COVID-19-Related Discrimination and Mental Distress: Mediating Role of Loneliness, Resilience, and Financial Worries

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is a well-executed study. The SEM analysis is solid, the writing is clear, and the inclusion of robust standard errors, sensitivity analyses, and a power analysis all reflect careful methodological thinking. The multi-mediator approach is a real strength.

 

My main suggestion for minor revision is to situate the contribution more accurately. The paper frames its application of stress process theory to discrimination and mental health as more novel than it is. The introduction and literature review present the mediating pathways (social, psychological, economic) as though this integrative, multi-domain approach is largely untested. But, there is a meaningful body of work and particularly in the sociology journal Society and Mental Health, that has applied stress process and stress proliferation frameworks to discrimination across various populations and stressor types in the COVID context. This includes research examining how discrimination operates through cascading secondary stressors and resource erosion to shape mental health, which is essentially the same theoretical architecture being used here.

 

This matters not because the study lacks value but because the actual contribution gets obscured when the framing overstates novelty. The paper's genuine strengths are the late-pandemic timing, the probability sample, and the specific combination of mediators in the COVID-19 context. Those are worth highlighting. But claims like the study contributing “in two important ways” (lines 48–50) and the emphasis on testing “three theoretically grounded mediators in a single structural equation model” as a distinctive approach would land better if situated against this existing work rather than implicitly positioned as breaking new ground.

 

I have several suggestions for addressing this:

1. The literature review (Section 2) would benefit from a paragraph engaging with prior stress process research on discrimination and mental health in the pandemic context that has already tested multi-mediator models, including work on disability, racial discrimination, and other stigmatized statuses. This would allow the authors to clarify what their study adds relative to that body of work, rather than relative to single-mediator or regression-based studies, which is a somewhat low bar.

 

2. Similarly, the framing in the introduction around mechanisms being “understudied” (line 43) needs qualifying. The mechanisms linking discrimination to mental health through stress proliferation and resource erosion have been studied. What's less studied is how those specific mechanisms operate in the during the late pandemic and that's a perfectly legitimate contribution worth owning more precisely.

 

3. The discussion could also be strengthened by connecting back to this broader literature. For instance, the finding that financial worries only mattered through the two-step pathway (via hardships and job disruptions) is interesting and could be discussed in light of what stress proliferation research has found about cascading economic stressors in other discrimination contexts.

 

Additional minor edits:

-Line 77: Citing Cronbach's alpha values in the introduction as a study contribution feels out of place. This is a methodological detail better suited to the measures section.

-The R² of 0.728 reported on line 334 seems very high for a model predicting mental distress. It would be worth double-checking this is for the full model and not an artifact of how Stata reports MLMV results, and perhaps briefly commenting on it.

-"Monte-Cario" (line 363) should be "Monte Carlo."

Author Response

Thank you for your review. Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript addresses an important topic: whether COVID-19-related discrimination remains associated with mental distress in the later stage of the pandemic, and whether loneliness, resilience, and financial worries help explain that relationship. The use of a community-based probability sample rather than a convenience sample is a strength, and the attempt to model multiple mediating pathways simultaneously is valuable. The paper is generally well organized, and the theoretical framing around stress process theory is appropriate.

At the same time, several issues need attention before the manuscript is ready for publication. The main concerns relate to the limits of causal interpretation with cross-sectional data, representativeness and weighting, measurement decisions, SEM specification and reporting, and some inconsistencies in wording and presentation.

1. The manuscript sometimes overstates what can be concluded from cross-sectional data The paper is careful in some places, but in other places the language still reads too causally. Terms such as “mediating role,” “pathways,” “contributes to,” and “intervention points” can imply temporal ordering and mechanism in a way that the design cannot fully support.

The authors should revise the manuscript more consistently to emphasize that the SEM estimates associations consistent with the proposed conceptual model, rather than demonstrating mediation in a causal sense. This is especially important in the abstract, discussion, and conclusion.

2. The use of address-based sampling is a strength, but the completed sample appears quite different from the county population, with overrepresentation of women, White respondents, and older adults. That is a major issue for generalizability, especially when the topic is discrimination.

The manuscript notes this imbalance, but the treatment is too limited. The authors should clarify whether survey weights were available or considered, why unweighted analyses were used if the aim is to describe a community-based probability sample , whether results are robust to any post-stratification or nonresponse adjustment and whether the low response rate could systematically bias estimates of discrimination and distress

At least the limitations section should discuss more explicitly that a probability-based sampling frame does not guarantee a representative analytic sample when response is selective.

3. The model includes multiple mediators, intermediate pathways, correlated residuals, and a fairly large number of covariates for a sample of 302. That is somewhat ambitious relative to the sample size.

The authors added a Monte Carlo power analysis, which helps, but more justification is still needed. In particular: Why was this degree of model complexity preferable to a more parsimonious model?  How stable are estimates for the smaller race categories and multiple categorical controls?  Were any model modifications driven by theory versus empirical fit? Were multicollinearity or sparse-cell issues assessed for the covariates?

4. The manuscript should better justify the measurement strategy for key constructs Several measures are brief or adapted, which is understandable, but the rationale should be developed more clearly.

 

• COVID-19-related discrimination: The 4-item adapted measure uses a 3-category response format (“no,” “maybe,” “yes”). The authors should explain more fully how “maybe” was coded and why treating the average score as continuous is appropriate.
• Financial worries: This is based on only two items, one of which concerns job security. This may not apply equally to retirees, unemployed respondents, or those outside the labor force, especially in an older sample with mean age 54.
• Material hardships and job disruptions are counts, but they seem to be modeled as continuous normal outcomes in SEM. That needs justification, especially given the reported skewness.
• Socioeconomic status is represented as a factor score combining education, income, and employment. The authors should describe the factor analysis more fully: extraction method, loadings, variance explained, and why this approach is preferable to including these components separately.

5. This topic is inherently tied to social stratification, yet the sample is predominantly White and subgroup analyses were not possible. That is a major substantive limitation. The discussion should say more explicitly that the meaning and prevalence of COVID-19-related discrimination likely differ across racial/ethnic groups, the findings may largely reflect patterns in a mostly White, older, Southern community sample. Thus, the study cannot speak strongly to populations most often highlighted in the COVID discrimination literature, especially Asians and other racialized minorities. Right now, the manuscript mentions this only briefly.

Minor comments

1. There are several wording and grammar issues throughout

A careful edit is needed. A few examples:

• “may have intensified exiting mental health disparities” should be “existing mental health disparities”
• “financial worries mediates” should be “financial worries mediate”  
• “Monte-Cario” should be “Monte Carlo”

3. Some citations and references need checking

A few issues stood out:

• “Center for Disease Control” should be “Centers for Disease Control and Prevention”
• Schmitt et al. (2014) appears twice in the reference list  
• Check whether all in-text citations appear in the reference list and vice versa

4. Table presentation could be improved

Table 2 is dense and difficult to read. Consider:

• separating structural paths of primary interest from covariate estimates
• moving the full covariate table to a supplement
• highlighting focal paths more clearly in the main text

Also, Table 1 says mental distress is “1–4,” but earlier the response options appear to be 1–5. Please verify the scaling and correct if needed.

5. The paper says missingness ranged from 0 to 9.3%, but it would help to know which variables had the most missingness, whether missingness was associated with key demographics and whether respondents with missing SES differed meaningfully from others
6. Why were political ideology and religion included, while other potentially relevant factors were not? A short explanation of the covariate selection strategy would improve clarity.

7.  The manuscript occasionally slides from a regional sample to broader language about 

Author Response

Thank you for your review. Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This study is considered to have derived meaningful results by exploring how COVID-19-related discrimination becomes a source of stress for individuals in relation to loneliness, resilience, and financial difficulties. Therefore, I believe the submitted manuscript may be accepted for publication in an academic journal, as it provides useful implications for academic, clinical, and policy purposes. However, areas requiring improvement and clarification within the manuscript were identified in this study. The following are areas I'd like to see revised:

 

1. If you intend to include multiple mediators in a structural equation model or a path model, in addition to the rationales for each path, the rationales for analyzing the mediating variables included in the model together must be valid. Therefore, such rationales must be presented in the introduction.
2. In fact, even though you used a statistical package primarily designed for analyzing structural equation models, the model presented here is a path analysis model rather than a structural equation model.
3. The explanation regarding statistical analysis needs to be described in more detail depending on the specific content of the analysis. Some of the currently described items could be omitted, as truly essential details related to each analysis are missing. I will comment again below regarding the issues with the statistical analysis in this study.
4. Please write Table 1 clearly so that it is easy to understand. At first glance, our readers might understand 'N' as the number of people included in that item. For example, it could be understood as the number of white people, but in this table, it appears to present the total number of people who responded to the survey with "yes," "no," or "other" options.
5. The most problematic aspect of this study is that it measured major psychological variables using a single item. It is unreasonable to validate path analysis models, let alone structural equation models, with such measurements. Such weaknesses could be presented as limitations of the study, given that it is classified as an exploratory study. However, the issue lies in the fact that parametric statistical analysis was performed on values ​​measured using a single Likert scale, even though these are not parametric parameters. Path analysis is not an analysis that can be performed non-parametrically. Furthermore, although not a principal analysis, beta values ​​were presented by converting nominal scales into dummy variables.
6. I would like to see a more statistically and logically sound discussion regarding these research findings presented in the discussion.

Author Response

Thank you for your review. Please see the attachment.

Author Response File: Author Response.pdf