Pith. sign in

REVIEW 5 major objections 4 minor 38 references

What Causes COVID-19 Fear? General Drivers of Fear During a Health Crisis

T0 review · 5 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims that the type of information source a person relies on is the dominant causal driver of COVID-19 fear, explaining 73.8% of its variance in U.S. survey data—about ten times the share of age or education.

desk verdict The descriptive CTIS analysis and the state-level clustering are worth a look, but the causal claim—that Source explains 73.8% of fear variance—is not supported by the text as written. read the letter →

arxiv 2508.20146 v3 pith:7JBMZIEW submitted 2025-08-27 cs.SI cs.CY

Daniele Baccega , Paolo Castagno , Antonio Fernández Anta , Juan Marcos Ramirez , Matteo Sereno This is my paper
classification cs.SIcs.CY
keywords FearSourceofinformationCOVID-19CausalinferenceOnlinesurveyUnitedStatesInfodemicDoWhy
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Using daily aggregated data from the Delphi US CTIS survey (May 2021–June 2022), the paper tries to establish that the type of information source a person relies on is the dominant causal driver of self-reported COVID-19 fear in the United States. A probabilistic causal model attributes 73.8% of the variance in fear to Source, against 6.9% for Age and 6.6% for Education. After causal adjustment, institutional and expert sources are estimated to increase fear, while politicians, religious leaders, and alternative channels reduce it. If the estimates hold, public-health messaging should be designed around the information ecosystem as much as around demographics. The paper also finds that state-level source-reliance patterns cluster along political lines, linking fear to the politics of information.

What carries the argument

The load-bearing object is a probabilistic causal graph with nodes Fear, Source, Age, and Education and directed edges Source→Fear, Age→Fear, Education→Fear, Age→Source, Education→Source, and Age→Education. Fitted to daily aggregated CTIS proportions, the graph's variance decomposition attributes 73.8% of Fear's variance to Source; backdoor adjustment over the same graph yields the average treatment effects of each source. The graph is what converts correlational survey patterns into the paper's causal claims.

What would settle it

Add state-level political orientation (e.g., 2020 presidential vote share) as an observed node in the causal graph of Figure 5 and recompute the variance shares; if Source's explained variance drops to the level of Age or Education, or the institutional-source ATE changes sign, the paper's dominance claim fails. A still more direct test would repeat the estimation on individual-level CTIS microdata with a political-affiliation covariate.

Watch

Extended reading notes

Core claim

On its own terms, the paper discovers that among the four variables in its causal graph—Fear, Source, Age, Education—Source is the dominant direct cause of COVID-19 fear, explaining 73.8% of the variance versus 6.9% for Age and 6.6% for Education. The average treatment effects estimated from the graph show institutional and expert sources raising fear and politicians, religious leaders, and alternative channels lowering it. Complementarily, K-means clustering of states by source reliance yields two clusters that closely match 2020 and 2024 U.S. election outcomes, with misclassifications mostly in swing states. The paper reads this as evidence that the information ecosystem, not demographics,

Load-bearing premise

The entire causal story depends on the assumption that daily aggregated survey proportions behave like individual-level observations for the DoWhy-GCM graph, and that the model has measured every factor that jointly drives both a person's choice of news source and their fear—most plausibly political identity, which the paper leaves out even though its own clustering shows source habits align with state politics.

Editorial extensions

If this is right

  • Crisis communication should weight information-source design as heavily as demographic targeting, since source choice explains roughly ten times more fear variance than age or education.
  • Institutional and expert sources will tend to amplify fear even after confounding adjustment, so official pandemic messaging needs explicit fear-mitigation framing rather than raw fact delivery.
  • Because state-level source usage aligns with political orientation, a single public-health message can produce opposite fear responses across red and blue states; targeted communication may be unavoidable.
  • The close tracking of fear and infection, coupled with source dominance, implies future crisis monitoring should track which sources populations are consuming in real time, not just infection counts.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The causal graph omits political orientation, even though the paper's own clustering shows source reliance tracks state politics; adding political identity as a common cause would likely shrink Source's variance share and could change the ATE signs.
  • Because the model runs on daily aggregates, not individual records, the 73.8% figure may reflect ecological correlation; re-estimating on individual-level CTIS microdata (or reconstructed joint distributions) is the clearest way to check the causal claim.
  • The findings invite a reverse-causality test: fear itself may shape source preferences (anxious people seek out experts) rather than only the reverse; a panel or instrumental-variable design could settle the direction.
  • A natural extension is to use a natural experiment, such as a local news blackout or a viral misinformation event, to compare fear changes in exposed versus unexposed areas with identical infection dynamics.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 4 minor

Summary. The paper analyzes COVID-19 fear during May 2021–June 2022 using aggregated Delphi US CTIS survey data. It constructs normalized singleton information-source proportions and fear scores, applies statistical tests and correlation analyses, then uses a DoWhy-based graphical causal model with variables Age, Education, Source, and Fear. The manuscript claims that Source explains 73.8% of the variance in Fear (vs. 6.9% for Age and 6.6% for Education), and the abstract reports average treatment effects of institutional versus political/religious/alternative sources on fear. It also clusters U.S. states by source reliance and reports that these clusters align closely with state-level political orientation.

Significance. If the causal claims were valid, the paper would make a substantive contribution to understanding how information ecosystems shape emotional responses during health crises, and the state-level clustering result is a falsifiable, interesting empirical finding. The paper is transparent about the survey questions and makes its aggregated data and code publicly available. However, the central causal claims are not supported by the material presented: the key estimating equation is self-referential, the causal model is applied at a different unit of analysis than the available aggregate data, and the abstract's ATE results are not reported in the body. These are load-bearing problems, not presentation issues.

major comments (5)
  1. [§3, Eq. (5)] Equation (5) defines phi_s'(v) as sum_{s in S} w_{s'|s} * phi_s'(v). The estimated quantity appears on both sides, making the definition tautological unless the equation is a typo. If the intended right-hand side uses phi_s(v) or another observable, the construction must be corrected; otherwise the singleton fear scores used throughout Figures 2–4 and Section 5 are undefined. This is not a minor notation issue: Eq. (5) is the basis for all downstream fear-score analyses.
  2. [§5, Fig. 5] The causal model is individual-level (Age, Education in {1,2,3}, Source in {1,...,9}, continuous Fear), but §3 supplies only daily aggregated proportions and scores. No routine is given for reconstructing individual rows from the aggregate data, and the text does not state whether DoWhy-GCM was run on daily time series, a pseudo-population, or state-day strata. Because the survey's Source item is multiple response with 256 combinations (Table 1 and §3), mapping a multi-select response to one of nine singleton categories requires row duplication or arbitrary selection, both of which distort the variance decomposition. The 73.8% figure therefore lacks a well-defined unit of analysis.
  3. [§5, Fig. 5 / §6] The DAG omits political orientation, yet §6 shows that state-level source reliance aligns strongly with partisan affiliation. If political identity or state political context influences both Source and Fear, the direct edge Source→Fear is confounded, and the causal variance shares and ATE directions claimed in the abstract are not identified by the stated model. At minimum, the authors should include a political-orientation variable or perform a sensitivity/negative-control analysis and state the ignorability assumption explicitly.
  4. [Abstract / §5] The abstract reports average treatment effects (institutional sources increase fear; politicians, religious leaders, and alternative channels reduce it). Section 5 reports only variance shares and does not display any ATE estimates, standard errors, or identification details. The reader cannot verify the abstract's quantitative claims. These estimates need to be reported with their estimating equation and uncertainty, or removed from the abstract.
  5. [§3 / §5] Even if Eq. (5) is corrected, the weights w_{s'|s} are fitted to the same aggregated fear scores later used as the outcome in the causal variance decomposition. Fitting and evaluating on the same data can inflate the apparent contribution of Source. The manuscript should state how overfitting is avoided (e.g., cross-validation, separate estimation samples) and provide a sensitivity analysis.
minor comments (4)
  1. [§3] Equation (4) uses phi_s'(v,t), which is not defined until Eq. (5). Please reorder the definitions or clarify that phi_s' is a placeholder in the optimization objective.
  2. [Table 1] The education categories 'Lower than high school' and 'High school / Bachelor's degree' appear to be approximate; please match the exact response labels of the Delphi US CTIS survey.
  3. [§6] The choice of k=2 is based on the Silhouette score, but the scores for k=3 and k=4 are not reported. Please provide the silhouette values used to select k.
  4. [Figures 3 and 7] Spearman correlation matrices are shown without confidence intervals or multiple-comparison corrections. Adding significance statements or shading would help interpret the many pairwise coefficients.

Circularity Check

2 steps flagged · score 8.0 of 10

Central fear-score construction is self-referential: Eq. (5) defines φ_s′ as a weighted sum of φ̂_{s′}, the very quantity it estimates, so the 73.8% Source variance claim inherits the tautology.

  1. self definitional [Section 3, 'Fear score of singleton information sources', Eqs. (4)-(5)]
    "Specifically, we use φ̂_{s′}(v,t)—the fear score among respondents who selected s′—as a proxy to estimate the weights. ... Using these weights, we then compute φ_s′ as φ_s′(v)=∑_{s∈S} w_{s′|s}·φ̂_{s′}(v) (5)"

    The target φ_s′(v) is defined as a weighted sum of φ̂_{s′}(v), which the text explicitly identifies as 'the fear score among respondents who selected s′'—the same quantity the equation is supposed to estimate. For fixed s′, the RHS is φ̂_{s′}(v) × ∑_s w_{s′|s}, a rescaling of its own input; it does not use the combination-level fear scores φ̂_s(v) at all. The 'disentangling' therefore never leaves the input space. If the RHS was meant to contain φ̂_s(v), the paper does not say so; as written, every downstream use of φ_s′ (including the causal decomposition of Section 5) is by construction a function of the very same per-source fear measure it claims to derive.

  2. fitted input called prediction [Section 5, 'Causal inference model']
    "Our analysis reveals that Source emerges as the most influential variable, explaining a substantial 73.8% of the variance in Fear."

    The 'Fear' entering the causal model is the φ_s′ built in Eq. (5), a per-source score constructed from the same singleton fear values φ̂_{s′} used as predictors in Eq. (4). Since each source has its own φ_s′ by construction, regressing Fear on Source and then reporting Source's variance share is a restatement of the construction: the outcome is indexed by the predictor. The 73.8% figure is a measure of how the fitted outcome separates by grouping, not an independent causal quantity.

full rationale

The paper's central quantitative claim—that information Source is the dominant driver of COVID-19 fear, explaining 73.8% of the variance—rests on the singleton fear score φ_s′ defined in Section 3. As written, Eq. (5) defines φ_s′(v) as a weighted sum of φ̂_{s′}(v), which the preceding sentence calls 'the fear score among respondents who selected s′'. That is exactly the quantity the equation is supposed to produce, so the definition is circular: the RHS is a linear rescaling of the input, not an independent estimate. The subsequent causal analysis in Section 5 uses this constructed φ_s′ as the continuous Fear variable, so the reported variance decomposition and the abstract's ATE signs inherit the same self-reference. The remainder of the paper—the correlational analyses, the temporal tracking of infection counts, and the state-level clustering that aligns with political orientation—does not depend on Eq. (5) and is not circular. There is no load-bearing self-citation chain: the authors' prior work [3] is used only for infection reconstruction, not for the fear-causal claim. The aggregate-to-individual mapping of the survey data is a separate validity concern, not a circularity. The score is 8 because the main causal result reduces, by the paper's own equations, to a self-referential construction; a corrected Eq. (5) would be needed to restore independent content.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The paper's central causal claim rests on five assumed inputs: an arbitrary fear-score weighting, fitted combination weights, a fitted causal model on aggregated data, a chosen cluster count, and an assumed recovery time for infection reconstruction. None of these is independently benchmarked.

free parameters (5)
  • Fear score weights w_f = {1, 0.34, -0.34, -1}
    Chosen by hand in Eq (3); the compensatory design is a modeling choice that determines the scale of all fear scores.
  • Combination weights w_{s'|s} = not reported
    Fitted per combination by least squares in Eq (4); they convert observed singleton fear scores into a final score, but the procedure is self-referential as written.
  • Causal model edge parameters (GCM) = not reported
    The 73.8%, 6.9%, 6.6% variance decomposition comes from a DoWhy-GCM fit on aggregated data; no coefficients or estimator details are given.
  • K-means number of clusters k = 2
    Selected via Silhouette score in Section 6; the two-cluster choice is crucial for the political alignment result.
  • SIRDS recovery time = 14 days
    Assumed in Section 3 to reconstruct infection counts; affects the infection trend used in correlations.
assumptions (5)
  • domain assumption Causal graph in Fig 5 is correct and complete, with no unobserved confounders.
    The validity of the Source->Fear estimate relies on this; political orientation and local epidemiology are omitted.
  • domain assumption Daily aggregated CTIS data can be treated as individual-level observations for causal inference.
    Section 5 does not describe how individual samples were reconstructed from aggregate proportions; causal inference on aggregates is an ecological assumption.
  • domain assumption The linear model in Eq (4) can disentangle singleton source contributions to fear.
    This is the identification assumption for the singleton fear scores; as written Eq (5) uses the target on both sides.
  • domain assumption Survey weights fully correct for Facebook-user and non-response bias.
    Section 3 relies on weights from CTIS; if weights are imperfect, all national estimates are biased.
  • domain assumption DoWhy-GCM correctly estimates causal effects from this data.
    The paper cites the package but does not justify its applicability to this aggregate, autocorrelated time series.

how reviews work

0 comments
Cite this review

Pith. "Pith review of What Causes COVID-19 Fear? General Drivers of Fear During a Health Crisis." pith.science (2026). https://pith.science/paper/7JBMZIEW

@misc{pith2026250820146,
  author       = {Pith},
  title        = {Pith review of: What Causes COVID-19 Fear? General Drivers of Fear During a Health Crisis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7JBMZIEW}},
  note         = {Machine review of arXiv:2508.20146}
}
read the original abstract

The COVID-19 pandemic triggered not only a global health crisis but also an infodemic, where exposure to heterogeneous information sources influenced public emotional responses. In this work, we investigate the determinants of self-reported fear of infection using data from the Delphi US CTIS survey. In particular, we analyze how demographic variables, epidemiological conditions, and exposure to different information sources shape fear levels. We introduce a Probabilistic Causal Model to estimate causal relationship strengths, identifying the variables that most strongly influence fear. Our results indicate that exposure to information sources accounts for a greater proportion of the variance in fear than demographic and epidemiological variables do. We further compute the Average Treatment Effect to quantify the impact of different information sources on fear. After causal adjustment, institutional and expert-driven sources are associated with increased fear levels, whereas politicians, religious leaders, and alternative information channels are associated with reduced fear. These findings highlight both the central role of the information ecosystem in shaping emotional responses during public health crises and the value of causal inference approaches for studying behavioral responses to pandemics.

Figures

Figures reproduced from arXiv: 2508.20146 by the authors.

Figure 1
Figure 1. Schematic representation of our study. Collectively, our findings underscore the critical role of informa￾tion sources in shaping public responses to large-scale crises. They also highlight the dangers of misinformation—particularly when disseminated by institutional or authoritative entities—which can distort public perception, exacerbate fear, and undermine trust in democratic institutions. The remainder of this a… view at source ↗
Figure 2
Figure 2. Temporal evolution of the singleton fear scores [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Spearman correlation coefficients among the time [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 5
Figure 5. Figure 5: Causal inference model used. driver of perceived fear remains challenging. Therefore, we employ causal inference methods to investigate these factors and determine which variable among Source, Age, and Education has the strongest influence on Fear. Specifically, causal…
Figure 7
Figure 7. Figure 7: Spearman correlation coefficients among the time [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 9
Figure 9. Figure 9: Clusters derived from K-means analysis based on [PITH_FULL_IMAGE:figures/full_fig_p008_9.png]
Figure 10
Figure 10. Figure 10: Clusters derived from K-means analysis based [PITH_FULL_IMAGE:figures/full_fig_p010_10.png]
Figure 12
Figure 12. Figure 12: Clusters derived from K-means analysis based on [PITH_FULL_IMAGE:figures/full_fig_p011_12.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

38 extracted references · 28 canonical work pages

  1. [1]

    Daniel Kwasi Ahorsu, Chung-Ying Lin, Vida Imani, Mohsen Saffari, Mark D Griffiths, and Amir H Pakpour. 2022. The fear of COVID-19 scale: development and initial validation. International journal of mental health and addiction 20, 3 (2022), 1537–1545

    work page 2022

  2. [2]

    Christina M Astley, Gaurav Tuli, Kimberly A Mc Cord, Emily L Cohn, Benjamin Rader, Tanner J Varrelman, Samantha L Chiu, Xiaoyi Deng, Kathleen Stewart, Tamer H Farag, et al . 2021. Global monitoring of the impact of the COVID- 19 pandemic through online surveys sampled from the Facebook user base. Proceedings of the National Academy of Sciences 118, 51 (20...

    work page doi:10.1073/pnas.2111455118 2021

  3. [3]

    Daniele Baccega, Paolo Castagno, Antonio Fernández Anta, and Matteo Sereno

    work page

  4. [4]

    Barnali Bhattacharjee and Tathagata Acharya. 2020. The COVID-19 pandemic and its effect on mental health in USA–a review with some coping strategies. Psychiatric Quarterly 91, 4 (2020), 1135–1145

    work page 2020

  5. [5]

    Mastakouri, and Dominik Janzing

    Patrick Blöbaum, Peter Götz, Kailash Budhathoki, Atalanti A. Mastakouri, and Dominik Janzing. 2024. DoWhy-GCM: An Extension of DoWhy for Causal Inference in Graphical Causal Models. Journal of Machine Learning Research 25, 147 (2024), 1–7. http://jmlr.org/papers/v25/22-1258.html

    work page 2024

  6. [6]

    Arcadio A Cerda and Leidy Y García. 2022. Factors explaining the fear of being infected with COVID-19. Health Expectations 25, 2 (2022), 506–512

    work page 2022

  7. [7]

    Carlos M Coelho, Panrapee Suttiwan, Nikolett Arato, and Andras N Zsido. 2020. On the nature of fear and anxiety triggered by COVID-19.Frontiers in psychology 11 (2020), 581314

    work page 2020

  8. [8]

    Saloni Dattani and Lucas Rodés-Guirao. 2024. 17 Key Charts to Understand the COVID-19 Pandemic. https://ourworldindata.org/key-charts-understand-covid- pandemic. Our World in Data (2024). Accessed: 2025-07-28

    work page 2024

Show all 38 references
  1. [9]

    Delphi Group, Carnegie Mellon University. [n. d.]. Delphi’s COVID-19 Trends and Impact Survey (CTIS). https://delphi.cmu.edu/epidemic-signals/ctis/. Ac- cessed: 2025-07-28

    2025

  2. [10]

    Delphi Group, CMU. 2020. COVIDcast. https://delphi.cmu.edu/covidcast/. Ac- cessed: 2025-07-28

    2020

  3. [11]

    Dolar Doshi, Parupalli Karunakar, Jagadeeswara Rao Sukhabogi, Jammula Surya Prasanna, and Sheshadri Vishnu Mahajan. 2021. Assessing coronavirus fear in Indian population using the fear of COVID-19 scale. International journal of mental health and addiction 19, 6 (2021), 2383–2391

    2021

  4. [12]

    Facebook. 2020. Facebook Data for Good: COVID-19 Symptom Survey – Request for Data Access. https://dataforgood.fb.com/docs/covid-19-symptom-survey- request-for-data-access/

    2020

  5. [13]

    J Fan, Y Li, K Stewart, AR Kommareddy, A Garcia, J O’Brien, A Bradford, X Deng, S Chiu, F Kreuter, et al. 2020. The University of Maryland Social Data Science Center Global COVID-19 Trends and Impact Survey. Partnership with Facebook (2020)

    2020

  6. [14]

    Yagmur Gokseven, Guzin Zeren Ozturk, Ecem Karadeniz, Ecem Sarı, Be- ray Gelmez Tas, and Hacı Mustafa Ozdemir. 2022. The fear of COVID-19 infection in older people. Journal of geriatric psychiatry and neurology 35, 3 (2022), 460– 466

    2022

  7. [15]

    Emanuele Guidotti. 2022. A worldwide epidemiological database for COVID-19 at fine-grained spatial resolution. Scientific Data 9, 1 (2022), 112. doi:10.1038/s41597- 022-01245-1

    2022 doi

  8. [16]

    Emanuele Guidotti and David Ardia. 2020. COVID-19 Data Hub. Journal of Open Source Software 5, 51 (2020), 2376. doi:10.21105/joss.02376

    2020 doi

  9. [17]

    John A Hartigan, Manchek A Wong, et al. 1979. A k-means clustering algorithm. Applied statistics 28, 1 (1979), 100–108. doi:10.2307/2346830

    1979 doi

  10. [18]

    Rosalind Jones, Davoud Mougouei, and Simon L Evans. 2021. Understanding the emotional response to Covid-19 information in news and social media: A mental health perspective. Human behavior and emerging technologies 3, 5 (2021), 832–842

    2021

  11. [19]

    Frauke Kreuter, Neta Barkay, Alyssa Bilinski, Adrianne Bradford, Samantha Chiu, Roee Eliat, Junchuan Fan, Tal Galili, Daniel Haimovich, Brian Kim, et al. 2020. Partnering with a global platform to inform research and public policy making: What needs to be in place to make a gl...

    2020 doi

  12. [20]

    Ruobing Li. 2022. Fear of COVID-19: What causes fear and how individuals cope with it. Health Communication 37, 13 (2022), 1563–1572

    2022

  13. [21]

    Linqi Lu, Jiawei Liu, Y Connie Yuan, Kelli S Burns, Enze Lu, and Dongxiao Li

  14. [22]

    Manuela Martínez-Lorca, Alberto Martínez-Lorca, Juan José Criado-Álvarez, Mª Dolores Cabañas Armesilla, and José M Latorre. 2020. The fear of COVID-19 scale: Validation in spanish university students. Psychiatry research 293 (2020), 113350

    2020

  15. [23]

    Frank J Massey Jr. 1951. The Kolmogorov-Smirnov test for goodness of fit. Journal of the American statistical Association 46, 253 (1951), 68–78

    1951

  16. [24]

    Jad Melki, Hani Tamim, Dima Hadid, Sally Farhat, Maha Makki, Lara Ghan- dour, and Eveline Hitti. 2022. Media exposure and health behavior during pandemics: The mediating effect of perceived knowledge and fear on compli- ance with COVID-19 prevention measures. Health communicat...

    2022

  17. [25]

    Gaëtan Mertens, Paul Lodder, Tom Smeets, and Stefanie Duijndam. 2023. Pan- demic panic? Results of a 14-month longitudinal study on fear of COVID-19. Journal of affective disorders 322 (2023), 15–23

    2023

  18. [26]

    Nadim Nachar et al. 2008. The Mann-Whitney U: A test for assessing whether two independent samples come from the same distribution. Tutorials in quantitative Methods for Psychology 4, 1 (2008), 13–20

    2008

  19. [27]

    Michael Nino, Casey Harris, Grant Drawve, and Kevin M Fitzpatrick. 2021. Race and ethnicity, gender, and age on perceived threats and fear of COVID-19: Evi- dence from two national data sources. SSM-population health 13 (2021), 100717

    2021

  20. [28]

    Judea Pearl. 2010. Causal Inference. In Proceedings of Workshop on Causality: Objectives and Assessment at NIPS 2008 (Proceedings of Machine Learning Research, Vol. 6), Isabelle Guyon, Dominik Janzing, and Bernhard Schölkopf (Eds.). PMLR, Whistler, Canada, 39–58. https://proce...

    2010

  21. [29]

    Shalini Quadros, Shalini Garg, Rupesh Ranjan, Guruprasad Vijayasarathi, and Mohammed A Mamun. 2021. Fear of COVID 19 infection across different cohorts: a scoping review. Frontiers in Psychiatry 12 (2021), 708430

    2021

  22. [30]

    Joshua A Salomon, Alex Reinhart, Alyssa Bilinski, Eu Jing Chua, Wichada La Motte-Kerr, Minttu M Rönn, Marissa B Reitsma, Katherine A Morris, Sarah LaRocca, Tamer H Farag, et al . 2021. The US COVID-19 Trends and Impact Survey: Continuous real-time measurement of COVID-19 sympt...

    2021 doi

  23. [31]

    Ketan Rajshekhar Shahapure and Charles Nicholas. 2020. Cluster quality analysis using silhouette score. In2020 IEEE 7th international conference on data science and advanced analytics (DSAA). IEEE, 747–748. doi:10.1109/DSAA49011.2020.00096

    2020

  24. [32]

    Samuel Sanford Shapiro and Martin B Wilk. 1965. An analysis of variance test for normality (complete samples). Biometrika 52, 3-4 (1965), 591–611

    1965

  25. [33]

    Amit Sharma and Emre Kiciman. 2020. DoWhy: An End-to-End Library for Causal Inference. arXiv preprint arXiv:2011.04216 (2020)

    2020 arXiv

  26. [34]

    infodemic

    Daniel H Solomon, Richard Bucala, Mariana J Kaplan, and Peter A Nigrovic. 2020. The “infodemic” of COVID-19. 1806–1808 pages

    2020

  27. [35]

    Charles Spearman. 1961. The proof and measurement of association between two things. (1961). doi:10.1037/11491-005

    1961 doi

  28. [36]

    NextGenerationEU

    Ann Marie Warren, Kiumars Zolfaghari, Michelle Fresnedo, Monica Bennett, Jamie Pogue, Anthony Waddimba, Michael Zvolensky, Per Carlbring, and Mark B Powers. 2021. Anxiety sensitivity, COVID-19 fear, and mental health: results from a United States population sample.Cognitive Be...

    2021

  29. [2021]

    Health Education & Behavior 48, 2 (2021), 132–139

    Source trust and COVID-19 information sharing: the mediating roles of emotions and beliefs about sharing. Health Education & Behavior 48, 2 (2021), 132–139

    2021

  30. [2024]

    Scientific Reports 14, 1 (2024), 19220

    Enhancing COVID-19 forecasting precision through the integration of compartmental models, machine learning and variants. Scientific Reports 14, 1 (2024), 19220. doi:10.1038/s41598-024-69660-5

    2024 doi

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.