Identifying And Exploring Bias In Public Opinion On Scarce Resource Allocation During The COVID-19 Pandemic

Identifying And Exploring Bias In Public Opinion On Scarce Resource Allocation During The COVID-19 Pandemic


The Americans with Disabilities Act (ADA) of 1990 sought to alleviate discrimination against people with disabilities by codifying disability-based protections into civil rights law. Coupled with the ADA Amendments Act of 2008, which clarified that the definition of disability should be interpreted broadly (encompassing well-controlled chronic conditions, such as diabetes, alongside more obvious disabilities), the ADA is an important tool for addressing discrimination against people with disabilities in health care.

In spite of these protections, people with disabilities experience more denial of care, more negative treatment by health care providers, and worse health outcomes compared to people without disabilities.13 Recent work suggests that these disparities result, in part, from clinicians’ beliefs that people with disabilities have lower quality of life than nondisabled people.2,3 Members of the general public may hold similar beliefs, and these attitudes might contribute to public policies that discriminate against people with disabilities. Yet existing scholarship has paid little attention to understanding the nature of disability bias among the general public.

This gap in the research limits understanding of the role of disability bias in public policy. Before distributing resources to specific recipients, policy makers engage in what scholars call “anticipatory feedback,” attempting to gauge the popularity and political feasibility of policies according to the public’s assessment of each group’s deservingness and their willingness to sustain sacrifices to benefit (or punish) members of target populations.4 Public perceptions also influence policy making more directly by shaping the views, agendas, and behaviors of the bureaucrats who implement policy.5,6

Previous research found that the public prioritizes people with disabilities when it comes to the allocation of cash assistance.7,8 Less is known, however, about contexts involving the allocation of scarce medical resources. In contrast to the redistribution of taxpayer funds in the form of cash assistance,9 one might expect that in a pandemic, many will understand themselves as being in direct competition with others for resources should they become sick. COVID-19 might therefore trigger feelings of anxiety, discomfort,10,11 and even anger toward people with disabilities—all emotions that are likely to be stimulated when people feel that they are competing for scarce resources or when they believe that others are being given special privileges.1012

Because of the large numbers of patients in the US who had COVID-19 during spikes, hospitals have faced acute shortages of ventilators, beds, medications, and other critical care resources, leading to the creation or activation of policies regarding crisis standards of care.13 Such policies are intended to provide guidance to health care providers about how to allocate scarce resources during a public health emergency. These policies have often incorporated potentially discriminatory provisions ranging from categorically excluding people with specific disabilities from care to applying prioritization criteria (such as expectations for long-term survival or lower resource use) that disadvantage people with disabilities by giving them lower priority.1316

Concerns about discrimination in the implementation of crisis standards of care have prompted state and federal oversight activities by both legislators and regulators during the pandemic, including guidance from the Department of Health and Human Services (HHS) Office for Civil Rights on the application of disability civil rights law to implementing crisis standards of care.1416 In response, crisis standards of care plans have evolved considerably to remove provisions that discriminate on the basis of disability.13

Regulators seeking evidence to inform civil rights enforcement in health care can benefit from empirical data on the nature of prejudice against specific populations. Public opinion research has examined preferences regarding crisis standards of care policy making, but these studies have typically used methodologies that are susceptible to social desirability bias;17 surveyed non-US populations; or inquired about broad allocative principles, such as whether patients should be deprioritized because of long-term life-expectancy, without tying them to specific disabilities.1820 This lack of specificity in studying public opinion is a major limitation of previous work, as the social construction of a particular disability may trigger different responses than broad ethical principles would.

To address these limitations and to examine the extent to which disability bias has emerged in the allocation of scarce resources during the COVID-19 pandemic, we conducted a conjoint experiment on a nationally representative sample of the US public. Conjoint experiments were initially developed by business scholars to uncover specific attributes driving product preferences.21 Political scientists have applied this methodology as a way to understand bias toward groups.2224 In our study, survey respondents made a series of choices between two personal profiles that randomly varied on multiple dimensions of interest. This method mitigates social desirability bias by allowing respondents to avoid acknowledging any specific characteristic as the reason for choosing a hypothetical individual for the allocation of benefits or burdens.2224 Respondents thereby reveal hidden preferences that they might not otherwise acknowledge. Findings from our conjoint experiment shed light on how various patient attributes affect the public’s beliefs regarding who should receive access to scarce medical resources during the COVID-19 pandemic.

Study Data And Methods

The Institutional Review Board of Miami University (Oxford, Ohio) approved this study.

Conjoint Study Design

We presented respondents with a choice scenario in which they would have to select which one of two patients with an equal likelihood of short-term survival would get a hospital’s last available ventilator (for a sample conjoint choice task, see online appendix exhibit 1).25 The patients in the scenario varied randomly along several characteristics, including gender (cisgender man, cisgender woman, transgender man, transgender woman), race (White, Black, Asian), age (chosen from the following ranges: 21–31, 41–51, 65–75), employment status before the COVID-19 outbreak (employed, unemployed), whether the patient followed Centers for Disease Control and Prevention (CDC) social distancing guidelines26 (followed, did not follow), and disability status.

If the patient had a disability, respondents were told the name of the disability and received a brief description, including information about its impact on life expectancy (lower long-term life expectancy, can have a normal life span) and typical onset (present from birth, can get it at any time, manifests in young adulthood) (appendix exhibit 2).25 We tested six disabilities (type 2 diabetes, congenital heart defect, paraplegia, intellectual disability, traumatic brain injury, bipolar disorder) that were selected because they vary in ways that allowed us to explore whether respondents’ choices varied between patients with physical or mental disabilities, those with lower and normal long-term life expectancy, and those with different times of disability onset. Although some previous public opinion research about ventilator allocation has asked respondents about broad allocative principles,1820 we believe that inquiring about specific disabilities is more likely to elicit a response consistent with real-world responses to specific allocative dilemmas.


Using the polling firm YouGov, we surveyed respondents from January 29 to February 4, 2021—a period during which COVID-19 case counts remained high and vaccine access was limited. By recruiting respondents through multiple methods (mainly targeted online advertising), YouGov maintains a panel of respondents who have agreed to take its surveys. The firm uses email solicitation to invite participants to complete surveys electronically.

YouGov ensured that respondents were representative of the total US adult population by matching its panelists to a target sample from the Census Bureau’s 2018 American Community Survey. After matching, YouGov used propensity score weighting and poststratification to correct for any remaining differences between the matched survey sample and the target sample (see notes in appendix exhibit 3 for details).25

Our final sample consisted of 2,500 respondents, using the weights provided by YouGov (see appendix exhibit 3).25 Each respondent was presented with four iterations of the choice scenario, resulting in 10,000 binary choices for analysis. Before conducting the study, we used the Sawtooth Guideline for Conjoint Power Analysis to ensure that our sample size was sufficient to detect a 5-percentage-point change in the dependent variable for main effects and interactions.27


We hypothesized that respondents would be more likely to deny ventilators to patients with disabilities than to patients without disabilities. We also hypothesized that respondents would deprioritize patients with lower long-term life expectancy and patients with noncongenital disabilities (that is, those that did not begin at birth). We classified each disability as physical or mental (see appendix exhibit 2)25 to test our hypothesis that mental disabilities would be deprioritized more than physical disabilities. Drawing on theories of intersectionality (which posit that intersecting axes of marginalization produce inequalities that are different from the results of each axis on its own), we hypothesized further that patients who were marginalized along multiple dimensions would experience compounded bias greater than the sum of each individual dimension added together.28 We categorized as marginalized the following patient groups: cisgender women, transgender men, transgender women, Black people, Asian people, people with disabilities, unemployed people, and people over age sixty-five. All hypotheses were preregistered with the Center for Open Science’s Open Science Framework.29

Consistent with established norms for conjoint experiments, we used multivariate linear regression incorporating all randomized traits as covariates. We also tested interactions between patients’ disability status and respondent demographics (gender, race and ethnicity, age cohort, education), as well as respondents’ self-reported political ideology and importance of religion (see appendix exhibit 4 for respondent characteristic questions).25 Our analyses used the weights provided by YouGov. As a robustness check, we produced unweighted analyses with standard errors clustered at the respondent level. These analyses produced substantively similar results (see appendix exhibits 12–16).25


We acknowledge several limitations. Although our representative sample can be generalized to the US public, the data should not be taken as representing particular categories of professionals such as clinicians, crisis standards of care policy makers, elected officials, or regulators. Because we aggregated individual disabilities into categories for analysis on the basis of their disability (physical, mental), impact on long-term life expectancy, and typical onset, specific disabilities may have driven some results. Therefore, we did not generalize conclusions to a broader category if a finding was driven only by a single disability (for example, if we found through incorporating individual disabilities into our regression that only diabetes showed an effect, whereas other acquired disabilities did not, we did not generalize the effect to all acquired disabilities).

Although we were able to identify the causal effect of the randomly assigned patient attributes on the likelihood of being allocated a ventilator, we could not identify causal effects of respondent attributes because we could not randomly assign respondents to particular demographic categories as we did the hypothetical patients. We could highlight correlations between respondents’ attributes (such as education level) and a lower likelihood of selecting certain kinds of patients, but we could not know for sure whether these attributes were causing the change or whether the relationship was driven by an unobservable factor.

Finally, we note that there are inherent trade-offs between forced-choice questions, such as those in our conjoint experiment, and open-ended ones. Open-ended questions would have allowed respondents to express detailed opinions on their preferred systems for fair allocation of a ventilator. However, the responses yielded from our conjoint design are less susceptible to social desirability bias and more likely to identify hidden preferences.

Study Results

Impact Of Disability And Other Patient Characteristics

As reflected in exhibit 1, respondents were 5.5 percentage points (p<0.001) less likely to allocate a ventilator to a patient with a disability than to a patient without a disability (details found in appendix exhibit 5).25 The level of deprioritization did not significantly vary by disability category (appendix exhibit 6).25 Respondents were also less likely to allocate a ventilator to transgender patients, patients ages 41–51 and 65–75, patients who had been unemployed before the pandemic, or patients who did not follow CDC guidelines (all at p<0.001 significance). Contrary to our hypothesis, respondents were 5.2 percentage points more likely to select Black patients (p<0.001) than they were to select White patients. Also contrary to our hypotheses, there were no significant differences in respondents’ allocation decisions for patients with lower versus normal life expectancies (p=0.511), acquired versus congenital onset (p=0.405), or physical versus mental disabilities (p=0.670) (see appendix exhibit 6).25 We also found no significant interactions between the marginalized patient identities other than disability (see above) and individual disabilities or between those identities and disability aggregated across types (appendix exhibits 7 and 8).25 We did, however, find a significant interaction between prior employment status and acquired disability (p=0.048; appendix exhibit 7).25

Exhibit 1 Marginal impact of patients’ demographic characteristics, employment, Centers for Disease Control and Prevention (CDC) guideline adherence, and disability status on their likelihood of being selected by respondents to receive a ventilator, 2021

SOURCE Original analysis of authors’ conjoint experiment data, January 29–February 4, 2021. NOTES Results are from a conjoint experiment on a representative sample of US residents. The x axis reflects marginal impact on likelihood of receiving a ventilator for each patient characteristic relative to the reference category in each domain. Both coefficients and confidence intervals are from a multivariate linear regression predicting choosing a patient for a ventilator. All p values are significant (p<0.01) except those for cisgender woman (p=0.189) and Asian race (p=0.225). Employment status refers to the patient’s status before the COVID-19 pandemic. “CDC guidelines” refers to the CDC’s social distancing guidelines for the COVID-19 pandemic. The sample conjoint task and definitions of disabilities used are in appendix exhibits 1 and 2 (see note 25 in text). The “any disability” coefficient comes from a separate regression in which all patient disability types were aggregated in an indicator variable.

Association Of Respondent Demographics With Disability Bias

We found that disability bias varied by respondent age and education level. Exhibit 2 shows that respondents from younger age cohorts displayed less bias toward disabled patients, with respondents in the youngest age cohort (Generation Z, born 1997–2012) almost 20 percentage points more likely (p<0.001) to select disabled patients than respondents in the oldest cohort (the Silent Generation, born before 1946) (see appendix exhibit 9).25Exhibit 2 also shows that respondents with college and postgraduate degrees deprioritized disabled patients to a greater extent than those whose education was limited to high school or less. Respondents with the highest level of education (postgraduate degrees) were 6.7 percentage points less likely (p=0.017) than those with a high school diploma or less to select a disabled patient (appendix exhibit 9).25

Exhibit 2 Marginal change in respondents’ likelihood of choosing a disabled person relative to a nondisabled person to receive a ventilator, by respondent age cohort and education level, 2021

Exhibit 2

SOURCE Original analysis of authors’ conjoint experiment data, January 29–February 4, 2021. NOTES Results are from a conjoint experiment on a representative sample of US residents. Confidence intervals (represented by whiskers) are calculated from two distinct multivariate regressions (one interacting disability with respondent age cohort, the other with respondent education level). All p values are significant (p<0.05) except Millennials (p=0.10) and Generation Z (p=0.270). Age cohorts were defined as follows: Silent Generation (born before 1946), Boomers (born 1946–64), Generation X (born 1965–80), Millennials (born 1981–96), and Generation Z (born 1997–2012). Education categories refer to the highest level of education completed by the respondent. Specific age and education questions are in appendix exhibit 4 (see note 25 in text).

Respondents for whom religion was important were considerably less likely to deprioritize patients with disabilities than nonreligious respondents (p=0.009), although this effect did not extend to bipolar disorder (p=0.658, see appendix exhibits 9 and 10)25 and was slightly short of conventional significance levels for mental disabilities more broadly (p=0.054, see appendix exhibit 9).25 We also found that conservative respondents deprioritized disabled patients overall more than liberal respondents (p=0.024), but paraplegia (p=0.007) and bipolar disorder (p=0.026) were the only specific disabilities for which there was a statistically significant difference between conservatives and liberals (see appendix exhibits 9 and 10).25 Hispanic respondents were less likely than White respondents to deprioritize patients with lower life expectancy (p=0.029) or with acquired (p=0.022) and physical (p=0.038) disabilities (see appendix exhibit 9).25 A robustness check that included additional controls for respondent-level characteristics did not substantively alter our results (see appendix exhibit 11).25

Racial Preference And Transgender Bias Associated With Political Ideology

Exhibit 3 examines the interaction between respondent political ideology and two patient attributes: race and gender (also see appendix exhibits 15 and 16).25 We found that liberal and moderate respondents prioritized Asian and Black patients relative to White patients. Liberal respondents, for instance, were 5.4 percentage points more likely (p=0.014) than conservative respondents to select Asian patients and 7.6 percentage points more likely (p=0.001) to select Black patients. When we examined the marginal impact of varying patient race by respondents’ political ideology (reflected in exhibit 3), we found that these differences reflected preference for Asian and Black patients on the part of liberal and moderate respondents, whereas conservative respondents were equally likely to select patients of any race. We also found that bias against transgender patients was driven by differences in ideology, with conservatives 18.6 percentage points less likely to select a transgender man (p<0.001) and 14.3 percentage points less likely to select a transgender woman (p<0.001) than they are to select a cisgender man.

Exhibit 3 Marginal change in respondents’ likelihood of choosing a patient to receive a ventilator, by patient race and gender and respondent political ideology, 2021

Exhibit 3

SOURCE Original analysis of authors’ conjoint experiment data, January 29–February 4, 2021. NOTES Results are from a conjoint experiment on a representative sample of US residents. The x axis reflects the marginal impact of a patient’s race and gender on the likelihood of receiving a ventilator by respondents’ political ideology relative to the reference groups. All p values are significant (p<0.05) for race except the impact of Asian patient race for respondents with conservative political ideology (p=0.152), Asian patient race for respondents who answered “not sure” regarding political ideology (p=0.481), Black patient race for respondents with conservative political ideology (p=0.365), and Black patient race for respondents who answered “not sure” regarding political ideology (p=0.922). For gender, the only p values that are significant at the 0.05 level are the impact of a patient being a transgender man for respondents with moderate political ideology (p=0.033) and respondents with conservative political ideology (p<0.001), as well as the impact of a patient being a transgender woman for respondents with moderate political ideology (p=0.007) and respondents with conservative political ideology (p<0.001). Results, including confidence intervals (represented by whiskers), are based on a single multivariate linear regression predicting choosing a patient for a ventilator. The specific question on respondents’ ideology are in appendix exhibit 4 (see note 25 in text).


Implications For Policy And Practice

To address concerns regarding disability discrimination in the allocation of scarce resources during COVID-19, in March 2020 the HHS Office for Civil Rights issued guidance informing providers that “assessments of quality of life, or judgments about a person’s relative ‘worth’ based on the presence or absence of disabilities” cannot be used to allocate resources.30 Subsequent guidance prohibited other discriminatory forms of prioritization in crisis standards of care policy making, including denial of care because of anticipated lower long-term life expectancy or greater resource use, while making clear that providers may prioritize based on short-term mortality risk.16

The public’s willingness to deprioritize disabled patients reflects bias against disabled people regardless of life expectancy.

In debates surrounding crisis standards of care policy making, some have argued that patients with diminished long-term life expectancy should be deprioritized relative to patients with a typical anticipated lifespan, claiming that policy makers should “consider the number of years of life saved” in addition to the number of lives saved.31 Our findings suggest that public support for deprioritizing disabled people with diminished life expectancy is not distinguishable from general bias against patients with disabilities, including those with normal life expectancy. Rather than a desire to “save the most life years” (itself impermissible according to recent HHS guidance),16 the public’s willingness to deprioritize disabled patients reflects bias against disabled people regardless of life expectancy. The data presented here did not allow us to identify underlying motivations; however, these findings would be consistent with discrimination motivated by quality-of-life judgments.

Although our experiment focused on ventilator allocation, it highlights the existence of biases against people with disabilities that may have implications for other areas of policy and practice, especially when resources are scarce. For example, disability bias may play a role in informing public policy decisions regarding access to and prioritization for organ transplantation and other scarce medical resources. Further research is needed to understand to what extent our results generalize to other contexts.

Our findings underscore the importance of bias mitigation in health policy making. Such measures might include, but should not be limited to, efforts to include people with disabilities on triage teams and hospital ethics committees, anti-bias training for triage team members, regular review of potential disparities in health outcome data, and greater investment in civil rights protection. Policy makers could also consider making additional investments in federally funded protection and advocacy programs, which provide legal assistance and advocate for systemic change in each state to protect the rights of people with disabilities.32 Although the protection and advocacy system has played an important role in enforcing disability rights laws during the COVID-19 pandemic, no dedicated funding stream currently supports protection and advocacy activities specific to health care more broadly; as a result, the resources available for populations not covered by other funding are limited. By authorizing an ongoing, health care–specific protection and advocacy funding stream, Congress could enhance efforts to address disability discrimination in health care.

Relationship Between Disability Bias And Respondents’ Education And Age

Our results also yield important insights about how different portions of the public view the deservingness of people with disabilities in health care contexts. Our finding that having a college or postgraduate degree is associated with greater disability bias, for example, stands in contrast to a body of work suggesting the opposite with respect to racial and anti-immigrant prejudice.33 It is unclear, however, whether greater disability bias is an unfortunate outcome of increasing education or whether people from backgrounds with lower levels of disability bias, perhaps due to greater personal or family experience with disability, are less likely to enter or complete higher education. The evidence from previous research about other forms of discrimination is mixed, with recent quasi-experimental studies reflecting contradictory findings on whether education causes an increase in prejudice.33,34 Future research could employ panel data to test stronger causal claims. Adding questions measuring disability bias to longitudinal studies of public opinion could also lay the groundwork for more meaningful quasi-experimental work about strategies for mitigating negative attitudes toward people with disabilities.

Our results show that completing higher education does not prevent disability bias and may instead be associated with it.

Regardless of causal origin, the presence of greater disability bias in college-educated populations should serve as a source of concern for civil rights policy makers. In the area of health care, policy makers often delegate to expert opinion, particularly on complex technical questions relating to clinical care. In the realm of bioethics, a long-standing body of work documents substantive disagreements between disability activists and bioethicists, with activists arguing that many bioethicists harbor troubling ideas about people with disabilities.13,35 Our results reinforce this concern, showing that completing higher education does not prevent disability bias and may instead be associated with it.

Our finding that younger age cohorts were less likely to deprioritize people with disabilities represents a promising sign for the future. We believe that this result is most likely a cohort rather than an aging effect, as older adults are more likely to have disabilities themselves. Insofar as personal experience with disability mitigates disability bias, the effect of population aging would predict the opposite result from the one we found.

We think that it is more likely that younger age cohorts have a different orientation toward people with disabilities than older cohorts and that this difference will persist over time. If so, this would suggest a future with less disability bias and greater equality of opportunity for people with disabilities. The inclusion of questions on disability bias in future public opinion surveys will help further validate our findings over time.

Role Of Political Ideology

Our findings show dramatically different responses to patient race and gender according to respondents’ political ideology. Deprioritization of transgender patients, for example, was found primarily among conservative respondents. Finding substantial bias against transgender people among political conservatives may reflect long-standing bias against transgender people, exacerbated by increasing attacks on transgender rights by conservative politicians and media.36,37

In contrast, the prioritization of Black and Asian patients was driven entirely by liberal and moderate respondents, whereas conservative respondents had no statistically significant preference. These results echo the ideologically divided debate surrounding state proposals to use race as a factor in the allocation of other COVID-19 resources, such as monoclonal antibodies.38

Arguments for taking race into account emerged in part because existing allocation protocols in crisis standards of care have well-documented racial biases.38 Even after long-term life expectancy is removed as an allocation criterion (which disadvantages Black Americans because of their higher rates of life-limiting comorbidities), racial biases in many of the prognostic scoring tools (such as the Sequential Organ Failure Assessment Score) used for assessing short-term mortality risk remain.38,39 For policy makers who have legal, ethical, or political concerns with using race as a factor, prior work suggests that prioritizing neighborhoods with greater social disadvantage can accomplish some of the same goals.40 However, some argue that use of place-based approaches alone is insufficient for allocation decisions, as such frameworks do not capture forms of disadvantage that are not geographically clustered (such as disability).13 Policy makers are still searching for tools to enhance equity that can garner legitimacy from public support across ideological divides.


Our findings provide support for long-standing concerns regarding disability bias in health care resource allocation.

Our findings provide support for long-standing concerns regarding disability bias in health care resource allocation. The bias we found against people with disabilities, older patients, and transgender patients provides empirical evidence to inform civil rights enforcement efforts and highlights the importance of expanding bias mitigation efforts in health policy making, especially in conditions of scarcity. Our investigation into public opinion helps identify the nature and intensity of disability bias in different portions of the public during COVID-19, but this research design could also be used as a blueprint for examining bias among clinicians and policy makers. Our research suggests that the fight for protecting the rights of people with disabilities is far from over and that policy makers and advocates should be particularly sensitive to potential biases against people with disabilities during public health crises such as the COVID-19 pandemic.


This work was supported by grants from the American Association of People with Disabilities and the World Health Organization (WHO). This research is part of an Epidemic Ethics/WHO initiative that has been supported by Foreign, Commonwealth & Development Office/Wellcome Grant No. 214711/Z/18/Z. Ari Ne’eman’s effort was supported by the National Institute of Mental Health, National Institutes of Health, under Award No. T32MH019733. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Ne’eman reports consulting income within the past twelve months from the Service Employees International Union, Inclusa, CareSource, and the Department of Health and Human Services Office for Civil Rights. The data presented here were not collected as part of his duties for any of these entities, including the Department of Health and Human Services, and the research, analysis, findings, and conclusions were not reviewed by them, nor do they necessarily represent their views. A version of these results was previously presented at the 2021 American Political Science Association Annual Meeting in Seattle, Washington, September 30–October 3, 2021, and the American Association on Intellectual and Developmental Disabilities 145th Annual Meeting (virtual), June 21–24, 2021 conventions. The authors thank Adrianna McIntyre, Bob Blendon, Jason Buxbaum, and Jennifer Hochschild for their helpful feedback in the development of this piece, and they thank Micah English for research assistance during the drafting of the manuscript. This is an open access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt, and build upon this work, for commercial use, provided the original work is properly cited. See


  • 1 Krahn GL, Walker DK, Correa-De-Araujo R. Persons with disabilities as an unrecognized health disparity population. Am J Public Health. 2015;105(Suppl 2):S198–206. Crossref, Medline, Google Scholar
  • 2 Iezzoni LI, Rao SR, Ressalam J, Bolcic-Jankovic D, Agaronnik ND, Donelan Ket al. Physicians’ perceptions of people with disability and their health care. Health Aff (Millwood). 2021;40(2):297–306. Go to the article, Google Scholar
  • 3 National Council on Disability. Medical futility and disability bias [Internet]. Washington (DC): NCD; 2019 Nov 20 [cited 2022 Jun 29]. Available from: Google Scholar
  • 4 Schneider AL, Ingram HM. Social constructions, anticipatory feedback strategies, and deceptive public policy. Policy Stud J. 2019;47(2):206–36. Crossref, Google Scholar
  • 5 Bell E, Ter‐Mkrtchyan A, Wehde W, Smith K. Just or unjust? How ideological beliefs shape street‐level bureaucrats’ perceptions of administrative burden. Public Adm Rev. 2021;81(4):610–24. Crossref, Google Scholar
  • 6 Naughton K, Schmid C, Yackee SW, Zhan X. Understanding commenter influence during agency rule development. J Policy Anal Manage. 2009;28(2):258–77. Crossref, Google Scholar
  • 7 van Oorschot W. Making the difference in social Europe: deservingness perceptions among citizens of European welfare states. J Eur Soc Policy. 2006;16(1):23–42. Crossref, Google Scholar
  • 8 Schneider A, Ingram H. Social construction of target populations: implications for politics and policy. Am Polit Sci Rev. 1993;87(2):334–47. Crossref, Google Scholar
  • 9 Cansunar A. Who is high income, anyway? Social comparison, subjective group identification, and preferences over progressive taxation. J Polit. 2021;83(4):1292–306. Crossref, Google Scholar
  • 10 Gething L. The Interaction with Disabled Persons Scale. Soc Behav Personal. 1994;9(5):23. Google Scholar
  • 11 Cottrell CA, Neuberg SL. Different emotional reactions to different groups: a sociofunctional threat-based approach to “prejudice.” J Pers Soc Psychol. 2005;88(5):770–89. Crossref, Medline, Google Scholar
  • 12 Dorfman D. Fear of the disability con: perceptions of fraud and special rights discourse. Law Soc Rev. 2019;53(4):1051–91. Crossref, Google Scholar
  • 13 Ne’eman A, Stein MA, Berger ZD, Dorfman D. The treatment of disability under crisis standards of care: an empirical and normative analysis of change over time during COVID-19. J Health Polit Policy Law. 2021;46(5):831–60. Crossref, Medline, Google Scholar
  • 14 Office of Sen. Kirsten Gillibrand. Gillibrand successfully leads bipartisan, bicameral call to protect civil rights for people with disabilities amidst COVID-19 pandemic; HHS issues new guidance to health care providers to enhance protections for those with disabilities [Internet]. New York (NY): Office of Sen. Kirsten Gillibrand; 2022 Feb 14 [cited 2022 Jun 29]. Available from: Google Scholar
  • 15 Office of Rep. Ayanna Pressley. Rep. Pressley calls on Governor Baker to rescind crisis of care standards that disproportionately harm communities of color & disability community [Internet]. Hyde Park (MA): Office of Rep. Ayanna Pressley; 2020 Apr 13 [cited 2022 Jun 29]. Available from: Google Scholar
  • 16 Department of Health and Human Services, Office for Civil Rights. FAQs for healthcare providers during the COVID-19 public health emergency: federal civil rights protections for individuals with disabilities under Section 504 and Section 1557 [Internet]. Washington (DC): HHS; 2022 Feb [cited 2022 Jun 6]. Available from: Google Scholar
  • 17 Social desirability bias is defined as “the tendency of some respondents to report an answer in a way they deem to be more socially acceptable than would be their ‘true’ answer…. The outcome of the strategy is overreporting of socially desirable behaviors or attitudes and underreporting of socially undesirable behaviors or attitudes.” Lavrakas PJ. Encyclopedia of survey research methods. London: SAGE Publications, Inc.; 2008. Vol 1, Social desirability; p. 826. Crossref, Google Scholar
  • 18 Buckwalter W, Peterson A. Public attitudes toward allocating scarce resources in the COVID-19 pandemic. PLoS One. 2020;15(11):e0240651. Crossref, Medline, Google Scholar
  • 19 Reeskens T, Roosma F, Wanders E. The perceived deservingness of COVID-19 healthcare in the Netherlands: a conjoint experiment on priority access to intensive care and vaccination. BMC Public Health. 2021;21(1):447. Crossref, Medline, Google Scholar
  • 20 Knotz CM, Gandenberger MK, Fossati F, Bonoli G. Public attitudes toward pandemic triage: evidence from conjoint survey experiments in Switzerland. Soc Sci Med. 2021;285:114238. Crossref, Medline, Google Scholar
  • 21 De la Cuesta B, Egami N, Imai K. Improving the external validity of conjoint analysis: the essential role of profile distribution. Polit Anal. 2022;30(1):19–45. Crossref, Google Scholar
  • 22 Horiuchi Y, Markovich ZD, Yamamoto T. Does conjoint analysis mitigate social desirability bias? Polit Anal. 2021 Sep 15. [Epub ahead of print]. Google Scholar
  • 23 Hainmueller J, Hopkins DJ. The hidden American immigration consensus: a conjoint analysis of attitudes toward immigrants. Am J Pol Sci. 2015;59(3):529–48. Crossref, Google Scholar
  • 24 Hainmueller J, Hangartner D, Yamamoto T. Validating vignette and conjoint survey experiments against real-world behavior. Proc Natl Acad Sci U S A. 2015;112(8):2395–400. Crossref, Medline, Google Scholar
  • 25 To access the appendix, click on the Details tab of the article online.
  • 26 Centers for Disease Control and Prevention [Internet]. Altanta (GA): CDC. Press release, CDC updates “How COVID is Spread” webpage; 2020 Oct 5 [cited 2022 Aug 23]. Available from: Google Scholar
  • 27 de Bekker-Grob EW, Donkers B, Jonker MF, Stolk EA. Sample size requirements for discrete-choice experiments in healthcare: a practical guide. Patient. 2015;8(5):373–84. Crossref, Medline, Google Scholar
  • 28 Crenshaw K. Demarginalizing the intersection of race and sex: a Black feminist critique of antidiscrimination doctrine, feminist theory, and antiracist politics. U Chi Legal F. 1989;1:139–67. Google Scholar
  • 29 Schneider M, Bell E, Ne’eman A, Strolovitch D. Who gets the last ventilator or compensatory school resources? Disability & deservingness during COVID-19 [Internet]. Charlottesville (VA): Center for Open Science; [cited 2022 Aug 23]. Available from: Google Scholar
  • 30 Department of Health and Human Services, Office for Civil Rights. Bulletin: civil rights, HIPAA, and the coronavirus disease 2019 (COVID-19) [Internet]. Washington (DC): HHS; 2020 Mar [cited 2022 Jun 6] Available at: Google Scholar
  • 31 Rajczi A, Daar J, Kheriaty A, Dastur C. The University of California crisis standards of care: public reasoning for socially responsible medicine. Hastings Cent Rep. 2021;51(5):30–41. Crossref, Medline, Google Scholar
  • 32 Department of Health and Human Services, Administration for Community Living. Protecting rights and preventing abuse of people with disabilities [Internet]. Washington (DC): HHS; 2022 May [cited 2022 Aug 10]. Available from: Google Scholar
  • 33 Lancee B, Sarrasin O. Educated preferences or selection effects? A longitudinal analysis of the impact of educational attainment on attitudes towards immigrants. Eur Sociol Rev. 2015;31(4):490–501. Crossref, Google Scholar
  • 34 Cavaille C, Marshall J. Education and anti-immigration attitudes: evidence from compulsory schooling reforms across Western Europe. Am Polit Sci Rev. 2019;113(1):254–63. Crossref, Google Scholar
  • 35 Stramondo JA. Bioethics, adaptive preferences, and judging the quality of a life with disability. Soc Theory Pract. 2021;47(1):199–220. Crossref, Google Scholar
  • 36 Kreitzer RJ, Smith CW. Reproducible and replicable: an empirical assessment of the social construction of politically relevant target groups. PS Polit Sci Polit. 2018;51(4):768–74. Crossref, Google Scholar
  • 37 Drabish K, Theeke LA. Health impact of stigma, discrimination, prejudice, and bias experienced by transgender people: a systematic review of quantitative studies. Issues Ment Health Nurs. 2022;43(2):111–8. Crossref, Medline, Google Scholar
  • 38 Schmidt H, Roberts DE, Eneanya ND. Rationing, racism, and justice: advancing the debate around “colourblind” COVID-19 ventilator allocation. J Med Ethics. 2022;48(2):126–30.35. Crossref, Medline, Google Scholar
  • 39 Miller WD, Han X, Peek ME, CharanAshana D, Parker WF. Accuracy of the Sequential Organ Failure Assessment Score for in-hospital mortality by race and relevance to crisis standards of care. JAMA Netw Open. 2021;4(6):e2113891. Crossref, Medline, Google Scholar
  • 40 Schmidt H, Shaikh SJ, Sadecki E, Gollust S. US adults’ preferences for race-based and place-based prioritisation for COVID-19 vaccines. J Med Ethics. 2022;48(7):497–500. Crossref, Medline, Google Scholar


Source link

Jorge Oliveira