitly stated and interpreted to make sure the reader does not have to scroll back up to another section of the paper to know what that outcome is and what it means—this makes the paragraphs in the main findings section hard to follow--I would suggest being much more explicit throughout.

- Where is the rational for why subjects were randomized the way that they were (end of page 4)? Why would elderly and those with medical conditions be one condition? We know that low income and poor populations are more likely to have medical conditions (diabetes, cardiovascular disease, obesity, COPD) that arises from systemic oppression and racism, so why separate poor and low income from medical conditions when they are related as much as being elderly can be related with medical conditions? I would definitely add text on how randomization occurred—decision process and theory of why/how these groups were created. Also, if there is a way to perform some sensitivity analyses which could show us what the results would look like if one group was just elderly, and the other group was low-income and with medical conditions, I would highly suggest including this—at the very least, provide some discussion on this in the manuscript.

Reviewer #2: In results you provide a robust description of the data on the use of the class inequality and also the use of the natural inequality. In the discussion there is less or no explanation of the data in results on the class inequality. Please discuss the class inequality data in discussion in some way.

Please explain how the custom script was developed and whether it was validated in some way by having another person read it or whether it was pre-tested on a sample before its use.

There is not section on limitations. The development of the custom script might be a limitation.

Also can you go to greater lengths in the discussion to clarify second and first outcomes as described in this section below- just make it clearer rather then simply saying the first outcome or the second outcome. I had to keep checking back to remind myself of these outcomes.

While the natural inequality condition led to significant changes in both outcomes, class

inequality condition was weaker in its effects. Despite the effect being in the same direction as

natural inequality, class inequality led to significant changes only in the second outcome (second outcome was what again?)

so were people who had the class inequity frame more or less likely to feel threatened by the virus and how did they feel about saving the economy?

Reviewer #3: This study broadly assesses the impact of message framing on the public's perceptions of threat of COVID-19, and their sensitivity to the outbreak’s human toll. The study obtains data from a large nationally representative sample accessed via a private marketing firm. The study is very interesting, timely and has policy-relevant implications.

My comments are discussed below:

Methods

The authors noted the use of linear regression modeling. However, their outcome variables are measured on a Likert Scale and as such are strictly ordinal variables. Did they test for violations of OLS assumptions? Perhaps they can consider ordinal logistic regression (especially if the assumptions of OLS are violated).

Discussion

One of the more interesting findings of this study is that a focus on the disproportionate negative impact of the virus on the elderly (and NOT on poor and low-income communities) minimizes people's perception of the threat of the virus. Why is this so? While the authors discuss the "elderly finding", the finding (or the lack of finding) for the poor and low-income arm is not discussed at all. This is an important omission that needs to be addressed.

Reviewer #4: This is a clearly written manuscript presenting a clear study asking how framing the COVID-19 pandemic in terms of disparate groups affected influences public response. While there is much to admire about this simple yet important question and design, there are a few places the paper can be improved.

The author should be much clearer about whether the treatment group is manipulating class inequality or race in the inequality frame. I pulled the supplementary materials to find the text of this manipulation and it is definitely a conflation of class and race (“poor and low-income communities, particularly minorities such as blacks and Hispanics, have been disproportionately affected”). The paper starts by discussing this as just class inequality, but occasionally also discusses race. There is a very broad literature in social science on group-cues (see e.g., Nelson and Kinder 1996) that is not discussed at all. This is curious, as sociology, political science, and communication all have spent much attention on looking at how framing information in terms of the groups affected – particularly as it relates to class and race – matters. More attention to framing and race/class would be warranted in this paper.

Neither the background nor the methods section motivates or pre-specifies the treatment effect heterogeneity findings – among those not at risk and with no family at risk. Since the authors present these results, they should explain why they sought to look at the data by these strata earlier in the manuscript.

Did the authors pre-register their hypotheses and study design?

You report separate models stratified by risk as noted above, but it would also be statistically sound to estimate an interaction term by at-risk vs. not-at risk and the treatment groups. The way the results by subgroup are reported on p. 9 is confusing -- it sounds like the authors are reporting the coefficients on the risk variable and not the treatment groups? I recommend revising the discussion of the treatment effect heterogeneity section so it is more clear and also reports the results of the models estimated with interaction terms.

Authors describe the “equal pandemic” frame as the control, but I would argue that most news coverage emphasizes groups at risk rather than the idea of “equal pandemic”. By using the “equal pandemic” as the omitted/reference category they assume in their interpretation that the inequality treatments are what are driving the results – but given there is no “no media exposure” control group, it could be that the “equal pandemic” is actually the distinct frame. Authors might want to comment on this in the discussion. Looking at the figures, for instance, I would actually not use the term “control” but instead call this condition “Equal pandemic”.

Reference:

Nelson, T. E., & Kinder, D. R. (1996). Issue frames and group-centrism in American public opinion. The Journal of Politics, 58(4), 1055-1078.

**********

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: Yes: Lauri Andress, Ph.D.

Reviewer #3: No

Reviewer #4: 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.

PLoS One. 2020 Dec 18;15(12):e0243599. doi: 10.1371/journal.pone.0243599.r002

Author response to Decision Letter 0

22 Sep 2020

We thank the editor and the reviewers for their constructive and helpful comments. We carefully reviewed the points raised by all four reviewers and revised the manuscript accordingly. Our responses to specific points raised are provided in the attached Response to Reviewers document.

Attachment

Submitted filename: PLOS-ONE-Response-to-Reviewers.docx

Click here for additional data file. (15.3KB, docx)

Decision Letter 1

Jim P Stimpson

13 Oct 2020

PONE-D-20-17174R1

Disparate impact pandemic framing decreases public concern for health consequences

PLOS ONE

Dear Dr. Yildirim,

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 provide a point-by-point response to the Editor Comments below. You may, if you wish, address other reviewer comments in your revision.

Please submit your revised manuscript by Nov 27 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,

Jim P Stimpson, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (if provided):

Please change “natural inequality” to a more descriptive term representing that construct such as “elderly and medical conditions inequality” in the text and tables/figures.

There is still confusion about whether racial/ethnic minorities are a focal group or not. It’s clear the data exist to analyze this population from the survey instrument but the analysis seems to combine this population with low income persons. If racial/ethnic minorities are not a focus and not genuinely analyzed, then it might be better to drop this reference in the text and tables/figures and instead list this as a limitation and an area of future work.

The paper would benefit from revising the language in the first paragraph of the Experimental Design, especially clearly defining the three groups. If the “equal pandemic” is indeed the control in the experiment, then a more detailed explanation of that group is needed.

Please add a table 2 x 2 perhaps with the treatment on one side of the table and outcomes across the top as suggested by reviewer 2.

[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 #2: (No Response)

Reviewer #4: (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 #2: Yes

Reviewer #4: No

**********

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

Reviewer #2: Yes

Reviewer #4: 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 #2: Yes

Reviewer #4: Yes

**********

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 #2: Yes

Reviewer #4: 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 #2: Please add a table 2 x 2 perhaps with the treatment on one side of the table and outcomes across the top

Reviewer #4: The abstract still only mentions disparities on the elderly, those with underlying medical conditions, and the poor and does not mention the disproportionate impact on people of color. This is a part of the study and should be in the abstract.

The literature review still lacks any reference to work framing inequality and/or the effects of group frames / cues on public opinion. The study does not seems situated in to the social scientific context around public understanding of social inequality.

Can you please explain how written consent would have been obtained in an online survey? Did they have to sign something? Most online surveys do not use written consent, in my experience.

What information was provided to respondents about how to answer whether "the respondent or someone in the respondent's family is at risk"? Is this the respondent's own judgment, or based on some kind of medical criteria? While I understand that people's response to the experimental message emphasizing risk to the elderly and those with medical conditions or poor/ POC may be more or less salient based on the respondent's own risk status, it's hard for me to evaluate the value of this stratification until I understand what participants' were referring to when evaluating their own risk. How many people are in each group (at risk vs. not at risk?) I couldn't find this important piece of information. Is the at-risk group simply much smaller than the not-at-risk group (presuming Lucid samples might not include many elderly people who may lack Internet access)? This would explain why results are stronger in the not-at-risk group?

I don't understand why a study design would be pre-registered but not these specific hypotheses - my understanding is that the point of pre-registering is to ensure that the hypotheses tested were a priori and not developed based on what the data actually show.

Language in response to my concern about reference groups (the fact that there is no real control group in this study) is informal - "see the story from the opposite angle". Please clarify for readers.

Please clarify what is meant by "custom script" in the discussion of manipulation check. This is in the response to another reviewer but is not in the manuscript text.

Could the reason that the elderly/medical condition frame reduces threat perception simply because this frame gave people more knowledge about who is at risk? Then, the effect wouldn't be really one about groups vs. universalizing, but about providing specific risk information. This would challenge the premise of the study.

In response to my comment about the fact that the equal pandemic condition is not really a control, the author indicated they no longer refer to the condition as control. However, on p. 11 they say "lower levels of threat perception compared to respondents who saw the control" (twice)

Overall, I'm just not sure that the data support the author's conclusion that dissemination about the unequal impact of the pandemic leads to these outcomes -- I'm not convinced the treatment is about inequality at all, but about discrete risk information. Yes, the two are aligned, but the paper is framed as being about inequality in COVID and not as a risk-information treatment. The class/race treatment not having much of an effect contributes to this conclusion as well. I am not sure I understand the argument as why the class/race manipulation didn't move opinion was because it was politicized. I am not in favor of the stratified analysis by risk, but it seems that a parallel exploratory analysis would divide by either class or race to assess whether there are heterogeneous effects for the other treatment by these salient characteristics. Either way, the lack of pre-specification of these hypotheses -- and scarce contextualization of this study into a body of literature or theory -- leaves me concerned about what this study is trying to accomplish and whether the analysis has achieved it.

**********

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 #2: Yes: Lauri A Andress

Reviewer #4: 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.

PLoS One. 2020 Dec 18;15(12):e0243599. doi: 10.1371/journal.pone.0243599.r004

Author response to Decision Letter 1

21 Nov 2020

I thank the editor and the reviewers for their constructive and helpful comments. As suggested by the editor, I carefully reviewed the points raised in the Editor Comments section and revised the manuscript accordingly. Please see the attached response to reviewers document for my responses.

Attachment

Submitted filename: PLOS-ONE-Response-to-Reviewers-Round-2.docx

Click here for additional data file. (7.4KB, docx)

Decision Letter 2

Jim P Stimpson

25 Nov 2020

Disparate impact pandemic framing decreases public concern for health consequences

PONE-D-20-17174R2

Dear Dr. Yildirim,

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,

Jim P Stimpson, PhD

Academic Editor

PLOS ONE

Acceptance letter

Jim P Stimpson

3 Dec 2020

PONE-D-20-17174R2

Disparate impact pandemic framing decreases public concern for health consequences

Dear Dr. Yildirim:

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. Jim P Stimpson

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 Appendix. Experimental texts, images, videos, and other related content.

    (PDF)

    Click here for additional data file. (1,011.3KB, pdf)
    S2 Appendix. Sample size calculations and sample characteristics.

    (PDF)

    Click here for additional data file. (181.6KB, pdf)
    S3 Appendix. Regression results.

    (PDF)

    Click here for additional data file. (140.6KB, pdf)
    S1 File

    (ZIP)

    Click here for additional data file. (18.9KB, zip)
    S1 Data

    (ZIP)

    Click here for additional data file. (366.5KB, zip)
    Attachment

    Submitted filename: PLOS-ONE-Response-to-Reviewers.docx

    Click here for additional data file. (15.3KB, docx)
    Attachment

    Submitted filename: PLOS-ONE-Response-to-Reviewers-Round-2.docx

    Click here for additional data file. (7.4KB, docx)

    Data Availability Statement

    All relevant data are within the manuscript and its Supporting Information files.

    Articles from PLoS ONE are provided here courtesy of PLOS

    RESOURCES