Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This article addresses a topic of considerable public health relevance and examines the important relationships among food insecurity, perceived stress, diet quality, and eating behaviors. The manuscript is generally well organized and the topic is timely and meaningful. However, upon review, there are several aspects that require further clarification and improvement to strengthen the rigor, coherence, and scholarly value of the paper.

1. The title highlights the role of perceived stress in food insecurity, diet quality, and eating behaviors, but it does not clearly reflect that the manuscript actually includes two distinct studies with different samples and different outcome variables, namely reward-based eating and comfort eating. In addition, the title does not indicate that the analyses are cross-sectional. As currently written, it may give readers the impression that the study establishes a clear mechanism or causal pathway, which is not fully supported by the observational design. A more precise title that better reflects the study design and scope would improve accuracy.

2. The abstract is generally clear and structured appropriately, but it does not sufficiently emphasize that both studies are cross-sectional and therefore limited to associative rather than causal interpretation. It also does not clearly explain the important differences between Study 1 and Study 2 in terms of sample source, measurement tools, and outcomes. Furthermore, the abstract would be strengthened by including more specific statistical information or effect estimates so that readers can better assess the magnitude and practical significance of the findings. The concluding language regarding perceived stress as a mechanism may also need to be moderated.

3. The Introduction is clear, well structured, and provides an appropriate rationale for examining the relationships among food insecurity, perceived stress, and eating behaviors. However, the theoretical distinction between reward-based eating and comfort eating could be explained more clearly, as these outcomes are related but not identical constructs. In addition, the conceptual logic for combining two methodologically different studies into one manuscript could be better justified. A clearer conceptual framework would strengthen the linkage between the literature review, study aims, and mediation hypotheses.

4. The Methods section provides relevant details regarding the samples and instruments, but several issues require further clarification. First, the two studies differ substantially in sample source, participant composition, measurement tools, outcome variables, and statistical modeling, and the manuscript should explain more explicitly why these two studies are presented together as evidence for a common mediation framework. Second, the authors may wish to acknowledge the potential limitation associated with the sampling approach in Study 2, as participants were recruited from an online panel, which may not fully represent the general population. Individuals who voluntarily participate in survey panels may differ systematically from the broader population, potentially affecting generalizability. Third, the different recall periods for food insecurity, perceived stress, and dietary outcomes, especially in Study 1, may weaken the temporal logic of the mediation analyses and should be discussed more carefully. Finally, the measurement of comfort eating as a single self-reported frequency item may warrant additional justification regarding its validity.

5. The Results section reports the main findings, but the interpretation of the statistical results could be more informative. The manuscript would benefit from clearer explanation of the practical meaning of the reported indirect effects and their magnitude. In addition, the mediation analysis indicated that the indirect effect of perceived stress on diet quality was not statistically significant, and the authors may wish to provide further discussion of possible explanations for this null finding and its implications for the proposed conceptual framework, particularly in relation to Table 4. Because the two studies use different analytic approaches, including a standard mediation model in Study 1 and a two-part zero-inflated mediation model in Study 2, the comparability of the findings should also be clarified.

6. The Discussion relates the findings to prior literature, but some of the interpretations appear stronger than warranted by the design. In particular, perceived stress is interpreted as a mechanism linking food insecurity and reward-based eating. However, because the analyses are based on cross-sectional data, the findings should be interpreted as associative rather than causal relationships. The manuscript would benefit from a clearer distinction between mediation as a statistical pattern and mediation as evidence of an underlying causal mechanism. In addition, the null mediation finding for diet quality deserves fuller discussion, and the integration of Study 1 and Study 2 should be presented more cautiously given their methodological differences.

7. The Conclusion appropriately emphasizes the potential public health importance of the findings, but the wording may still be somewhat too strong given the cross-sectional nature of the data and the differences between the two studies. Statements suggesting that perceived stress is a key pathway should be framed more cautiously as preliminary or associative evidence. The recommendation to incorporate stress screening and management into food insecurity-related programs is reasonable, but it should be made clear that stronger longitudinal or experimental evidence is needed before firm intervention implications can be drawn.

8. The tables are informative, but there are several areas where presentation could be improved. In Table 1, the number of income categories appears relatively large, while several groups contain small sample sizes. The authors may consider collapsing adjacent income categories to enhance interpretability and improve the stability of the estimates. In addition, the manuscript would benefit from a visual conceptual figure or mediation pathway diagram to help readers understand the hypothesized relationships and the differences between the two studies. Greater consistency in the presentation of Study 1 and Study 2 results would also improve readability.

Author Response

Reviewer 1

This article addresses a topic of considerable public health relevance and examines the important relationships among food insecurity, perceived stress, diet quality, and eating behaviors. The manuscript is generally well organized and the topic is timely and meaningful. However, upon review, there are several aspects that require further clarification and improvement to strengthen the rigor, coherence, and scholarly value of the paper.

We thank Reviewer 1 for their thoughtful and constructive feedback. We appreciate the opportunity to clarify and strengthen several aspects of the manuscript and have revised the paper accordingly.

1. The title highlights the role of perceived stress in food insecurity, diet quality, and eating behaviors, but it does not clearly reflect that the manuscript actually includes two distinct studies with different samples and different outcome variables, namely reward-based eating and comfort eating. In addition, the title does not indicate that the analyses are cross-sectional. As currently written, it may give readers the impression that the study establishes a clear mechanism or causal pathway, which is not fully supported by the observational design. A more precise title that better reflects the study design and scope would improve accuracy.

                  Thank you for this helpful suggestion. We have revised the title to more accurately reflect the study design, scope, and samples. The updated title now indicates that the findings are based on two cross-sectional studies using U.S. samples.

The Role of Perceived Stress on Food Insecurity, Diet Quality, and Eating Behaviors

Perceived Stress and Associations Between Food Insecurity, Diet Quality, and Eating Behaviors: Evidence from Two Cross-Sectional Studies in U.S. Samples

2. The abstract is generally clear and structured appropriately, but it does not sufficiently emphasize that both studies are cross-sectional and therefore limited to associative rather than causal interpretation. It also does not clearly explain the important differences between Study 1 and Study 2 in terms of sample source, measurement tools, and outcomes. Furthermore, the abstract would be strengthened by including more specific statistical information or effect estimates so that readers can better assess the magnitude and practical significance of the findings. The concluding language regarding perceived stress as a mechanism may also need to be moderated.

                  Thank you for this feedback on how we can improve the abstract. We have addressed the comments and made edits to the abstract, specifically emphasizing the cross-sectional nature of the studies, differences between samples used for each study, including relevant coefficients, and tailoring language to reduce causal implications (lines 13-37).

Abstract: Background/Objectives: The present investigation examined whether perceived stress mediates associations between food insecurity and diet quality and eating behaviors. Methods: Study 1 used cross-sectional data from the National Heart, Lung, and Blood Institute Growth and Health Study (N = 624), in which participants Black and white women completed self-report measures of food security, perceived stress, diet quality, and re-ward-based eating. Study 2 used cross-sectional data from a national sample census-matched to the U.S. sample by age, gender, income, race/ethnicity, and census region (N = 1,993), with self-report measures of food security, perceived stress, and comfort eating. Mediation analyses tested the indirect effect of perceived stress on associations between food insecurity and diet quality, reward-based eating, and comfort eating, controlling for sociodemographic factors. Results: In Study 1, food insecurity was positively correlated with perceived stress (r = 0.30) and negatively correlated with diet quality (r = -0.11). Perceived stress mediated the relationship between food insecurity and higher reward-based eating (indirect effect = 0.14, 95% CI [0.08, 0.22]) but did not mediate the association between food insecurity and diet quality (Indirect effect = -0.04. 95% CI [-0.11, 0.03]). In Study 2, food insecurity was positively correlated with perceived stress (r = 0.42) and comfort eating (r = 0.19). Using a two-part mediation model, perceived stress mediated the association between food in-security and the frequency of comfort eating among individuals who reported at least one day of comfort eating, with the strongest indirect effect observed among food-insecure individuals (conditional indirect effect = 0.75, 95% CI [0.49, 1.13]). Conclusions: Across two cross-sectional studies, higher perceived stress statistically mediated the relationship between food insecurity and two forms of maladaptive eating behaviors, suggesting that stress-related processes may represent a potential mechanism underlying these relationships. Future work is needed to further evaluate the causal relationships between these constructs.  emerged as a psychological pathway linking food insecurity to maladaptive eating behaviors.

3. The Introduction is clear, well structured, and provides an appropriate rationale for examining the relationships among food insecurity, perceived stress, and eating behaviors. However, the theoretical distinction between reward-based eating and comfort eating could be explained more clearly, as these outcomes are related but not identical constructs. In addition, the conceptual logic for combining two methodologically different studies into one manuscript could be better justified. A clearer conceptual framework would strengthen the linkage between the literature review, study aims, and mediation hypotheses.

Thank you for the suggestions on how we can improve the introduction section. We have addressed the comments below.

We have provided a theoretical distinction between reward-based eating and comfort eating in the Introduction (lines 65-71) and Study 2 (lines 308-310).

1. Introduction

In the current examination of this framework, two different forms of eating behavior are tested. Reward-based eating refers to patterns of eating driven by heightened responsiveness to palatable foods, including a lack of control, reduced satiety, and preoccupation with food, and may occur in the absence of negative affect. In contrast, comfort eating refers specifically to eating in response to negative emotions as a coping strategy. Although these constructs are related and may co-occur, they reflect distinct behavioral responses and may differentially relate to stress-related mechanisms.

3. Study 2

Comfort eating is a distinct construct from reward-based eating in that it can occur in response to negative emotions, whereas reward-based eating can occur in their absence.

We have elaborated on the logic for combining two different studies to test a shared conceptual framework. Please see the response to comment 4 for the detailed response and subsequent line edits.

4. The Methods section provides relevant details regarding the samples and instruments, but several issues require further clarification. First, the two studies differ substantially in sample source, participant composition, measurement tools, outcome variables, and statistical modeling, and the manuscript should explain more explicitly why these two studies are presented together as evidence for a common mediation framework. Second, the authors may wish to acknowledge the potential limitation associated with the sampling approach in Study 2, as participants were recruited from an online panel, which may not fully represent the general population. Individuals who voluntarily participate in survey panels may differ systematically from the broader population, potentially affecting generalizability. Third, the different recall periods for food insecurity, perceived stress, and dietary outcomes, especially in Study 1, may weaken the temporal logic of the mediation analyses and should be discussed more carefully. Finally, the measurement of comfort eating as a single self-reported frequency item may warrant additional justification regarding its validity.

Thank you for bringing these points to our attention. We have addressed each of the suggestions in their respective portions of the manuscript.

We have addressed the first comment and specified throughout the manuscript (i.e., introduction (lines 61-63), overview of study aims (lines 75-80); discussion (lines 462-466)) that the two studies are not intended to be direct replications but are instead presented together as a larger conceptual test in which perceived stress serves as a potential mediator in the relationship between food insecurity and diet quality and maladaptive eating behaviors.

1. Introduction

These studies provide preliminary evidence that perceived stress may serve as a key psychological mechanism that explains how food insecurity is related to diet quality and eating behaviors. Less is known about whether perceived stress also contributes to reward-driven eating and comfort eating. The current studies extend prior work by examining how stress-related mechanisms may impact diet quality and eating behaviors. Conceptually, perceived stress may represent a broader pathway linking food insecurity to multiple diet- and eating-related outcomes…The current work extends prior research by examining this conceptual framework across distinct outcomes and samples.

1.1 Overview of Study Aims

Across two studies, we tested perceived stress as an underlying mechanism linking food insecurity to poorer diet quality and eating behaviors. Across two studies, we tested a shared conceptual framework in which perceived stress statistically mediated associations between food insecurity and diet- and eating-related outcomes. The studies were not intended as direct replications. Instead, they were designed as complementary tests of whether stress-related processes operate similarly across samples and across related, but distinct, outcomes, including diet quality, reward-based eating, and comfort eating.

4. Strengths and Limitations

Second, the findings highlight the potential of perceived stress as a key psychosocial mechanism through which food insecurity may influence eating behavior. Identifying stress as a mediator underscores the importance of psychological pathways in addition to material constraints in shaping responses to food insecurity., and identifies a potential intervention target.  Despite differences in sample characteristics, measurements, and outcomes, findings from these two studies were consistent in supporting perceived stress as a mechanism linking food insecurity to diet- and eating-related outcomes. This convergent evidence across distinct contexts, although preliminary, supports the role of stress-related processes implicated in food insecurity.

We have addressed the second comment regarding the limitation with relying on a sample recruited for an online survey in the strengths and limitations section (lines 514-519).

4.3 Strengths and Limitations

Fourth, data for Study 2 were collected using an online survey. Past research has emphasized that web surveys are susceptible to self-selection bias as well as under coverage of individuals who do not have access to the internet (Bethlehem, 2010). Future research should replicate these findings using probability-based samples or mixed-methods data collection approaches to better capture individuals who may be underrepresented in web-based surveys.

We have addressed the third comment regarding the time frame for the focal variables in the strengths and limitations section (lines 496-502).

4.3 Strengths and Limitations

Although sociodemographic variables like income, education level, and household size were statistically accounted for, additional unmeasured variables (e.g., food bank usage, mental health symptoms) may confound the observed associations., and thus longitudinal work is necessary. Furthermore, the cross-sectional nature of the data for both studies made it difficult to fully assess the temporal ordering between the focal variables. Measures of food insecurity, perceived stress, and dietary outcomes were assessed over different recall periods, which may weaken the temporal logic of the mediation analyses. Future research can leverage longitudinal designs that assess the focal variables over comparable recall periods to better establish temporal ordering.

We have addressed the fourth comment regarding the limitations with a single-item comfort eating measure (lines 445-449).

3.3 Study 2 Discussion

Comfort eating was assessed using a single-item self-report measure, which may not fully capture the complexity of this behavior and may limit construct validity. Although the single-item measure was used to reduce participant burden in, future work should consider using multi-item validated scales to more comprehensively assess comfort eating.

 

5. The Results section reports the main findings, but the interpretation of the statistical results could be more informative. The manuscript would benefit from clearer explanation of the practical meaning of the reported indirect effects and their magnitude. In addition, the mediation analysis indicated that the indirect effect of perceived stress on diet quality was not statistically significant, and the authors may wish to provide further discussion of possible explanations for this null finding and its implications for the proposed conceptual framework, particularly in relation to Table 4. Because the two studies use different analytic approaches, including a standard mediation model in Study 1 and a two-part zero-inflated mediation model in Study 2, the comparability of the findings should also be clarified.

Thank you for this comment and the recommendations on how we can improve the results of the studies. We have removed tables depicting results from mediation analyses and instead included visual depictions of results in mediation path diagrams (lines 717-752).

We have included clearer explanation of the practical meaning of the reported indirect effects and their magnitude. We have revised the Results section to more explicitly describe the practical meaning of the indirect effects in both studies. Specifically, we now clarify how the mediated pathways reflect associations between food insecurity, perceived stress, and outcomes in substantive terms rather than relying solely on statistical significance. For Study 1, we additionally included an effect size estimate to aid interpretation of the indirect effect within a standard linear mediation framework (lines 197-200; 209-216). Because Study 2 used a two-part zero-inflated model that yields nonlinear and conditional indirect effects, comparable standardized effect size estimates were not computed (lines 387-389).

2.1.4 Statistical Analyses

The magnitude of the indirect effect was quantified using the squared standardized indirect effect (Lachowicz et al., 2018), which reflects the proportion of variance in the outcome explained by the mediator.

2.2 Study 1 Results

Study 1 tested if greater perceived stress mediated the relationship between food insecurity and diet quality (Figure 2). (Table 4). The indirect effect of perceived stress was not significant. Although food insecurity was associated with higher perceived stress, higher perceived stress was not significantly associated with diet quality. Study 1 also examined if perceived stress mediated the relationship between food insecurity and greater reward-based eating (Figure 3). Results showed the indirect effect was significant, wherein food insecurity status was associated with higher greater perceived stress, and greater higher perceived stress was linked to greater higher reward-based eating drive. The squared standardized indirect effect of perceived stress indicated a small effect size (u = 0.02) according to Cohen’s effect size benchmarks.

3.1.4 Statistical Analyses

Due to the two-part model including nonlinear components and conditional indirect effects, a standardized effect size comparable to those from linear mediation models were not computed.

We have updated the results section for Study 2 to elaborate on the findings from the two-part mediation model (lines 400-409).

3.2 Study 2 Results

Table 5 displays zero-order correlations for Study 2. Food insecurity was positively correlated with perceived stress and comfort eating. Study 2 leveraged a two-part mediation model to distinguish between the occurrence of comfort eating (any vs. none) and the frequency of comfort eating among individuals reporting at least one day (Table 6) (figure 4). This model tested whether perceived stress mediated the association between food in-security and the frequency of comfort eating. Results from the binary portion of the model indicated that food insecurity was associated with a greater likelihood of any comfort eating (versus not comfort eating) via higher perceived stress. Results indicated that Furthermore, among individuals who reported at least one day of comfort eating, higher perceived stress was significantly associated with a greater number of comfort-eating days. This conditional indirect effect was observed among both food secure and food insecure individuals, indicating that perceived stress was associated with greater frequency of comfort eating across groups. However, the indirect effect was slightly larger among individuals experiencing food insecurity, suggesting that the stress-related pathway linking food insecurity to comfort eating may be stronger in this group. This indirect effect was present both among individuals without food insecurity and among those experiencing food insecurity, with the strongest effect observed in the food-insecure group.

We have also added language to contextualize the magnitude of the indirect effects in the Discussion section (lines 246-250; 449-453), noting that these effects are small and represent one of several potential pathways linking food insecurity to eating behaviors.

2.3 Study 1 Discussion

Lastly, although the indirect effect was statistically significant, the magnitude was small per Cohen’s effect size benchmarks. This suggests that perceived stress represents one of several potential pathways linking food insecurity to reward-based eating. These findings should therefore be interpreted as evidence of a partial, rather than comprehensive, explanatory mechanism.

4.2 Study 2

Effect sizes for the indirect effects estimated from the two-part zero-inflated mediation model were not calculated because the model contained non-linear components (binary outcome and a zero-inflated outcome). Thus, the strength of the indirect effects should be interpreted in relative rather than standardized terms.

Thank you for the helpful suggestion to elaborate on the null findings. We have expanded the Discussion to provide additional interpretation of the null indirect effect and its implications for our conceptual framework (lines 274-298).

4.1 Study

These findings may also suggest that the psychological stress pathway linking food insecurity to poor outcomes is more relevant for certain eating behaviors than for overall dietary quality. Behaviors such as reward-based eating may be more directly responsive to perceptions of stress whereas diet quality may be more strongly shaped by structural and economic constraints. This pattern suggests that the conceptual framework in which perceived stress operates as a mediator may be more relevant for eating behaviors than for overall diet-related outcomes, although further work is needed to confirm this distinction.

Thank you for the important recommendation of discussing limitations related to differences in analytic approaches. We have edited the Limitations section to reflect this comment (lines 519-527).

4.3 Strengths and Limitations

Finally, the two studies used different analytic approaches, including a standard mediation model in Study 1 and a two-part zero-inflated mediation model in Study 2, which limits the comparability of findings across studies. Because the zero-inflated model involves nonlinear components, standardized effect size estimates are not directly comparable to those from linear mediation models. Results should be interpreted as preliminary support for a broader conceptual framework in which perceived stress operates as a mediator, rather than as evidence of differences in effect magnitude across eating-related outcomes Future research using comparable analytic approaches across outcomes may help clarify differences in effect size for perceived stress.

6. The Discussion relates the findings to prior literature, but some of the interpretations appear stronger than warranted by the design. In particular, perceived stress is interpreted as a mechanism linking food insecurity and reward-based eating. However, because the analyses are based on cross-sectional data, the findings should be interpreted as associative rather than causal relationships. The manuscript would benefit from a clearer distinction between mediation as a statistical pattern and mediation as evidence of an underlying causal mechanism. In addition, the null mediation finding for diet quality deserves fuller discussion, and the integration of Study 1 and Study 2 should be presented more cautiously given their methodological differences.

Thank you for bringing this issue to our attention. We have addressed the issue of causal language throughout the introduction and discussion sections.

1. Introduction

The current cross-sectional work studies extend prior research by examining this conceptual framework across distinct outcomes and samples. work by examining how stress-related mechanisms may impact diet quality and eating behaviors (lines 71-73).

1.1 Overview of Study Aims

Study 1 relied on a sample of Black and white women from the National Heart, Lung, and Blood Institute Growth and Health study (NGHS) to test the hypotheses that stress would mediate the association relationship between food insecurity and diet quality and reward-based eating (lines xx-xx). Study 2 relied on a census-matched U.S. sample from the Eating in America Study to test the hypothesis that perceived stress would mediate the association relationship between food insecurity and more frequent comfort eating among a U.S. census-matched sample (lines 82-85).

2. Study 1

The first aim of Study 1 was to investigate if greater perceived stress mediated the association relationship between food security status and poor diet quality (lines 90-91).

The second aim of Study 1 was to examine if perceived stress mediated the association relationship between food insecurity a form of maladaptive eating, reward-based eating (lines 105-106).

To our knowledge, research has yet to examine the association relationship between food insecurity and reward-based eating in humans or the underlying mechanism that may drive this association. The current study sought to address this gap in the literature by testing perceived stress as a potential statistical mediator (lines 118-121).

2.3 Study 1 Discussion

Study 1 revealed that greater perceived stress mediated the association relationship between food insecurity and reward-based eating (lines 224-225).

These results also identify perceived stress as a novel psychosocial mechanism yield a statistical pattern suggesting perceived stress as a novel psychosocial mediator that might underlie this association (lines 228-229).

However, perceived stress did not mediate the association relationship between food insecurity and diet quality (lines 224-225).

3. Study 2

Study 2 assessed if perceived stress mediated the association relationship between food insecurity and comfort eating (lines 278-279)

Thus, the current study examined if the association relationship between food insecurity and comfort eating was mediated by greater perceived stress (lines 290-292).

3.3 Study 2 Discussion

Study 2 leveraged data from a large census-matched U.S. sample to examine whether higher perceived stress mediated the association relationship between food insecurity and the frequency of comfort eating (lines 404-406).

These findings provide preliminary evidence support for psychological stress may be a mechanism linking as a mediator of the association between experiences of food insecurity and comfort eating (lines 408-409).

4. Strengths and Limitations

Second, the findings highlight the potential of perceived stress as a key psychosocial mechanism mediator through which food insecurity may influence eating behavior. Identifying stress as a mediator support for stress as a mediator underscores the importance of psychological pathways, in addition to material constraints, in shaping influencing responses to food insecurity, and identifies a potential intervention target (lines 435-438).

Finally, Study 2 leveraged a census-matched U.S. sample, increasing the generalizability of findings that identify perceived stress as a key psychological driver potential mediator of comfort eating in both food-secure and food-insecure individuals (lines 455-457).

First, data for both studies were cross-sectional and thus, no causal or mechanistic claims can be made (lines 458-459).

Additionally, thank you for pointing out the importance of elaborating on the null mediation finding for diet quality and differing methodology of Study 1 and Study 2. We have addressed the need for greater elaboration on the null mediation finding and specified that the differing measurement period for diet quality and reward-based eating, may have contributed to the null mediation finding for diet quality but not reward-based eating (lines 252-265).

2.3 Study 1 Discussion

Second, the discrepancy in recall periods for measures assessing food insecurity, perceived stress, and diet quality may have reduced our ability to detect any indirect effects of perceived stress, thus contributing to the null associations reported here. Food insecurity was assessed over the past 12 months, whereas diet quality only reflected the participant’s intake during a three-day period. As a result, the diet measure may not have adequately captured the cumulative impact of food insecurity on diet quality across the same period. Given the cyclical nature of food insecurity and food assistance programs, participants may have reported their diet quality in a period of relative abundance, when diet quality may not have been compromised. This temporal mismatch may have obscured the association between food insecurity and diet quality, and any potential indirect effects of perceived stress. Future research should assess food insecurity, stress, and dietary outcomes over aligned time frames using longitudinal designs to provide a robust test of the proposed mediation pathway.

7. The Conclusion appropriately emphasizes the potential public health importance of the findings, but the wording may still be somewhat too strong given the cross-sectional nature of the data and the differences between the two studies. Statements suggesting that perceived stress is a key pathway should be framed more cautiously as preliminary or associative evidence. The recommendation to incorporate stress screening and management into food insecurity-related programs is reasonable, but it should be made clear that stronger longitudinal or experimental evidence is needed before firm intervention implications can be drawn.

Thank you for this comment. We have revised the wording of the Conclusion to appropriately align with the cross-sectional nature of our data and underscore the necessity of longitudinal and experimental evidence to further support presented claims (lines 496-513).

5. Conclusions

Food insecurity is associated with a range of adverse can lead to a host of health outcomes complications, which that can may arise not only from limited economic resources, but also from the psychological strain of living in such conditions. By utilizing a psychological approach, these results provide preliminary evidence that perceived stress may act as a pathway through which food insecurity negatively impacts diet quality and eating behaviors. The present findings identify perceived stress as a common correlate of both food insecurity and maladaptive eating behaviors and provide preliminary, associative evidence of a pathway linking food insecurity to eating behaviors through perceived stress. Given the cross-sectional nature of the data and differences across studies, these findings should be interpreted cautiously. Replication of these results using longitudinal analyses or experimental methods would provide stronger rationale for including stress screening and stress management resources in conjunction with food insecurity programs. Future research using longitudinal and experimental designs is needed to clarify causal relationships and to determine whether targeting perceived stress may be an effective component of interventions addressing food insecurity.

8. The tables are informative, but there are several areas where presentation could be improved. In Table 1, the number of income categories appears relatively large, while several groups contain small sample sizes. The authors may consider collapsing adjacent income categories to enhance interpretability and improve the stability of the estimates. In addition, the manuscript would benefit from a visual conceptual figure or mediation pathway diagram to help readers understand the hypothesized relationships and the differences between the two studies. Greater consistency in the presentation of Study 1 and Study 2 results would also improve readability.

Thank you for the feedback on how to improve table presentation and manuscript readability. We have addressed the first comment by including statistical notations as superscripts and collapsing adjacent income categories in Table 1. We also applied these suggestions to Table 3 by including statistical notations as superscripts, combining “Non-Binary” and “Other” gender categories due to small sample size, and collapsing adjacent income and education categories. Mediation pathway diagrams were created to address the second comment and aid readers in understanding hypothesized relationships. We have also restructured the manuscript by presenting Study 1 and Study 2 in separate, logical order.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript investigates whether perceived stress mediates the association between food insecurity and diet-related outcomes, including diet quality, reward-based eating, and comfort eating. The study uses two datasets: (1) a cohort from the National Heart, Lung, and Blood Institute Growth and Health Study and (2) a large census-matched U.S. sample. The topic is relevant and timely, as food insecurity represents an important public health concern and psychological mechanisms linking food insecurity to dietary behaviours remain insufficiently understood. The use of two independent datasets and the application of mediation models represent strengths of the study.

Overall, the manuscript is well structured and addresses an important research question. However, several conceptual, methodological, and analytical issues should be clarified to strengthen the robustness and interpretation of the findings.

The following comments are intended to help improve the clarity and scientific rigor of the manuscript.

 

Major comments

1. Both Study 1 and Study 2 rely on cross-sectional data. Although the authors acknowledge this limitation in the discussion, mediation analyses are often interpreted as implying causal pathways. Because the variables were measured at the same time point, the directionality between food insecurity, perceived stress, and eating behaviours cannot be definitively established. It would be helpful if the authors: more clearly emphasize that the mediation models represent statistical associations rather than causal pathways, and; further discuss possible bidirectional relationships between stress and eating behaviours.

 

2. In both studies, food insecurity was dichotomized so that a score of zero indicated food security and any score above zero indicated food insecurity. While the authors justify this approach by citing literature suggesting that marginal food insecurity may affect health, dichotomization may also reduce statistical power and obscure differences between levels of food insecurity (e.g., marginal, low, very low).

The authors may wish to: explain this decision in greater detail; discuss potential implications for interpretation; consider sensitivity analyses using the full scale or ordinal categories.

 

3. In Study 1, food insecurity was measured over the previous 12 months, whereas diet quality was derived from three days of dietary records. This mismatch in reference periods may weaken the ability to detect associations between food insecurity and diet quality. The authors acknowledge this limitation briefly, but the discussion could further elaborate on how this discrepancy may have contributed to the null mediation effect observed for diet quality.

 

4. Study 1 uses data from the NGHS cohort composed exclusively of Black and White women recruited during adolescence and followed into adulthood. Although this design allows examination of long-term outcomes, it limits the generalizability of findings to: men; other racial or ethnic groups; younger populations.

The authors discuss this limitation but may wish to expand on how these characteristics influence the interpretation of the results.

 

5. The statistical analyses involve Bayesian mediation models with multiple imputation using rBlimp. While this approach is appropriate, the description of the analytic strategy could be clarified. In particular: provide more detail on the specification of the mediation models; clarify how indirect effects were interpreted in the zero-inflated model used in Study 2; briefly explain the rationale for the number of imputations used. Providing a short conceptual explanation would improve accessibility for readers less familiar with Bayesian mediation analysis.

 

6. The manuscript proposes perceived stress as a key psychological mechanism linking food insecurity and maladaptive eating behaviours. However, other psychological processes may also contribute, including: depressive symptoms; food-related anxiety; emotional regulation difficulties. The discussion could benefit from acknowledging these additional pathways and situating perceived stress within a broader theoretical framework.

 

Minor comments

1. The manuscript is generally well written, but a few minor typographical issues were noted. For example: “percieved stress” appears in the Results section and should be corrected to “perceived stress.”; “lack of control overeating” should likely read “lack of control over eating.”
2. In Study 2, comfort eating was assessed with a single question asking participants how many days they engaged in comfort eating during the past 30 days.

Although this measure is practical, the authors may wish to discuss potential limitations associated with using a single-item measure for a complex behaviour.

3. The discussion could benefit from a more explicit comparison of the findings with previous literature on: stress-related eating; emotional eating in food-insecure populations.

 

This manuscript addresses an important and understudied question regarding the psychological mechanisms connecting food insecurity and eating behaviours. With clarification of several methodological points and refinement of the discussion, the study could make a valuable contribution to the literature.

Author Response

Reviewer 2

This manuscript investigates whether perceived stress mediates the association between food insecurity and diet-related outcomes, including diet quality, reward-based eating, and comfort eating. The study uses two datasets: (1) a cohort from the National Heart, Lung, and Blood Institute Growth and Health Study and (2) a large census-matched U.S. sample. The topic is relevant and timely, as food insecurity represents an important public health concern and psychological mechanisms linking food insecurity to dietary behaviours remain insufficiently understood. The use of two independent datasets and the application of mediation models represent strengths of the study.

Overall, the manuscript is well structured and addresses an important research question. However, several conceptual, methodological, and analytical issues should be clarified to strengthen the robustness and interpretation of the findings.

The following comments are intended to help improve the clarity and scientific rigor of the manuscript.

We thank Reviewer 2 for their comments and suggestions. We have carefully addressed each point raised and provide detailed responses below.

Major comments

1) Both Study 1 and Study 2 rely on cross-sectional data. Although the authors acknowledge this limitation in the discussion, mediation analyses are often interpreted as implying causal pathways. Because the variables were measured at the same time point, the directionality between food insecurity, perceived stress, and eating behaviours cannot be definitively established. It would be helpful if the authors: more clearly emphasize that the mediation models represent statistical associations rather than causal pathways, and; further discuss possible bidirectional relationships between stress and eating behaviours.

Thank you for this comment and the recommendation on how we can better convey this important limitation. We have made edits in the abstract (lines 13-37) as well as the limitations paragraph (lines 440-454).

Abstract

Background/Objectives: The present investigation examined whether perceived stress statistically mediated mediates the association between food insecurity and diet quality, as well as maladaptive eating behaviors (i.e., reward-based eating, comfort eating)… Conclusions: Across two cross-sectional studies, higher perceived stress statistically mediated the relationship between food insecurity and two forms of maladaptive eating behaviors, suggesting that stress-related processes may represent a potential mechanism underlying these relationships. Future work is needed to further evaluate the causal relationships between these constructs.

4.3 Strengths and Limitations

Results should be interpreted considering several limitations. First, data for both studies were cross-sectional and thus, no causal claims can be made. The mediation models reported should therefore be interpreted as statistical associations among focal variables rather than evidence for causal pathways. Although socio-demographic variables like income, education level, and household size were statistically accounted for, additional unmeasured variables (e.g., food bank usage, mental health symptoms) may confound the observed associations. Cross-sectional data may also obscure bidirectional relationships among the focal variables. For example, higher levels of perceived stress may contribute to reward-based and comfort eating, while these eating behaviors may, in turn, influence levels of perceived stress. Furthermore, the cross-sectional nature of the data for both studies made it difficult to fully assess the temporal ordering between the focal variables. Measures of food insecurity, perceived stress, and dietary outcomes were assessed over different recall periods, which may weaken the temporal logic of the mediation analyses. Future research can leverage longitudinal designs that assess the focal variables over comparable recall periods to better establish temporal ordering and assess potential bidirectional relationships.

In both studies, food insecurity was dichotomized so that a score of zero indicated food security and any score above zero indicated food insecurity. While the authors justify this approach by citing literature suggesting that marginal food insecurity may affect health, dichotomization may also reduce statistical power and obscure differences between levels of food insecurity (e.g., marginal, low, very low).

The authors may wish to: explain this decision in greater detail; discuss potential implications for interpretation; consider sensitivity analyses using the full scale or ordinal categories.

Thank you for this comment. Dichotomization was used to capture the presence versus absence of food insecurity, consistent with prior work indicating that even marginal food insecurity is associated with adverse health outcomes. However, we acknowledge that dichotomization may obscure variability across levels of food insecurity and reduce statistical precision. Thus, we conducted sensitivity analyses using the full food insecurity scale. Results were substantively consistent with the primary analyses that included a binary food insecurity measure, with significant indirect effects via perceived stress observed for reward-based eating (Study 1) and comfort eating (Study 2). Because the pattern of findings remained unchanged, we retained the dichotomous specification in the manuscript for interpretability and consistency with prior literature. We have updated the statistical analyses sections for both studies to reflect this comment (lines 190-194; 340-344).

2.1.4 Statistical Analyses

At the recommendation of a reviewer, sensitivity analyses were conducted using the full food insecurity scale specified as an ordinal predictor rather than a dichotomized predictor. Results were substantively consistent, with no meaningful changes in statistical significance or interpretation. Accordingly, results from models using the dichotomized food insecurity variable are presented.

2.2.4 Statistical Analyses

Similar to Study 1, sensitivity analyses were conducted using the full food insecurity scale specified as an ordinal predictor. Results were substantively consistent, with no meaningful changes in statistical significance or interpretation. Results from models using the dichotomized food insecurity variable are reported.

In Study 1, food insecurity was measured over the previous 12 months, whereas diet quality was derived from three days of dietary records. This mismatch in reference periods may weaken the ability to detect associations between food insecurity and diet quality. The authors acknowledge this limitation briefly, but the discussion could further elaborate on how this discrepancy may have contributed to the null mediation effect observed for diet quality.

Thank you for bringing this limitation to our attention. We have expanded upon this limitation in the discussion section for Study 1 (lines 238-251).

4.1 Study 1

Second, the discrepancy timeline  in recall periods for measures assessing food insecurity, perceived stress, and diet quality may have reduced our ability to detect any indirect effects of perceived stress, thus contributing to the null associations reported here. might not have allowed for an appropriate test of the hypothesis. Food insecurity was assessed over the past 12 months, whereas diet quality only reflected only the participant’s intake during a three-day period. As a result, the diet measure may not have adequately captured the cumulative impact of food insecurity on diet quality across the same period. Given the cyclical nature of food insecurity and food assistance programs [48], participants may have reported their diet quality in a period of relative abundance, when diet quality may not have been compromised. This temporal mismatch may have obscured the association between food insecurity and diet quality and, relatedly, any potential indirect effects of perceived stress. Future research should assess food insecurity, stress, and dietary outcomes over aligned time frames using longitudinal designs to provide a robust test of the proposed mediation pathway.  Future work may consider longitudinal assessments that capture food insecurity and diet patterns over the same period. 

Study 1 uses data from the NGHS cohort composed exclusively of Black and White women recruited during adolescence and followed into adulthood. Although this design allows examination of long-term outcomes, it limits the generalizability of findings to: men; other racial or ethnic groups; younger populations.

The authors discuss this limitation but may wish to expand on how these characteristics influence the interpretation of the results.

Thank you for commenting on this crucial limitation. We have elaborated on this limitation in the Study 1 specific discussion section (lines 215-223).

4.1 Study 1

Study 1 revealed that greater perceived stress mediated the relationship between food insecurity and reward-based eating. Findings from this study align with animal models showing that food insecurity may alter alters reward processes and, to our knowledge, yields the first evidence of a correlational link between food insecurity and reward-based eating in humans. These results also identify perceived stress as a novel psychosocial mechanism that might underlie this association. Results from Study 1 should be interpreted considering the characteristics of the analytic sample. This study relied on a sample composed exclusively of middle-aged Black and white women from Richmond, CA, which limits the generalizability of these findings to other demographic groups. It is possible that the strength or nature of the observed associations may differ across men, individuals from other racial and ethnic backgrounds, and younger populations. As such, the current findings should be interpreted within the context of this specific sample.  Future research could consider replicating the associations in diverse samples of individuals with food insecurity. Furthermore, future research should identify additional psychological and biological mediators that underly the association between food insecurity and greater reward-based eating drive.

The statistical analyses involve Bayesian mediation models with multiple imputation using rBlimp. While this approach is appropriate, the description of the analytic strategy could be clarified. In particular: provide more detail on the specification of the mediation models; clarify how indirect effects were interpreted in the zero-inflated model used in Study 2; briefly explain the rationale for the number of imputations used. Providing a short conceptual explanation would improve accessibility for readers less familiar with Bayesian mediation analysis.

Thank you for the recommendations on how we can improve our statistical analyses sections. We have made the suggested edits in both studys’ respective sections (lines 190-198; 319-348).

2.1.4 Statistical Analyses

Hypotheses and data analytic plans were pre-registered (https://shorturl.at/amKNY). Zero-order correlations were conducted to examine the relationship between all variables of interest. Missing data were addressed using Bayesian multiple imputation implemented in rBlimp. Mediation models were also estimated in rBlimp following a standard mediation framework.  A Bayesian multiple imputation approach was utilized to address missing data using rBlimp; mediation models were also conducted using rBlimp. Food insecurity status was included as a binary predictor and perceived stress as a continuous mediator. Diet quality and reward-based eating were included as outcomes in separate models. In these models, the indirect effect represented the product of the association between food insecurity and perceived stress (path a) and the association between perceived stress and the outcome (path b). Models included age, race, household size, income, and education as covariates. Race and education were dummy coded. The models were estimated using 20 imputations to ensure stable parameter estimates under missing data, with 5,000 burn-in cycles followed by 10,000 posterior iterations across two Markov Chain Monte Carlo chains. Indirect, direct, and total effects were estimated. An indirect effect was considered statistically significant if the 95% credible interval did not contain zero.

2.2.4. Statistical Analyses

Hypotheses and data analytic plans were pre-registered (https://shorturl.at/amKNY). Zero-order correlations were conducted between continuous variables to examine the relationship among between all variables of interest. Similar to Study 1, missing data were handled using a Bayesian multiple imputation approach in rBlimp.  was utilized to address missing data using rBlimp. Mediation models with a zero-inflated outcome were also conducted using rBlimp. Mediation models were estimated using a two-part count framework to account for the large number of zero values in the comfort eating outcome. Food insecurity status was included as a binary predictor and perceived stress included as a continuous mediator. Comfort eating was modeled using two components: (1) a binary component estimating the likelihood of reporting any comfort eating (i.e., any vs. none), and (2) a count component estimating the number of comfort-eating days among participants who reported at least one day of comfort eating. Number of days of comfort eating was included as a zero-inflated outcome to account for the high number of zeroes reported. A two-part, zero-inflated model was used: a binary model that pre-dicted whether participants reported any comfort eating, and a count model that predicted the number of days among participants who reported at least one day spent comfort eating. Two Indirect effects were computed for the zero binary and count parts of the model. For the binary component, the indirect effect represents the extent to which food insecurity is associated with the likelihood of any comfort eating through perceived stress. For the count component, the indirect effect represents the extent to which food insecurity is associated with the frequency of comfort eating through perceived stress among those who engage in the behavior. Because the count model is nonlinear, conditional indirect effects were estimated separately for individuals with and without food insecurity. Indirect effects were computed as the product of the association between food insecurity and perceived stress (path a) and the association between perceived stress and each outcome component (path b). Models included age, gender, race, household size, income, and education as covariates, with race, gender, and education dummy coded .Models included age, gender, race, household size, income, and education as covariates. The models were estimated using two 2 imputations, with 10,000 burn-in cycles followed by 20,000 posterior iterations across two Markov Chain Monte Carlo chains. A conditional indirect effect was considered statistically significant if the 95% credible interval did not include zero.

The manuscript proposes perceived stress as a key psychological mechanism linking food insecurity and maladaptive eating behaviours. However, other psychological processes may also contribute, including: depressive symptoms; food-related anxiety; emotional regulation difficulties. The discussion could benefit from acknowledging these additional pathways and situating perceived stress within a broader theoretical framework.

Thank you for these thoughtful suggestions. We have addressed the situation of perceived stress within a broader theoretical framework by specifying perceived stress as a potential mediator, rather than a potential key psychological mechanism. Additionally, we included an acknowledgement of other potential pathways that could interact with perceived stress to influence maladaptive eating behaviors.

3.3 Study 2 Discussion

However, consistent with broader theoretical models of stress and coping, perceived stress is unlikely to influence comfort eating in isolation. Other psychological processes such as depressive symptoms, food-related anxiety, and emotion regulation difficulties may interact with perceived stress, thereby contributing to comfort eating (lines 393-396).

4. Strengths and Limitations

Second, the findings highlight the potential of perceived stress as a key psychosocial mechanism mediator through which food insecurity may influence eating behavior (lines 413-414).

Finally, Study 2 leveraged a census-matched U.S. sample, increasing the generalizability of findings that identify perceived stress as a key psychological driver potential mediator of comfort eating in both food-secure and food-insecure individuals (lines 424-426).

Minor comments

The manuscript is generally well written, but a few minor typographical issues were noted. For example: “percieved stress” appears in the Results section and should be corrected to “perceived stress.”; “lack of control overeating” should likely read “lack of control over eating.”

Thank you for bringing these minor typographical issues to our attention. We have addressed the specific instances mentioned as well as throughout the manuscript.

1.1 Study 1

Reward-based eating is characterized by a lack of control over eating overeating, reduced satiety from eating, and preoccupation with food.

2.1.2 Measures

The 9-item Reward-Based Eating Drive scale [22] assessed the strength of re-ward-based motivations for eating including a lack of control over eating overeating, preoccupation with food, and lack of satiation (e.g., “When it comes to foods I love, I have no will power.”).

3.1 Study 1

The indirect effect of perceived stress was not significant. Study 1 also examined if perceived percieved stress mediated the relationship between food insecurity and greater reward-based eating.

In Study 2, comfort eating was assessed with a single question asking participants how many days they engaged in comfort eating during the past 30 days.

Although this measure is practical, the authors may wish to discuss potential limitations associated with using a single-item measure for a complex behaviour.

We appreciate this comment and the opportunity to address it in the strengths and limitations section (lines 399-403).

4.2 Study 2 Discussion

Results from Study 2 should be interpreted with certain limitations. Comfort eating was assessed using a single-item self-report measure, which may not fully capture the complexity of this behavior and may limit construct validity. Although the single-item measure was used to reduce participant burden in, future work should consider using multi-item validated scales to more comprehensively assess comfort eating.

The discussion could benefit from a more explicit comparison of the findings with previous literature on: stress-related eating; emotional eating in food-insecure populations.

Thank you for this comment. We addressed the explicit comparison of findings with the previous literature on stress-related eating in the discussion for Study 2 on the significant mediation effect of perceived stress for the association between food insecurity and the frequency of comfort eating (lines 388-391).

3.3 Study 2 Discussion

Research on the perceived benefits and frequency of comfort eating indicates that negative mood management, reduced boredom, and improved cognitive competence are associated with comfort eating frequency, but not perceptions of comfort eating as pleasurable and accompanied by positive emotions. Thus, individuals who perceive greater psychological stress may be more likely to manage their stress by engaging in comfort eating.

This manuscript addresses an important and understudied question regarding the psychological mechanisms connecting food insecurity and eating behaviours. With clarification of several methodological points and refinement of the discussion, the study could make a valuable contribution to the literature.

We thank the reviewer for this positive and constructive assessment. We have clarified the methodological details and refined the Discussion accordingly, which we believe enhances the rigor and contribution of the manuscript.

Reviewer 3 Report

Comments and Suggestions for Authors

The topic of this paper seems relevant and actual. However, some points need to be corrected/revised before publishing.

Title: 

It would be useful to mention the country and that the data come from national representative sample.

Abstract:

Study aim are clearly described. However, in the first sentence the authors should clarify what they mean by eating behavior. As it looks it mainly refers to maladaptive eating behaviors (last sentence). Methods and Results are given in an appropriate way.

Introduction:

Background (lines 34-60) seems adequate, however, there is a need for more explanation for the presentation of two separate studies.

In addition, as I see these studies are not only seperated but also based on a logical order. Therefore, it would be more appropriate to present them separately as Study 1 (background, study aims, methods, results, discusssion) and Study 2 (background, study aims, methods, results, discusssion) with a final conclusion. This structure would be more transparent.

Figures of the mediation models are also  missing from both studies with the coefficients.

Overall, the paper needs reconsideration in several points but I encourage the authors to do them carefully.

Author Response

Reviewer 3

The topic of this paper seems relevant and actual. However, some points need to be corrected/revised before publishing.

Title:

It would be useful to mention the country and that the data come from national representative sample.

                  Thank you for this helpful suggestion regarding the title of the manuscript. We agree that specifying the country and sample characteristics improves clarity. We have revised the title to indicate that the findings are based on two cross-sectional studies using U.S. samples. To maintain clarity, we did not include additional detail regarding the census-matched nature of one sample in the title. However, this information is described in the Abstract and Methods sections.

The Role of Perceived Stress on Food Insecurity, Diet Quality, and Eating Behaviors

Perceived Stress and Associations Between Food Insecurity, Diet Quality, and Eating Behaviors: Evidence from Two Cross-Sectional Studies in U.S. Samples

Abstract:

Study aim are clearly described. However, in the first sentence the authors should clarify what they mean by eating behavior. As it looks it mainly refers to maladaptive eating behaviors (last sentence). Methods and Results are given in an appropriate way.

Thank you for the feedback on how to improve our abstract. We have made updates to the first sentence to specify the eating behaviors being examined (lines 13-15).

Abstract: Background/Objectives: The present investigation examined whether perceived stress mediates statistically mediated the association between food insecurity and diet quality, and as well as maladaptive eating behaviors (i.e., reward-based eating, comfort eating).

Introduction:

Background (lines 34-60) seems adequate, however, there is a need for more explanation for the presentation of two separate studies.

Thank you for this helpful comment. We have revised the manuscript to more clearly articulate the rationale for including two studies. In the Introduction (lines 63-73) and Overview of Study Aims (lines 75-80), we now clarify that both studies test a shared perceived stress mediation framework across distinct outcomes and samples. We also added language in the Strengths and Limitations (lines 416-420; 459-467) sections to further contextualize how the studies provide complementary evidence.

In addition, as I see these studies are not only seperated but also based on a logical order. Therefore, it would be more appropriate to present them separately as Study 1 (background, study aims, methods, results, discusssion) and Study 2 (background, study aims, methods, results, discusssion) with a final conclusion. This structure would be more transparent.

Thank you for the recommendation on how to better structure the manuscript. We have made the suggested edits by presenting Study 1 (Methods, Results, Discussion) and Study 2 (Methods, Results, Discussion) separately and in logical order. We present a general introduction and broad overview of study aims as well as a general strengths and limitations section.

Figures of the mediation models are also  missing from both studies with the coefficients.

Thank you for the recommendation to include mediation models. We have removed the tables depicting mediation results and included figures for each of the mediation models that were tested with coefficients (lines 634-641).

Overall, the paper needs reconsideration in several points but I encourage the authors to do them carefully.

We appreciate the reviewer’s careful evaluation of the manuscript. We have addressed the points raised and revised the manuscript to improve clarity and rigor.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I have carefully evaluated the revised manuscript and the authors’ detailed responses to the reviewers’ comments. I acknowledge the substantial effort undertaken to improve the manuscript, particularly in enhancing the methodological description, refining the interpretation of findings, and addressing several previously identified limitations.

The revision represents a clear improvement over the previous version. Nonetheless, while many concerns have been addressed, I find that several issues remain only partially resolved. These pertain primarily to the precision of causal language, depth of theoretical framing, completeness of methodological justification, and clarity of statistical interpretation for a broader readership.

In its current form, the manuscript is approaching publication standard but still requires further refinement.

 

Major Comments

A central concern relates to the continued presence of language that may be interpreted as implying causality. Although the authors have made commendable efforts to temper their wording, certain expressions, such as references to “mechanisms” or “key psychological pathways,” remain conceptually strong given the cross-sectional nature of the data. I strongly encourage the authors to ensure full consistency throughout the manuscript, explicitly acknowledging the impossibility of establishing directionality and considering the plausibility of bidirectional relationships.

The justification for the dichotomisation of food insecurity has improved, and the inclusion of sensitivity analyses using ordinal measures is a positive addition. However, the implications of this analytical decision are not sufficiently interrogated. In particular, the potential loss of information, attenuation of variability, and masking of dose–response relationships should be more explicitly discussed. A more critical reflection would strengthen the methodological transparency of the study.

The issue of temporal misalignment between the measurement of food insecurity and dietary intake has been appropriately acknowledged and reasonably justified. I consider this point satisfactorily addressed.

Regarding generalisability, the authors now recognise the limitations of the sample composition. However, the discussion remains somewhat superficial. It would be important to go beyond stating limited generalisability and instead reflect on how the observed associations might differ across populations, including the possibility of variation in both magnitude and direction of effects.

The statistical analysis section has been substantially expanded and now demonstrates improved rigour. The description of the Bayesian mediation approach, including imputation procedures and model estimation, is appropriate. Nevertheless, I find that the presentation remains highly technical and may not be fully accessible to readers without advanced statistical expertise. The manuscript would benefit from the inclusion of brief interpretative explanations, particularly regarding the meaning of credible intervals and indirect effects.

Although the authors acknowledge that perceived stress accounts for only a small proportion of the observed associations, the discussion of alternative psychological pathways remains limited. A more comprehensive theoretical integration, including constructs such as depression, anxiety, or coping mechanisms, would provide a more nuanced interpretation of the findings.

In Study 2, the clarification of the comfort eating measure is appreciated. However, the limitations associated with the use of a single-item measure are not sufficiently addressed. This represents an important methodological constraint that should be explicitly discussed, particularly in terms of construct validity.

 

Minor Comments

The manuscript would benefit from further refinement of language and style. Some sections remain overly dense, and simplifying sentence structure would enhance readability. Additionally, certain parts of the discussion are still somewhat repetitive and could be streamlined.

Tables and figures are generally clear, but their interpretability could be improved by more explicitly highlighting key findings.

Finally, the manuscript would benefit from a clearer articulation of practical implications, particularly in relation to public health or intervention design.

Author Response

Reviewer 2

 

I have carefully evaluated the revised manuscript and the authors’ detailed responses to the reviewers’ comments. I acknowledge the substantial effort undertaken to improve the manuscript, particularly in enhancing the methodological description, refining the interpretation of findings, and addressing several previously identified limitations.

 

The revision represents a clear improvement over the previous version. Nonetheless, while many concerns have been addressed, I find that several issues remain only partially resolved. These pertain primarily to the precision of causal language, depth of theoretical framing, completeness of methodological justification, and clarity of statistical interpretation for a broader readership.

 

In its current form, the manuscript is approaching publication standard but still requires further refinement.

 

We appreciate Reviewer 2’s comments and suggestions on how we can further refine the manuscript. Below we include line edits for each of the comments.

 

Major Comments

 

A central concern relates to the continued presence of language that may be interpreted as implying causality. Although the authors have made commendable efforts to temper their wording, certain expressions, such as references to “mechanisms” or “key psychological pathways,” remain conceptually strong given the cross-sectional nature of the data. I strongly encourage the authors to ensure full consistency throughout the manuscript, explicitly acknowledging the impossibility of establishing directionality and considering the plausibility of bidirectional relationships.

 

Thank you for the recommendation to tailor language regarding mediation. We have made edits throughout the manuscript. Major edits include the abstract (lines 35-36), introduction (lines 64-65), discussions (lines 292-293, 484-485), and strengths and limitations (510-514).

 

Abstract

 

Across two cross-sectional studies, higher perceived stress statistically mediated the relationship between food insecurity and two forms of maladaptive eating behaviors, suggesting that perceived stress is an important correlate of these relationships. stress-related processes may represent a potential mechanism underlying

 

1. Introduction

 

These studies provide preliminary evidence that perceived stress may serve as a statistical mediator between food insecurity and diet- and eating-related outcomes (Figure 1).

 

2.3 Study 1 Discussion

 

This suggests that perceived stress represents one of several potential pathways linking food insecurity to reward-based eating. These findings indicate that perceived stress explains only a portion of the association, leaving substantial variance to be explained by other unmeasured factors. These findings should therefore be interpreted as evidence of a partial, rather than comprehensive, explanatory mechanism.

Future research should assess food insecurity, stress, and dietary outcomes over aligned time frames using longitudinal designs to provide a robust test of the proposed mediation model pathway

 

3.3 Study 2 Discussion

 

Results indicated that perceived stress partially explained the association between food insecurity and more frequent comfort eating, with a stronger effect observed among participants experiencing food insecurity compared with those who were food secure. These findings provide preliminary support that psychological stress may be a mechanism linking experiences of food insecurity with comfort eating… Notably, perceived stress was also associated with higher frequency of comfort eating among food-secure individuals, suggesting that stress is a general psychological factor that is correlated with comfort eating. influencing eating behaviors.  

 

4. Strengths and Limitations

 

Second, the findings identify highlight the potential of perceived stress as a statistical mediator through which  mediator of the association between food insecurity and eating behaviors. These results offer preliminary support for the conceptual model being tested wherein psychological factors like perceived stress, in addition to material and structural constraints, are correlated with food insecurity. Identifying support for stress as a mediator underscores the importance of psychological pathways in addition to material constraints in shaping responses to food insecurity. Despite differences in sample characteristics, measurements, and outcomes, findings from these two studies were consistent in supporting perceived stress as a mechanism linking food in-security to diet- and eating-related outcomes. This convergent evidence across distinct contexts, although preliminary, supports the role of stress-related processes implicated in food insecurity.

 

We have also explicitly acknowledged the impossibility of establishing directionality and considering the plausibility of bidirectional relationships in the strengths and limitations section (lines 521-530).

 

4. Strengths and Limitations

 

Results should be interpreted considering several conceptual limitations. First, data for both studies were cross-sectional and the directionality of these associations cannot be firmly established. thus, no causal or mechanistic claims can be made. The mediation models reported should therefore be interpreted as statistical associations among focal variables rather than evidence for causal pathways. Cross-sectional data may also obscure the possibility of bidirectional or reverse associations relationships among the focal variables. For example, higher levels of perceived stress may contribute to reward-based and comfort eating, while these eating behaviors may, in turn, influence levels of be associated with perceived stress. It is also possible that reward-based or comfort eating are associated with perceived stress in the opposite direction, reflecting a potential reverse association. Furthermore, the cross-sectional nature of the data for both studies made it difficult to fully assess the temporal ordering between the focal variables. Future research can leverage longitudinal or repeated measures designs that assess the focal variables over comparable recall periods to better establish temporal ordering, and assess potential bidirectional relationships, and probe reverse associations.

 

The justification for the dichotomisation of food insecurity has improved, and the inclusion of sensitivity analyses using ordinal measures is a positive addition. However, the implications of this analytical decision are not sufficiently interrogated. In particular, the potential loss of information, attenuation of variability, and masking of dose–response relationships should be more explicitly discussed. A more critical reflection would strengthen the methodological transparency of the study.

 

Thank you for the suggestion on how we can further improve the transparency of the study regarding our decision to dichotomize the food insecurity variable. Specifically, we have elaborated on the implications of this methodological decision as well as suggestions for future research (lines 623-628).

 

4. Strengths and Limitations

 

Additionally, food insecurity was dichotomized in both studies. This approach improved interpretability but may have reduced variability in food insecurity severity. Consequently, potential dose-response associations between food insecurity and the outcomes may have been obscured. Future research may consider refining the current conceptual model to assess how food insecurity severity may alter relationships to perceived stress, diet, and eating behaviors.

 

The issue of temporal misalignment between the measurement of food insecurity and dietary intake has been appropriately acknowledged and reasonably justified. I consider this point satisfactorily addressed.

 

Thank you for this comment and for the suggestions on how to improve the discussions of this important limitation.

 

Regarding generalisability, the authors now recognise the limitations of the sample composition. However, the discussion remains somewhat superficial. It would be important to go beyond stating limited generalisability and instead reflect on how the observed associations might differ across populations, including the possibility of variation in both magnitude and direction of effects.

 

Thank you for this comment to elaborate on this limitation. We have made edits to Study 1 Discussion section (lines 286-290).

 

2.3. Study 1 Discussion

 

Results from Study 1 should be interpreted considering the characteristics of the analytic sample. The sample for Study 1 was composed exclusively of middle-aged Black and white women from Richmond, CA, which limits the generalizability of these findings to other demographic groups. It is possible that the strength or nature of the observed associations may differ across men, individuals from other racial and ethnic backgrounds, and younger populations. These associations may vary in magnitude or direction across groups, as differences in structural conditions, resource usage, and coping resource availability may shape how food insecurity relates to eating behaviors. Future research could consider replicating the associations in diverse samples of individuals with food insecurity and assess how stress and coping to food insecurity may differ across people.

 

The statistical analysis section has been substantially expanded and now demonstrates improved rigour. The description of the Bayesian mediation approach, including imputation procedures and model estimation, is appropriate. Nevertheless, I find that the presentation remains highly technical and may not be fully accessible to readers without advanced statistical expertise. The manuscript would benefit from the inclusion of brief interpretative explanations, particularly regarding the meaning of credible intervals and indirect effects.

 

Thank you for the suggestions on how can further improve the analyses sections and have addressed both points below.

 

We have made edits to the discussion on the interpretation of credible intervals (lines 223-228).

 

2.1.4 Statistical Analyses

Whereas frequentist confidence intervals reflect the proportions of intervals that would contain the population parameter across repeated samples, the Bayesian credible interval represent the probability that the parameter lies within the interval given the observed data. Both confidence intervals and credible intervals similarly assess whether parameter estimates differ from zero.

 

We have also made edits to discuss the meaning of indirect effects (lines 215-217; lines 419-420).

 

2.1.4 Statistical Analyses

In these models, the indirect effect represented the product of the association between food insecurity and perceived stress (path a) and the association between perceived stress and the outcome (path b). In the context of this study, the indirect effect reflects the extent to which the association between food insecurity and the outcome is statistically explained by perceived stress.

 

3.1.4 Statistical Analyses

Indirect effects were computed as the product of the association between food insecurity and perceived stress (path a) and the association between perceived stress and each outcome component (path b). Like Study 1, the indirect effect calculated from these models represented the extent the association between food insecurity and comfort eating is statistically accounted by perceived stress.

 

Although the authors acknowledge that perceived stress accounts for only a small proportion of the observed associations, the discussion of alternative psychological pathways remains limited. A more comprehensive theoretical integration, including constructs such as depression, anxiety, or coping mechanisms, would provide a more nuanced interpretation of the findings.

 

We agree with the reviewer that this point can be addressed further. Although we discuss it briefly in the Study 2 Discussion section, we elaborate this limitation in the Strengths and Limitations section to acknowledge the presence of other psychological factors likely at play (lines 531-541).

 

4. Strengths and Limitations

 

Additionally, the current conceptual model focused on perceived stress as the sole psychological mediator. The current work did not assess relevant constructs like mental health symptoms or coping strategies. Meta-analytic evidence indicates that food insecurity is correlated with depression and anxiety (Arenas et al., 2019), both of which may also contribute to reward-based eating and comfort eating behaviors (Finch & Tomiyama, 2015; Muha et al., 2024). Theoretical models of stress and coping further suggest that individuals may engage in cognitive and behavioral coping responses, such as emotion regulation or engagement in social support resources, in response to stress-related experiences (Biggs et al. 2017). Future research could benefit from incorporating multiple statistical mediating or moderating factors alongside perceived stress to provide a comprehensive account of these associations.

 

In Study 2, the clarification of the comfort eating measure is appreciated. However, the limitations associated with the use of a single-item measure are not sufficiently addressed. This represents an important methodological constraint that should be explicitly discussed, particularly in terms of construct validity.

 

Thank you for this helpful comment. We have revised the manuscript to more explicitly address the limitations associated with the use of a single-item measure of comfort eating (lines 487-492).

 

3.3 Study 2 Discussion

 

Comfort eating was assessed using a single-item self-report measure, which may not fully capture the complexity of this behavior and may limit construct validity. The measure did not assess situational or emotional contexts that may influence comfort eating (52), potentially restricting the ability to capture variability in the behavior. As a result, this single-item measure may provide an incomplete representation of comfort eating. This limitation may introduce random measurement error and reduce the ability to adequately capture the construct of interest.

 

Minor Comments

 

The manuscript would benefit from further refinement of language and style. Some sections remain overly dense, and simplifying sentence structure would enhance readability. Additionally, certain parts of the discussion are still somewhat repetitive and could be streamlined.

 

Thank you for the suggestion to improve the overall language and style of the manuscript. We have made edits throughout to reduce redundancy. The Strengths and Limitations section has received major edits to improve the readability and reduce the density of the content.

 

2.3 Study 1 Discussion

 

Study 1 revealed that greater perceived stress mediated the association between food insecurity and reward-based eating. Findings from this study align with animal models showing that food insecurity may alter reward processes and, to our knowledge, yields the first evidence of a correlational link between food insecurity and reward-based eating in humans. These results also yield a statistical pattern suggesting perceived stress as a novel psychosocial mediator that may underlie this association. Results from Study 1 should be interpreted considering the characteristics of the analytic sample. This study relied on a sample The sample for Study 1 was composed exclusively of middle-aged Black and white women from Richmond, CA, which limits the generalizability of these findings to other demographic groups. It is possible that the strength or nature of the observed associations may differ across men, individuals from other racial and ethnic backgrounds, and younger populations. As such, the current findings should be interpreted within the context of this specific sample. Future research could consider replicating the associations in diverse samples of individuals with food insecurity. Lastly, although the indirect effects were statistically significant, their magnitude was small per Cohen’s effect size benchmarks. This suggests that perceived stress represents one of several potential pathways linking food insecurity to reward-based eating. These findings indicate that perceived stress explains only a portion of the association, leaving substantial variance to be explained by other unmeasured factors. These findings should therefore be interpreted as evidence of a partial, rather than comprehensive, explanatory mechanism. Future research may consider identifying additional psychological and biological mediators that underly the association between food insecurity and greater reward-based eating drive.

 

3.3. Study 2 Discussion

 

Study 2 leveraged data from a large census-matched U.S. sample to examine whether higher perceived stress mediated the association between food insecurity and the frequency of comfort eating. Results indicated that perceived stress partially explained the association between food insecurity and more frequent comfort eating, with a stronger effect observed among participants experiencing food insecurity compared with those who were food secure. These findings provide preliminary support that psychological stress may be a mechanism linking experiences of food insecurity with comfort eating. Research on the perceived benefits and frequency of comfort eating indicates that negative mood management, reduced boredom, and improved cognitive competence are associated with comfort eating frequency, but not perceptions of comfort eating as pleasurable and accompanied by positive emotions [51]. Thus, individuals who perceive greater psychological stress may be more likely to manage their stress by engaging in comfort eating. Consistent with broader theoretical models of stress and coping, perceived stress is unlikely to influence comfort eating in isolation. Other psychological processes such as depressive symptoms, food-related anxiety, and emotion regulation difficulties may interact with perceived stress, thereby contributing to comfort eating. Notably, perceived stress was also associated with higher frequency of comfort eating among food-secure individuals, suggesting that stress is a general psychological factor that is correlated with comfort eating. influencing eating behaviors.

 

4. Strengths and Limitations

 

Results should be interpreted considering several limitations. First, data for both studies were cross-sectional and thus, no causal or mechanistic claims can be made. The mediation models reported should therefore be interpreted as statistical associations among focal variables rather than evidence for causal pathways. Although sociodemographic variables like income, education level, and household size were statistically accounted for, additional unmeasured variables (e.g., food bank usage, mental health symptoms) may confound the observed associations. Cross-sectional data may also obscure bidirectional relationships among the focal variables. For example, higher levels of perceived stress may contribute to reward-based and comfort eating, while these eating behaviors may, in turn, influence levels of perceived stress. Furthermore, the cross-sectional nature of the data for both studies made it difficult to fully assess the temporal ordering between the focal variables. Measures of food insecurity, perceived stress, and dietary outcomes were assessed over different recall periods, which may weaken the temporal logic of the mediation analyses. Future research can leverage longitudinal designs that assess the focal variables over comparable recall periods to better establish temporal ordering and assess potential bidirectional relationships.

           

            Additional methodological limitations should be considered. First, Second, all focal variables were assessed via self-report questionnaires. This approach is subject to recall error and social desirability bias, particularly for sensitive constructs such as food insecurity and dietary behaviors. Future studies could use ecological momentary assessment to capture food insecurity, perceived stress, dietary intake, and eating behaviors repeatedly across the day, thereby reducing retrospective bias and improving ecological validity. Third, Study 1 relied on a sample of Black and white women, which limits generalizability to other racial and ethnic groups. Although the NGHS design permits examination of racial differences while holding socioeconomic factors equal, the absence of other racial/ethnic populations constrains the extent to which findings generalize beyond Black and white women. Given that food insecurity disproportionately affects several racial and ethnic minority groups, future studies should include more diverse samples and consider oversampling populations at elevated risk. Second, data for Study 2 were collected using an online survey. Past research has emphasized that web Web surveys are susceptible to self-selection bias and may also underrepresent as well as under coverage of individuals who do not have access to the internet [52]. Future research should replicate these findings using probability-based samples or mixed-methods data collection approaches to better capture underrepresented populations. individuals who may be underrepresented in web-based surveys. Finally, methodological limitations should be considered. The two studies used different mediation analytic approaches including a standard mediation model in Study 1 and a two-part zero-inflated mediation model in Study 2, which limits the comparability of findings indirect effects across studies. Because the zero-inflated model involves non-linear components, standardized effect size estimates are not directly comparable to those from linear mediation models. Results should be interpreted as preliminary support for a broader conceptual framework in which perceived stress operates as a mediator, rather than as evidence of differences in effect magnitude across eating-related outcomes. Future research using comparable analytic approaches across outcomes may help clarify differences in effect size for perceived stress. 

 

Tables and figures are generally clear, but their interpretability could be improved by more explicitly highlighting key findings.

 

Thank you for this helpful suggestion. We have revised the tables and figures to improve interpretability by more explicitly highlighting key findings. Specifically, we added brief interpretive statements to the notes of the mediation figures to clarify the main results in substantive terms. We also revised table notes to provide additional context for readers to more easily interpret the reported values.

 

Table 1.

 

Values represent Pearson correlation coefficients. Means, standard deviations (SD), and ranges are reported for each variable of interest. Analyses are based on non-imputed data.

 

Table 2.

 

Values represent Pearson correlation coefficients. Means, standard deviations (SD), and ranges are reported for each variable of interest. Analyses are based on non-imputed data.

 

Figure 2.

 

Results indicated that food insecurity was positively associated with perceived stress, but perceived stress was not significantly associated with diet quality. The indirect effect of perceived stress was not significant.

 

Figure 3.

 

Results indicated that food insecurity was positively associated with perceived stress, and perceived stress was significantly associated with higher reward-based eating. The indirect effect of perceived stress was significant. The magnitude of the indirect effect was small (u= 0.02) per Co-hen’s 1988 benchmarks.

 

Table 3

Table is based on non-imputed data. Values are presented as mean (SD) for continuous variables and n (%) for categorical variables. Missing data for education level (n = 227).

 

Table 4.

 

Values represent Pearson correlation coefficients. Means, standard deviations (SD), and ranges are reported for each variable of interest. Analyses are based on non-imputed data.

 

Figure 4.

Perceived stress statistically mediated the association between food insecurity and comfort eating. Higher food insecurity was associated with greater perceived stress, which in turn was associated with both a higher likelihood of any comfort eating (Panel A) and greater frequency of comfort eating among those reporting at least one day (Panel B).

 

Finally, the manuscript would benefit from a clearer articulation of practical implications, particularly in relation to public health or intervention design.

 

We appreciate this suggestion to provide a clearer articulation of practical implications. We have made edits to the conclusion section to reflect this comment (lines 436-439).

 

5. Conclusions

 

The present findings identify perceived stress as a common correlate of both food insecurity and maladaptive eating behaviors and provide preliminary, associative evidence linking food insecurity to eating behaviors through perceived stress. Given the cross-sectional nature of the data and differences across studies, these findings should be interpreted cautiously. These findings may have practical implications for public health and intervention design. Incorporating stress screening or stress management components into food assistance programs may represent a feasible and scalable approach to addressing both material and psychosocial aspects of food insecurity.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The authors did a great job in revising their paper, they met all my requests successfully. Now the paper is much improved abd suitable for publication in its present form.

Author Response

The authors did a great job in revising their paper, they met all my requests successfully. Now the paper is much improved and suitable for publication in its present form.

We thank the comments and suggestions that were made by Reviewer 3 that substantively improved the quality and rigor of the manuscript.

Author Response File: Author Response.pdf