Review Reports - The Cost of Perceived Overqualification: How Self-Control Depletion and Self-Leadership Shape Nurses’ Negligence Behavior

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript tests a moderated-mediation model in 428 Nigerian nurses surveyed in two waves. In the model, perceived overqualification (PO) leads to self-control depletion (SCD), which leads to negligence behaviour (NB), with self-leadership (SL) as moderator. The question is relevant. However, the statistical reporting is internally inconsistent, and the design does not support the mechanistic claims. The study, in my opinion, requires revisions beyond a traditional round of major review.

My majors concerns

Statistical inconsistencies.
- (a) Section 2.2 cites PROCESS Model 15, but Section 2.6 cites Model 59. Only Model 59 moderates the a-path tested in H5.
- (b) For H5, the text reports t = 7.545, but Table 6 reports t = 1.335.
- (c) The 95% CIs for both interactions exclude their own point estimates: PO×SL = 0.071 [0.114, 0.326]; SCD×SL = 0.068 [0.091, 0.249].
- (d) Both interactions are positive, which implies steeper slopes at high SL. Yet the conditional direct effects decline (0.452 → 0.257). Given SD(SL) = 0.877, this decline implies an interaction of about −0.11.
- (e) At mean SL, the conditional indirect effect should be about a×b = 0.415×0.128 = 0.053, not 0.117.
- (f) In Model 59 the indirect effect is quadratic in W, so a single linear index of moderated mediation is undefined. Explain how 0.033 was obtained.
- (g) Table 5 F-values are incompatible with R². For example, R² = 0.165 with one predictor and n = 428 implies F ≈ 84, not 60.2. Model 2 coefficients also cannot be reproduced from the Table 3 correlations, which imply R² ≈ 0.23 and similar PO and SCD coefficients.
- Please provide the full PROCESS output and state whether coefficients are standardized.
Design and inference.
- PO, SCD and SL were measured simultaneously at T1; only the outcome was lagged. This does not support mediation claims. Causal wording ("triggers", "cost", "shapes") should be replaced with associational language.
- Nesting of nurses within 238 hospitals (roughly 3 respondents per hospital) is ignored. Multilevel models or cluster-robust SEs are needed.
- Explain how T1 and T2 responses were matched while guaranteeing anonymity.
Measurement.
- (a) "Negligence behaviour" derives from the EVLN neglect construct and a public-sector employee scale. Its content validity for safety-relevant nursing behaviour is not demonstrated; consider established constructs such as missed nursing care.
- (b) Report the response format. If it is a 5-point scale, means of about 4.0 for SCD and NB imply widespread self-reported carelessness, which requires interpretation.
- (c) The SPOQ has 9 items; justify using 8.
- (d) The Abbreviated Self-Leadership Questionnaire is Houghton et al. (2012), not 2004. The sample item (self-talk) is a constructive-thought strategy, contradicting the stated focus on behaviour-focused strategies (lines 263–265).
- (e) SL correlates positively with PO, SCD and NB, and its main effect on NB is positive (0.327). This contradicts the buffering rationale and must be discussed.
Ethics. Approval came only from the authors' university in Northern Cyprus. Provide the approval number and evidence of Nigerian ethical approval (NHREC and/or hospital committees, as it is strange to have approval from another country).

Minor comments

Drop the Baron & Kenny causal-steps logic and rely on bootstrap inference.
Lines 251–254 are self-contradictory: depleted nurses are "driven to manage work activities effectively", yet resort to negligence.
Reference 72 duplicates reference 38. Author–year and numbered citations are mixed. Verify that refs 4 and 15 support the claims attributed to them.
Consider additional covariates: experience, unit, hospital type, staffing.
Kurtosis values are positive in the text but negative in Table 2. A t of 0.603 corresponds to p ≈ 0.55, not 0.498.
Author contributions omit data collection and analysis.

Author Response

Reviewer One

General comment

Response

We sincerely thank the Editor and the three Reviewers for carefully reading our manuscript and for their constructive, detailed comments, which substantially improved the accuracy, transparency, and theoretical clarity of the paper. We have revised the manuscript thoroughly in response to every comment.

Furthermore, we would like to clarify that our manuscript did not make causal inferences. We adopted a time-lagged approach, which is increasingly used in survey-based studies to mitigate common method bias associated with survey data. Moreover, administering survey questions at different time intervals (i.e., temporal separation) is well established in the literature. The time-lagged approach used in this study has been used by several scholars (e.g., Mehmood et al., 2023).

However, based on the reviewer’s comment, we removed the mention of cross-sectional and changed it to time-lagged to clarify the data collection approach. The changes (highlighted in yellow text) can be seen on pages 7 (Methods) and 19 (Limitations) of the revised manuscript.

Several recent studies published in high-impact-factor and reputable journals have used the same approach as in our study. Please see:

Mehmood, K., Iftikhar, Y., Khan, A. N., & Kwan, H. K. (2023). The nexus between high-involvement work practices and employees’ proactive behavior in public service organizations: a time-lagged moderated-mediation model. Psychology Research and Behavior Management, 1571-1586.

My majors concerns

Comment 1a

Statistical inconsistencies.

(a) Section 2.2 cites PROCESS Model 15, but Section 2.6 cites Model 59. Only Model 59 moderates the a-path tested in H5.

Response

We are grateful to the reviewer for this particular observation. Thank you for identifying this inconsistency. PROCESS Model 59 is the model that was used for H5–H7 (moderation analysis), and the reference to Model 15 in the Data analysis section was an error. We removed all references to Model 15.

The changes are highlighted in yellow on page 9 of the revised manuscript.

Comment 1b

(b) For H5, the text reports t = 7.545, but Table 6 reports t = 1.335.

Response

We are grateful for the reviewer's observation; we unknowingly used the data file, which included incomplete responses, for “hypotheses testing”. We would like to clarify that this was an error in the manuscript. However, with the inconsistencies identified by the reviewer, we discovered that initial data with incomplete responses was used for the “hypothesis testing”. Based on this, we rechecked and reran the analysis with the final sample size of 428. All data analysis outputs are now consistent, and we have submitted the revised raw data output files as requested by the reviewer for further verification.

The H5 results are now consistent with Table 6, and the changes are highlighted in yellow on pages 14 and 15 of the revised manuscript.

Furthermore, based on other reviewers’ requests to add additional covariates to the model, we added them from Table 1. However, the results of all covariates in the model are insignificant. They did not change the study results, so we retained only “unit/ward” in the final analysis, based on the other reviewers’ request.

Comment 1c

(c) The 95% CIs for both interactions exclude their own point estimates: PO×SL = 0.071 [0.114, 0.326]; SCD×SL = 0.068 [0.091, 0.249].

Response

We thank the reviewer for identifying the inconsistency in the manuscript. As mentioned in the previous response, we reran the hypothesis testing using Model 59; we corrected the statistical output for both interaction effects based on the Model 59 results, and we now report the confidence intervals in Table 6. We also verified that the reported confidence intervals are consistent with the corresponding coefficient estimates and standard errors.

Again, at the other reviewer's request and comment, we added unit/ward (though insignificant) as an additional covariate in the model. We kindly ask the reviewer to verify this in the revised manuscript.

In the revised manuscript, the significant interaction now reads:

“PO×SL = -0.105 [-0.206, -0.005]; SCD×SL = -0.129 [-0.239, -0.019]

The changes are highlighted in yellow on pages 14 and 15 of the revised manuscript and also in the revised Table 6.

Comment 1d

(d) Both interactions are positive, which implies steeper slopes at high SL. Yet the conditional direct effects decline (0.452 → 0.257). Given SD(SL) = 0.877, this decline implies an interaction of about −0.11.

Response

We have rerun the analysis using Model 59 (embedded in PROCESS Macro plug-in version 5.0). As the reviewer correctly noted, on re-examination of the output, the outcome (“the sign of the PO × SL coefficient had been reported incorrectly; the correct coefficient is = b=-0.105”]. The interaction coefficient, the conditional direct effects, and the interaction plot (now Figure 3) are now consistent with one another and have been regenerated from the corrected output. As mentioned before, we added the raw analysis output as a supplementary file for further verification.

Comment 1e

(e) At mean SL, the conditional indirect effect should be about a×b = 0.415×0.128 = 0.053, not 0.117.

Response

After mean-centering and rerunning the analysis, the conditional indirect effect in Model 59 at the mean of SL equals a × b, and the previously reported value (0.117) did not satisfy. The conditional indirect effects have been re-estimated as mentioned before, and the value at the mean of SL (0.089) now equals a × b from Table 6. We have stated this identity explicitly in the revised manuscript and in Table 6. We kindly ask the reviewer to verify.

Comment

(f) In Model 59, the indirect effect is quadratic in W, so a single linear index of moderated mediation is undefined. Explain how 0.033 was obtained.

Response

As the reviewer pointed out, model 15, not 59, provides the index of moderated mediation. We have removed the index of moderated mediation from the revised manuscript. We have therefore removed the previously reported index (0.033) in Table 6 (pages 15 and 16 of the revised manuscript). H7 is now evaluated by estimating the conditional indirect effects at low (−1 SD), mean, and high (+1 SD) SL and by testing the difference between the high and low conditional indirect effects with a percentile bootstrap confidence interval computed from the same 5000 bootstrap samples.

We kindly ask the reviewer to verify this in the revised manuscript.

Comment 1g

(g) Table 5 F-values are incompatible with R². For example, R² = 0.165 with one predictor and n = 428 implies F ≈ 84, not 60.2. Model 2 coefficients also cannot be reproduced from the Table 3 correlations, which imply R² ≈ 0.23 and similar PO and SCD coefficients.

Response

We would like to clarify that these inconsistencies came from using data with incomplete responses for the “hypotheses testing”. However, by rerunning the hypothesis testing with the final sample that excluded the incomplete responses that were discarded (as stated under the methods section of the manuscript), we have corrected this, and we kindly ask the reviewer to verify.

Sub comment

Please provide the full PROCESS output and state whether coefficients are standardized.

Response

As requested, we provided the complete PROCESS output for Models 4 and 59 in Supplementary File S1. All coefficients are unstandardized.

Comment

Design and inference.

PO, SCD and SL were measured simultaneously at T1; only the outcome was lagged. This does not support mediation claims. Causal wording ("triggers", "cost", "shapes") should be replaced with associational language.

Response

As requested by the reviewer, we changed all references to triggers, costs, and shapes to association.

Comment

Nesting of nurses within 238 hospitals (roughly 3 respondents per hospital) is ignored. Multilevel models or cluster-robust SEs are needed.

Response

We welcome this comment and acknowledge that nurses were recruited across multiple hospitals and that potential within-hospital dependence is a relevant consideration. However, we would like to clarify that the survey was anonymized and that the final dataset was anonymized because the study was designed and analyzed at the individual nurse level rather than the hospital level. Several studies published in reputable nursing journals have also adopted this same approach. We kindly ask the reviewer to see the similar studies below:

Aliza, K., Shaheen, S., Malik, M. J., Zulfiqar, S., Batool, S. A., Ahmad-ur-Rehman, M., & Javed, A. (2022). Linking ostracism with employee negligence behavior: A moderated mediation model. The Service Industries Journal, 42(11-12), 872-896.

Bochoridou, A., Chatziioannou, A., Gkorezis, P., & Triantou, A. (2025). Perceived overqualification and service behavior: a moderated mediation model of nurses’ silence toward patient safety and praise from patients. Journal of Health Organization and Management, 39(6), 842-860.

Furthermore, to reflect the reviewer’s comment, we have acknowledged this under the limitations section of the manuscript

Comment

Explain how T1 and T2 responses were matched while guaranteeing anonymity.

Response

To address the reviewer’s comment, we have explained how T1 and T2 responses were matched while ensuring anonymity. Specifically, we assigned a unique code identifier to respondents who completed the survey at T1. At T2, this code was used to match participants, and those who did not complete the survey at T1 were not allowed to participate at T2.

We have added this explanation to the methods section of the revised manuscript. The changes are highlighted in yellow text on page 7 of the revised manuscript.

Measurement.

Comment 3a

(a) "Negligence behaviour" derives from the EVLN neglect construct and a public-sector employee scale. Its content validity for safety-relevant nursing behaviour is not demonstrated; consider established constructs such as missed nursing care.

Response

While we welcome this specific comment, we would like to clarify that the negligence behavior items used in this study have been used in the nursing context before. Moreover, we conducted CFA analysis to ensure the items are reliable and consistent. The measurement model results (reliability and validity) were presented in the manuscript prior to this revision. We kindly ask the reviewer to reconsider this comment and see the study below:

Aliza, K., Shaheen, S., Malik, M. J., Zulfiqar, S., Batool, S. A., Ahmad-ur-Rehman, M., & Javed, A. (2022). Linking ostracism with employee negligence behavior: A moderated mediation model. The Service Industries Journal, 42(11-12), 872-896.

Furthermore, because data collection has been completed, we were unable to add a measure of missed nursing care to the present study, and it is not in the current research context.

Comment 3b

(b) Report the response format. If it is a 5-point scale, means of about 4.0 for SCD and NB imply widespread self-reported carelessness, which requires interpretation.

Response

All items were rated on a five-point Likert scale (1 = strongly disagree to 5 = strongly agree). This has been added to the “Measures” (section 2.1) of the manuscript. Moreover, as stated in section 2.3, this study adopted a well-established approach in the literature, the marker variable technique (including a theoretically unrelated construct in the survey), and the correlations (insignificant) between the marker variable and the study's construct were very low. This suggests they paid sufficient attention while completing the survey and are unlikely to have filled out the questionnaire carelessly.

Comment 3c

Response

One item was removed during CFA because its loading was well below 0.60. This is a standard practice in survey-based research. For clarity, we added a statement about the omission under the measure section of the manuscript.

The changes are highlighted in yellow on page 9 of the revised manuscript.

Comment 3d

(d) The Abbreviated Self-Leadership Questionnaire is Houghton et al. (2012), not 2004. The sample item (self-talk) is a constructive-thought strategy, contradicting the stated focus on behaviour-focused strategies (lines 263–265).

Response

We are grateful to the reviewer for this constructive comment regarding the abbreviated self-leadership questionnaire. We have corrected the citation for the abbreviated version of the questionnaire and added it to the reference list. The change can be seen on page 9 of the revised manuscript.

Furthermore, because the SL measure captures behavioral awareness and volition, task motivation, and constructive cognition, we revised the self-leadership section to state that self-leadership is examined as a global self-regulatory capability, and we removed the statement that the study focuses on behavior-focused strategies.

The changes are highlighted in yellow on page 6 of the revised manuscript.

Comment 3e

(e) SL correlates positively with PO, SCD and NB, and its main effect on NB is positive (0.327). This contradicts the buffering rationale and must be discussed.

Response

We already addressed this comment in our response to the reviewer's earlier comments. We have rerun the analysis, and the buffering effect is in the opposite direction. Again, we have attached the raw output as supplementary files.

Comment

Ethics. Approval came only from the authors' university in Northern Cyprus. Provide the approval number and evidence of Nigerian ethical approval (NHREC and/or hospital committees, as it is strange to have approval from another country).

Response

We would like to clarify that the authors are of Nigerian origin; however, our institutional affiliation is with the University of Mediterranean Karpasia. The study was conducted as part of the authors’ academic research at their affiliated university, and we obtained ethical approval from the university's Ethics Committee, which we have provided to the assistant editor. Moreover, we obtained permission from the hospitals before administering the survey.

Minor comments

Minor comment 1

Drop the Baron & Kenny causal-steps logic and rely on bootstrap inference.

Response

Based on the reviewer’s suggestion, we removed the causal-steps logic. Mediation inference is now based on the bootstrap confidence interval of the indirect effect.

Minor comment 2

Lines 251–254 are self-contradictory: depleted nurses are "driven to manage work activities effectively", yet resort to negligence.

Response

We thank the reviewer for the comment. We have revised the argument and removed the contradiction. The sentences have been rewritten: nurses experiencing insufficient psychological resources are now described as protecting their remaining resources by reducing the effort and attention they invest in work activities, so that negligence behavior may reflect a resource-conserving response rather than a deliberate choice

The changes are highlighted in yellow on page 6 of the revised manuscript.

Minor Comment 3

Reference 72 duplicates reference 38. Author–year and numbered citations are mixed. Verify that refs 4 and 15 support the claims attributed to them.

Response

We are grateful to the reviewer for the observations. (i) The duplicate reference has been removed: the citation that pointed to reference 72 now points to reference 38, and the subsequent references have been renumbered. (ii) All author–year citations have been converted to numbered citations. (iii) We verified references 4 and 15. Reference 4 reports that nursing students in Ghana perceived negligence as the most common cause of medical errors; we revised the sentence to state this precisely. Reference 15 validates the SPOQ in Italian nurses and shows that their competencies are frequently underused; the sentence citing it has been revised accordingly, and its second citation (on self-regulatory demands in nursing) has been replaced with reference 45, which addresses self-control and decision fatigue in healthcare settings.

The changes are highlighted in yellow on pages 1 and 2, and in the reference list.

Minor Comment 4

Consider additional covariates: experience, unit, hospital type, staffing.

Based on this comment, we added all additional covariates into the model. However, none of them were significant, and they did not change the results. To accommodate the reviewer’s comment, we included only unit, in addition to gender and age; yet, all covariates remained non-significant and did not change the results.

Minor comment 5

Kurtosis values are positive in the text but negative in Table 2. A t of 0.603 corresponds to p ≈ 0.55, not 0.498.

Response

We thank the reviewer for this observation. The text now reports the ranges exactly as in the measurement table (skewness −0.210 to 0.162; kurtosis −0.665 to −0.220).

Minor comment 6

Author contributions omit data collection and analysis.

Response

Based on the reviewer’s comment, we have revised the authors’ contribution section to reflect the reviewer’s comment. The changes are highlighted in yellow on page 20 of the revised manuscript.

The changes are highlighted in yellow on page 20 of the revised manuscript.

We are very grateful to the reviewer for the constructive comments that helped improve our manuscript.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript addresses a relevant and potentially valuable topic for nursing practice and patient safety. The two-wave design and the integration of Conservation of Resources theory and Social Cognitive Theory are strengths. However, several methodological, conceptual, and statistical aspects require clarification, particularly the conceptualisation and measurement of negligence behaviour; the sampling and participant flow; the consistency between the hypothesised moderated mediation model and the PROCESS model used; and some inconsistencies in the reported statistical results. Addressing these issues and strengthening the reporting and interpretation of the findings would substantially improve the manuscript.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

The manuscript is generally understandable; however, moderate language editing is recommended before publication. Careful language revision would improve readability, precision, and overall presentation.

Author Response

Reviewer Two

General comment:

The manuscript addresses an interesting and relevant topic by examining the relationships among perceived overqualification (PO), self-control depletion (SCD), self-leadership (SL), and negligence behaviour among nurses in Nigeria. The topic is potentially relevant to nursing management, workforce well-being, and patient safety, and the use of a two-wave design represents a strength compared with a single-wave cross-sectional survey. The manuscript is generally well-structured, and the theoretical framework, based on Conservation of Resources (COR) theory and Social Cognitive Theory (SCT), provides a potentially coherent basis for the proposed relationships. However, several methodological, statistical, conceptual, and reporting issues should be addressed before the manuscript can be considered for publication. In particular, clarification is needed regarding the conceptualisation and measurement of negligence behaviour, the sampling and two-wave data collection procedures, the statistical model used to test the moderated mediation hypotheses, and some inconsistencies in the reported statistical results. I therefore recommend Major Revision, as the main concerns appear potentially addressable through clarification, reanalysis where necessary, and improved reporting.

Response

We sincerely thank Reviewer 2 for this positive and constructive assessment, including the recognition of the time-lagged design and the theoretical framework. We have addressed each comment as described below.

Comment 1

The manuscript describes negligence behaviour as an important issue in nursing practice and patient safety; however, the construct as measured appears to include behaviours such as procrastination, leaving tasks unfinished, doing less than expected, careless behaviour, and reduced willingness to make additional effort. These behaviours may overlap conceptually with other constructs, including counterproductive work behaviour, withdrawal behaviour, reduced work effort, organisational cynicism, or disengagement. The authors should therefore provide a clearer conceptual definition of negligence behaviour and explain why the selected items specifically represent negligence rather than these related constructs. This clarification is particularly important because the manuscript makes strong connections between negligence behaviour and patient safety. The authors should explicitly explain how the measured construct translates into nursing negligence in clinical practice and whether the scale has previously been validated specifically in healthcare or nursing populations.

Response

We thank the reviewer for this constructive comment. We have added a paragraph to the Introduction that defines negligence behavior as a passive, low-effort pattern of conduct in which attention to, and effort on, assigned duties lapses, and that distinguishes it from counterproductive work behavior (volitional acts that harm or are intended to harm) and withdrawal (physical or psychological removal from the work role).

To our knowledge, negligence behavior has been validated in the nursing context. Please see;

Furthermore, in the Measures section, we frame negligence behavior as a general work-behavior risk factor rather than a direct measure of clinical negligence and acknowledge this as a limitation.

Comment 2:

The use of the Conservation of Resources theory and the Social Cognitive Theory is potentially appropriate; however, the theoretical integration should be strengthened. The manuscript should explain more explicitly why perceived overqualification is expected to contribute to self-control depletion, and subsequently why self-control depletion is expected to increase negligence behaviour. Similarly, the proposed protective role of self-leadership should be more clearly connected to the mechanisms described by SCT. For each hypothesis, the theoretical argument should go beyond merely describing prior empirical associations and explicitly identify the psychological mechanism linking the variables. This would make the proposed moderated mediation model more theoretically convincing.

Response

We are thankful to the reviewer for this helpful suggestion. We have improved the rationale behind H3, we draw on the conservation motive in COR theory, and we explain why vigilance-dependent nursing tasks are vulnerable to lapses. For H5–H7, we link self-leadership to the three self-regulatory subfunctions of SCT (self-observation, judgment, and self-reaction) and explain how constructive thought strategies may reduce the regulatory cost of overqualification (H5), and how self-observation, self-goal setting, and self-reward may weaken the expression of overqualification and depletion in conduct (H6 and H7).

The changes are highlighted in yellow on pages 5 and 7 of the revised manuscript.

Comment 3:

The sampling procedure requires further detail. The manuscript states that nurses working in hospitals in Lagos and Abuja were recruited, that 238 hospitals were considered, and that purposive sampling and random cluster sampling were employed. However, the exact sequence of the sampling procedures is not entirely clear. I suggest the authors specify: how the hospitals were identified; how many hospitals were initially selected; how the selected hospitals were distributed between Lagos and Abuja; how nurses were selected within each hospital; what constituted the sampling clusters; whether inclusion/exclusion criteria were applied; how nurses were contacted and invited to participate; and whether participation was voluntary and anonymous. Furthermore, since participants were recruited from multiple hospitals, with nurses potentially nested within hospitals, the authors should clarify whether clustering was accounted for in the statistical analysis. If not, the authors should provide a methodological justification and discuss this aspect as a limitation.

Response

We appreciate this comment from the reviewer. We have revised the Methods section to clarify the sampling and recruitment procedures. We purposively selected Lagos and Abuja as the study locations, and used the Nigerian Health Facility Registry (HFR) to establish the sampling frame of public and private hospitals. Hospitals were eligible if they were listed in the HFR, located in Lagos or Abuja, operated as public or private hospitals, and provided direct patient-care services involving registered nurses. We used purposive sampling to select eligible and accessible hospitals, of which 238 ultimately participated.

Individual nurses were the survey units and were recruited according to the inclusion and exclusion criteria already provided in the manuscript prior to this revision. We also removed the term “random cluster sampling” because it did not precisely characterize the recruitment procedure and now explicitly describe the procedure followed. Moreover, several similar studies in the nursing context have used similar approach as this study. Please see:

Finally, regarding potential nesting, the individual nurse was the unit of analysis, and the 428 respondents were distributed across 238 hospitals, averaging approximately 1.8 respondents per hospital. Given these very small and sparse within-hospital observations, hospital-level effects were not estimated. Moreover, the survey was anonymous, precluding retrospective estimation of intraclass correlations or cluster-adjusted estimates. However, potential within-hospital dependence cannot be completely excluded; the authors acknowledge this limitation and encourage future research to retain non-identifying hospital codes and assess clustering where warranted.

The changes to the method section of the manuscript are highlighted in yellow on pages 8 and 9 of the revised manuscript.

Comment 4:

The two-wave survey design is a strength, but the description of the participant flow requires improvement. The manuscript reports that 788 questionnaires were administered, yielding 449 usable responses, and that 428 responses were successfully matched across the two waves. However, the exact figures for each stage of data collection and the calculation of the reported response rate (54.31%) should be presented more transparently. The authors should clearly indicate: the number of participants at Time 1; the number of participants remaining at Time 2; the number of incomplete questionnaires excluded; the number of unmatched questionnaires; the final matched sample; the criteria used to match responses across the different waves; and the response/retention rate at each stage. A participant flow diagram might improve transparency.

Response

We are grateful to the reviewer for highlighting the strength of our data collection approach. This comment is similar to Reviewer 1's comment and has been addressed. As stated in the revised manuscript, we used a unique code identifier to match the administered survey. Unmatched surveys were not collected “on-site”. We discarded responses with incomplete/irregular answers.

Comment 5:

The manuscript correctly identifies the sources of the measurement instruments; however, further information is required regarding the adaptation and validation of the scales for the Nigerian nursing context. Specifically, the authors should clarify whether any items were modified from the original instruments, whether the wording was adapted to the healthcare setting, and whether a pilot study was conducted prior to the main survey. Furthermore, the authors should describe the adaptation process and provide a more robust justification for construct validity within a nursing population. The manuscript reports factor loadings, composite reliability (CR), and average variance extracted (AVE), which is appropriate. However, the assessment of discriminant validity would benefit from a clear justification for the decision to rely solely on the Fornell-Larcker criterion.

Response

We are grateful for this constructive comment from the reviewer. PO, SCD and SL were all adopted. However, in the negligence behavior measure, “organization” was changed to “hospital”. This was already stated in the supplementary file prior to this review.

Furthermore, we are grateful to the reviewer for confirming that the measurement model analysis is appropriate. As in studies published in high-impact factor journals (see below), and because we used AMOS for the CFA analysis, we used the widely regarded Fornell-Larcker criterion for the discriminant analysis.

Stanikzai, M. E., & Mittal, E. (2025). Effects on organizational performance: Mediation through employees’ performance layers and moderation of AI decision-making. Social Sciences & Humanities Open, 11, 101592.
Gli, D. D., Tweneboah-Koduah, E. Y., & Amu, P. Y. (2025). Hotel sustainability practices, social identity and customer citizenship behaviour (CCB) in an emerging country: a moderation-mediation approach. Journal of Hospitality and Tourism Horizons, 1(4), 379-396.

Comment 6:

The analysis of early and late respondents appears to rely primarily on age. The authors should explain the rationale for selecting age as the indicator of response bias and clarify which variables were compared between early and late respondents.

Response

Again, we are grateful for the reviewer's constructive comment. However, it is standard in survey-based studies to use at least one demographic variable when comparing early and late responses. While we randomly chose age, we would like to clarify that, as in the model, all demographic information listed in Table 1 yielded the same results. Therefore, choosing other demographic variables does not change the outcomes.

Comment 7:

The manuscript refers to the traditional Baron and Kenny criteria when discussing mediation. Given that the authors use a bootstrapping approach, the indirect effect and its bootstrap confidence interval should serve as the primary basis for inference regarding mediation, rather than requiring that all traditional Baron and Kenny conditions be satisfied. The authors should therefore revise the mediation analysis and its interpretation to clearly emphasise the statistical significance and confidence interval of the indirect effect. This is especially important because the absence of a statistically significant total effect does not necessarily preclude a significant indirect effect.

Response

This comment aligns with Reviewer 1's comment; we have already addressed and removed references to Baron and Kenny’s conditions. The changes are highlighted in yellow on page 15 of the revised manuscript.

Furthermore, we would also like to clarify that the total effect is significant in our study, contrary to the comment above.

Comment 8:

The manuscript states that PROCESS Model 15 was used to test the moderated mediation model. However, the hypotheses appear to propose two different moderation effects: self-leadership moderates both the relationship between perceived overqualification and self-control depletion (H5) and the relationship between self-control depletion and negligence behaviour (H6). The authors should therefore clarify whether PROCESS Model 15 is the appropriate model for testing all the proposed relationships. Please specify which variables were entered as X, M, Y, and W and indicate the interaction terms included in the analysis. If a different PROCESS model is more appropriate for the hypothesised relationships, the authors should revise the analysis accordingly and ensure that the description and interpretation of H5H7 are consistent with the statistical model used.

Response

The reference to Model 15 was an error; the actual model used for the data analysis is Model 59. This comment is also similar to Reviewer 1's comment. We have already addressed it.

Furthermore, to reflect the reviewer comment, we have revised the data analysis section of the manuscript by specifiying which variable was entered as X, M, Y and W. The changes are highlighted in yellow text on page 11 of the revised manuscript.

“Furthermore, Model 59 was selected for the moderation analysis because it specifies that self-leadership moderates the first stage of the indirect path (perceived overqualification on self-control depletion; H5), the direct path (perceived overqualification on negligence behavior; H6), and the indirect association (H7). In Model 59, perceived overqualification was entered as the independent variable (X), self-control depletion was entered as the mediator (M), self-leadership as the moderator (W), and negligence behavior as the outcome (Y)”

Comment 9:

There appear to be numerical inconsistencies between the text and the statistical tables that need to be carefully checked against the original PROCESS output. For example, the results reported for H5 include a coefficient and standard error that would not yield the t-value reported in the text. Similarly, the coefficient reported for H6 appears inconsistent with the reported confidence interval, as the point estimate falls outside the stated 95% confidence interval. errors; t values; p values; confidence intervals; interaction effects; and conditional indirect effects against the original statistical output. The discrepancies between the narrative text, tables, and statistical software output should be corrected consistently throughout the manuscript.

Response

We appreciate this comment, which helps improve the Results section of the manuscript. However, we have already addressed this comment, as it is similar to Reviewer 1's comment.

The changes are highlighted in yellow on pages 14, 15, 16 & 17 of the revised manuscript. We kindly ask the reviewer to check the aforementioned sections of the revised manuscript.

Comment 10:

The manuscript does not appear to provide a clear a priori sample size or power calculation. Given that the study tests interaction and conditional indirect effects, which generally require larger sample sizes than simple main-effect analyses, the authors should provide a justification for the sample size. If an a priori power analysis was conducted, the authors should report the software used, the statistical model, the assumed effect size, the alpha level, the desired power, and the number of predictors. If no power analysis was performed before data collection, this should be acknowledged, and the authors should consider providing a sensitivity or post hoc assessment where appropriate.

Response

Prior to this revision, our study conducted power analysis. However, the manuscript length was already 10,000 words. Therefore, we did not include the calculation in the manuscript. At the reviewer's request, we have now included the G*Power analysis in the revised manuscript, showing that our study sample size is appropriate.

The changes are highlighted in yellow on page 9 of the revised manuscript.

Comment 11:

The Discussion would also benefit from a clearer distinction between findings that are directly supported by the data and interpretations derived from COR theory and SCT.

Response

We thank the reviewer for this constructive comment. Based on this comment, we have further strengthened the findings of our study derived from COR theory and SCT. The changes are highlighted in yellow on pages 19 and 20 of the revised manuscript.

Comment 12:

The practical implications are relevant, particularly regarding nurses' self leadership and organisational strategies aimed at reducing negligence behaviour. However, some recommendations appear stronger than the empirical evidence permits. For example, the current study does not establish that selecting nurses based on self leadership would reduce negligence behaviour. It may be more appropriate to frame self-leadership as a potentially modifiable competency that could be addressed through educational, professional development, coaching, or organisational interventions, while recognising that intervention studies would be required to establish effectiveness.

Response

We are grateful for this constructive comment from the reviewer. We have removed the recommendation to use self-leadership in evaluation and selection. Self-leadership is framed as a potentially modifiable competency that could be addressed through education, continuing professional development, coaching, or organizational interventions, and we note that intervention studies are required to establish effectiveness. The other recommendations have been worded more cautiously.

Comment 13:

The Limitations section should be revised to provide a more complete account of the methodological constraints. In addition to the limitations already acknowledged by the authors, consideration should be given to: the potential non-independence of observations due to nurses being nested within hospitals; the use of purposive/cluster sampling and the resulting limitations for generalizability; possible selection bias related to nurses who were available on the days of data collection; the self-reported nature of all principal constructs; possible social desirability or recall bias; the use of an adapted negligence behaviour scale; attrition or unmatched responses between the two waves; and the distinction between temporal separation and causal inference. The authors should also revise the statement that refers to the study as cross-sectional to make it consistent with the reported two-wave design.

Response

We are grateful for the reviewer comment, and it aligns with reviewer 1's comment. We have already addressed this comment. The Limitations section has been revised to reflect the reviewer’s comment. However, we would like to clarify that negligence behavior has been validated in the nursing context. Please see the study below:

Furthermore, we have outlined the nesting of nurses within hospitals and unmeasured hospital-level factors such as staffing; purposive sampling and possible selection bias from recruiting nurses on duty; and non-response and attrition between waves. The design has been corrected from cross-sectional to time-lagged.

Comment 14:

The ethical statement requires clarification. The data were collected from nurses working in Nigeria, whereas the manuscript identifies the university of Mediterranean Karpasia as the institution that provided ethical approval. The authors should clearly explain the relationship between the approving institution and the Nigerian study setting and specify whether additional ethical approvals or permissions were obtained from the relevant Nigerian authorities, hospitals, or institutional review boards. Where applicable, the manuscript should provide the ethics approval number, date, and the institutional bodies involved. This information is particularly important for a multicentre study involving healthcare professionals recruited from hospitals.

Response

This comment is also raised by reviewer 1, and we have addressed it. Once again, we would like to clarify that the authors are of Nigerian origin; however, our institutional affiliation is with the University of Mediterranean Karpasia. The study was conducted as part of the authors’ academic research at their affiliated university, and we obtained ethical approval from the university's Ethics Committee, which we have provided to the assistant editor. Moreover, we obtained permission from the hospitals before administering the survey.

Comment 15:

The manuscript would benefit from a concise and clearly identifiable Conclusion section summarising the principal findings and their implications for nursing practice and research. At present, the manuscript moves from the Discussion to Implications and Limitations without a distinct concluding section. The conclusion should remain closely aligned with the actual findings and should avoid causal claims that cannot be supported by the study design.

Response

We are grateful for this suggestion from the reviewer. We have now added a new Section 4 (Conclusions) that summarizes the principal findings and their implications for nursing practice and research.

The changes are highlighted in yellow text on page 22 of the revised manuscript.

We are very grateful to the reviewer for the constructive comments that helped improve our manuscript.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript studies the relationship between perceived overqualification and negligence behavior, with the mediating effect of self-control depletion and the moderating role of self-leadership between self-control depletion and negligence behavior, as well as between perceived overqualification and negligence behavior among nurses. The authors supported the study primarily with Conservation of Resources theory and Social Cognitive Theory and tested the proposed relationships using survey data collected from nurses in Lagos and Abuja, Nigeria. The sample consists of 428 respondents, with perceived overqualification, self-control depletion, and self-leadership measured at Time 1 and negligence behavior measured approximately one month later at Time 2.

The topic is related to healthcare management and patient safety. Examining perceived overqualification as a potential antecedent of nurses’ negligence behavior represents an interesting extension of the overqualification literature into a healthcare setting. The attempt to integrate self-control depletion as a mediating mechanism and self-leadership as a boundary condition is also valuable. The manuscript reports support for the direct relationships and mediation, while H5 is unsupported and H6 and H7 are reported as supported.

However, after reviewing the manuscript, my concern is the inconsistency and interpretation of the PROCESS analyses that were implemented, not the topic, theoretical foundation, or overall research idea. These problems are not only presentation problems but also directly affect the conclusions related to moderation and moderated mediation. I therefore suggest that the manuscript requires major revision. he authors can improve the statistical section by following the comments below;

The PROCESS model is inconsistent between the Methods and Results sections.

The Methods part mentioned that PROCESS Model 4 is used for mediation and Model 15 for moderation. However, in the Results section explains that Model 59 used for the moderation analysis. I believe this is a methodological inconsistency because Models 15 and 59 require different conditional-process structures. The authors should classify exactly which model was estimated, prepare a diagram corresponding to that PROCESS model, and make sure that the hypotheses, equations, tables, figures, and interpretation all correspond to the same requirement.

I strongly recommend that the authors provide either the applicable PROCESS output or, at minimum, the whole table which shows the equations for the mediator and outcome models, all interaction terms, conditional effects, and bootstrap confidence intervals.

The statistics reported for H5 are also inconsistent.

In the manuscript, the relationship between perceived overqualification and self-leadership predicting self-control depletion is reported as β = 0.101, SE = 0.076, t = 7.545, p > .05, with 95% CI. And then, Table 6 also reports the same interaction coefficient as β = 0.101, SE = 0.076, with t = 1.335 and p = .183, which is mathematically consistent with the reported coefficient and SE.

The t = 7.545 value seems instead to correspond to the main effect of perceived overqualification. The manuscript should therefore be modified, and all reported statistics must be checked directly against the original PROCESS output.

Although the conclusion that H5 is not supported appears consistent with the confidence interval including zero, the reporting error raises concerns about the accuracy of the remaining moderation statistics.

The H6 interaction coefficient and its confidence interval can’t both be correct.

For H6, the manuscript reports an interaction coefficient of β = 0.071, SE = 0.020, t = 3.550, p < .001, but gives a 95% confidence interval of approximately (0.114, 0.326).

A coefficient of 0.071 cannot lie outside its own confidence interval. Based on the coefficient and standard error, the reported interval seems unlikely to be correct. This needs to be proved directly using the original PROCESS output.

Because H6 is central to the argument that self-leadership buffers the relationship between perceived overqualification and negligence, this issue must be resolved before H6 can be considered supported.

The direction of the H6 interaction also needs clarification.

The manuscript reports a positive interaction coefficient for perceived overqualification and self-leadership; It is indicating that the relationship between perceived overqualification and negligence becomes weaker as self-leadership increases 0.452 at low self-leadership, 0.355 at the mean, and 0.257 at high self-leadership. Under conventional coding, that declining would normally correspond to a negative interaction coefficient, not a positive one.

The interaction indicates a weaker PO–negligence slope at higher self-leadership. The authors therefore need to examine whether the interaction coefficient sign, coding of self-leadership, or figure has been reported incorrectly.

This is a major problem because the theoretical interpretation of self-leadership as a moderating effect depends directly on the way of this interaction.

H7 and the moderated mediation analysis need clarification and verification.

The manuscript states that the indirect effect of perceived overqualification on negligence through self-control depletion is weaker as self-leadership increases. The reported conditional indirect effects decrease from 0.145 at low self-leadership to 0.117 at the mean and 0.090 at high self-leadership, while an index of moderated mediation of 0.033, 95% CI (0.022, 0.091), is reported.

Author Response

Reviewer Three

General comment

Response

We sincerely thank Reviewer 3 for the positive assessment of the topic, the theoretical foundation, and the research idea, and for the precise statistical comments.

Comment 1

The PROCESS model is inconsistent between the Methods and Results sections.

Response

We are grateful to the reviewer for this important observation. However, Reviewers I & 2 have also spotted the same. Reviewers 1 and 2 also noted these inconsistencies, and we have addressed them. The model estimated for the moderation analyses was PROCESS Model 59; the reference to Model 15 in the Methods was an error and has been removed. Table 6 of the revised manuscript now reports the output of Model 59: all interaction terms, the covariates, the conditional direct and indirect effects, and the bootstrap confidence intervals. The complete PROCESS output is provided as Supplementary materials with the revision.

The results of the moderated analysis are highlighted in yellow on pages 16 and 17 of the revised manuscript. We kindly ask the reviewer to verify.

Comment 2

The statistics reported for H5 are also inconsistent.

We are grateful to the reviewer for this observation. These were raised by other reviewers. All have been addressed. The reporting of H5 is now consistent in Table 6 with the interpretation in section 2.6 of the revised manuscript. The changes are highlighted in the revised manuscript (pages 16 and 17).

Comment 3

The H6 interaction coefficient and its confidence interval can’t both be correct.

For H6, the manuscript reports an interaction coefficient of β = 0.071, SE = 0.020, t = 3.550, p < .001, but gives a 95% confidence interval of approximately (0.114, 0.326).

Response

We are grateful for yet another important observation from the reviewer. The conclusion regarding H6 is now based on the corrected values. The changes are highlighted on page 16 of the revised manuscript. We kindly ask the reviewer to verify.

Comment

The direction of the H6 interaction also needs clarification.

This is a major problem because the theoretical interpretation of self-leadership as a moderating effect depends directly on the way of this interaction.

Response

This comment has also been raised by the other reviewer. In the revised manuscript, the coefficient, conditional effects, and interaction plot (Figure 3) are now consistent, and we regenerated the figure from the corrected output. The changes are highlighted in yellow on pages 16, 17, and 18 of the revised manuscript.

Comment 5

H7 and the moderated mediation analysis need clarification and verification.

Response

Other reviewers have raised this comment as well, and we have addressed it. The corrected output for the conditional indirect effect is now reported on pages 16 and 17 of the revised manuscript. In addition, Model 15 (not 59) reports the index of moderated mediation. We have removed the reference to the index of moderated mediation.

We thank the Editor and the Reviewers once again for their time and valuable guidance in improving our manuscript.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have made considerable efforts to address the previous review comments, particularly by correcting statistical inconsistencies and rerunning the PROCESS analyses. Nevertheless, several fundamental concerns remain unresolved.

The operationalization of “nursing negligence” remains problematic. The four-item self-report measure captures general work disengagement and carelessness rather than clinically established nursing negligence or missed nursing care. CFA reliability and model fit do not establish clinical content validity. Consequently, the manuscript's patient-safety implications remain insufficiently supported.
Perceived overqualification, self-control depletion, and self-leadership were measured simultaneously at T1, with only negligence behavior assessed at T2. This design does not establish the temporal sequence required for mechanistic interpretation. The indirect associations should not be presented as evidence of an established psychological mechanism.
The authors corrected several previous inconsistencies, but the moderated mediation interpretation remains insufficiently justified. H5 was not supported, and differences between significant and nonsignificant conditional indirect effects do not themselves demonstrate significant moderated mediation. The bootstrap confidence interval for the contrast between conditional indirect effects should be reported explicitly.
Clustered sampling. Nurses were recruited from 238 hospitals, yet hospital-level dependence was not accounted for. The absence of hospital identifiers prevents appropriate cluster-robust inference and represents a substantial unresolved methodological limitation.
Institutional approval from the authors' university is documented, but the applicability of local Nigerian research ethics requirements and the nature of hospital-level authorizations require clearer statement.
The Conclusions contain contradictory statements regarding self-leadership moderation. Several statistical entries and interpretations also require further verification.

Although the research question is relevant, these limitations substantially affect the credibility and clinical interpretation of the findings. I cannot recommend publication in the current form.

Author Response

Reviewer One

General revision comment

Response

We sincerely thank the reviewer for the careful second reading of our manuscript and for acknowledging our efforts to correct the statistical inconsistencies and re-run the PROCESS analyses. We have taken each remaining concern seriously and have addressed them with the responses to each comment given below:

Comment 1

The operationalization of “nursing negligence” remains problematic. The four-item self-report measure captures general work disengagement and carelessness rather than clinically established nursing negligence or missed nursing care. CFA reliability and model fit do not establish clinical content validity. Consequently, the manuscript's patient-safety implications remain insufficiently supported.

Response

We thank the reviewer for this constructive comment. However, we would like to clarify that the operationalization of negligence behavior in this study captures nurses’ general work conduct (i.e., general lapses in work effort and attentiveness) rather than clinically established nursing negligence. We stated this in the first revision and emphasized it in the last revised manuscript.

As the reviewer pointed out, CFA and model fit do not establish clinical content validity. We would respectfully clarify that our study was designed to examine negligence behavior as defined in employee behavior research, that is, a passive reduction in effort and attention to one’s work duties, a construct that has been studied with this and related measures in the nursing context (please see Aliza et al., 2022).

Aliza, K., Shaheen, S., Malik, M. J., Zulfiqar, S., Batool, S. A., Ahmad-ur-Rehman, M., & Javed, A. (2022). Linking ostracism with employee negligence behavior: A moderated mediation model. The Service Industries Journal, 42(11-12), 872-896.

Our aim was to examine this general work-behavior construct among nurses, not to measure clinical or legal negligence. We accept, however, that some statements in the previous version implied a direct link to patient safety that the data cannot support. We have therefore added text to the Introduction that defines the construct as general work conduct and distinguishes it from clinical negligence. We have also removed statements implying direct patient-safety effects in the Abstract, Introduction, Discussion, Implications, and Conclusions (i.e., the entire manuscript). For instance:

The changes are highlighted in yellow on pages on page 2 of the revised manuscript:

“In the current research, negligence behavior is conceptualized as a passive, low-effort pattern of conduct in which nurses fail to perform their required job duties sufficiently (i.e., reduced work effort) and allow their effort on assigned duties to lapse or leave tasks unfinished [7]. Negligence behavior differs from several related constructs. Unlike withdrawal behavior, which involves psychological or physical removal from the job role, such as absenteeism or lateness [8]. Further, unlike counterproductive behavior, which involves actions that harm or are intended to harm [9], negligence is passive and does not require intent to harm. Moreover, negligence behavior occurs while the individual is present. Accordingly, negligence behavior is conceptualized as general work behavior that may compromise the quality of nursing care. Importantly, negligence behavior as conceptualized here is a general work behavior drawn from the employee behavior literature [1], [10], [11]. It is distinct from clinical negligence[12]. Therefore, the current research does not measure patient outcomes or clinical errors. Hence, its implications concern nurses’ general work conduct, which may be relevant to patient safety. ”

On page 3 of the revised manuscript:

“By investigating the buffering role of self-leadership, this research aims to extend the understanding of the conditions under which adverse perceived consequences of overqualification can be minimized, highlighting the potential value of self-leadership for nurses’ work conduct”.

Comment 2

Perceived overqualification, self-control depletion, and self-leadership were measured simultaneously at T1, with only negligence behavior assessed at T2. This design does not establish the temporal sequence required for mechanistic interpretation. The indirect associations should not be presented as evidence of an established psychological mechanism.

Response

While we welcome this comment from the reviewer, we would note that the one-month separation of the outcome was intended to reduce common method variance rather than to establish the ordering of the T1 and T2 variables. This is a standard approach. Accordingly, we maintain our prior stance that this is a standard approach in survey-based research. Our study is not making causal inferences, as mentioned by the reviewer. In survey-based studies, it is recommended that independent and dependent variables be separated by a time lag. This approach encourages honest answers and prevents respondents from using cues from previous answers when answering subsequent questions. This study uses the same approach. It is a standard approach in survey-based studies. We kindly ask the reviewer to see Podakoff et al . (2003).

Furthermore, several studies in the literature (mediation model with survey data) collected data at the same point and still refer to mediator variables as mediating mechanisms. Our approach was to encourage honest answers and to mitigate common method bias, as widely established in the literature.

However, we still accommodated the reviewer’s comment. We have revised the presentation of the established psychological mechanism. Specifically, “mechanism” has been changed to “intervening variable”. The changes are highlighted in the revised manuscript on page 3 and throughout the entire manuscript:

“The current study introduces self-control depletion as a potential intervening variable that could help explain how perceived overqualification is related to negligence behavior. Accordingly, the current study aims to offer new insights by examining whether self-control depletion accounts for the association between perceived overqualification and negligence behavior.”

Comment 3

The authors corrected several previous inconsistencies, but the moderated mediation interpretation remains insufficiently justified. H5 was not supported, and differences between significant and nonsignificant conditional indirect effects do not themselves demonstrate significant moderated mediation. The bootstrap confidence interval for the contrast between conditional indirect effects should be reported explicitly.

Response

We thank the reviewer for this comment. We agree that a difference between a significant and a non-significant conditional indirect effect does not, by itself, demonstrate moderated mediation. We have therefore reported the full set of evidence relevant to H7, including the contrast the reviewer requested.

First, the hypothesized moderation is directly supported at the second stage of the indirect path. Because H5 was not supported, self-leadership was expected to condition the indirect association through the link between self-control depletion and negligence behavior. In the results, the self-control depletion × self-leadership interaction was significant and negative (b = −0.129, SE = 0.056, 95% CI [−0.239, −0.019]). The association between self-control depletion and negligence behavior therefore weakens significantly as self-leadership increases, which is the process specified in H7 (consistent with our proposed H7).

Second, the conditional indirect effects follow the hypothesized pattern consistently. The indirect effect declined from 0.127 (95% CI [0.054, 0.220]) at low self-leadership to 0.089 (95% CI [0.047, 0.138]) at median self-leadership and 0.038 (95% CI [−0.019, 0.114]) at high self-leadership. At high self-leadership, the indirect association is less than one-third of its size at low self-leadership and is no longer distinguishable from zero.

Third, as the reviewer recommended, we estimated the bootstrap contrast between the conditional indirect effects at high and low self-leadership within PROCESS Model 59, using 5000 resamples. The contrast was −0.090 (BootSE = 0.062, 95% CI [−0.212, 0.030]). The contrast is negative, as hypothesized, but its confidence interval includes zero, so the difference in the overall indirect effect does not reach conventional statistical significance.

We reported the contrast in Table 6 of the revised manuscript, as the reviewer recommended. We highlighted the changes in yellow on pages 15 & 16 of the revised manuscript.

Comment 4

Nurses were recruited from 238 hospitals, yet hospital-level dependence was not accounted for. The absence of hospital identifiers prevents appropriate cluster-robust inference and represents a substantial unresolved methodological limitation.

Response

In scientific research, context matters. We would like to clarify again that our study is at the individual level, as are several other studies in the nursing literature. Our study is conducted at the individual level.

We also agreed that hospital-level dependence is a limitation and stated this explicitly in the limitations section in the last revision, reflecting the reviewer’s comment. We kindly ask the reviewer to check the limitations section of the revised manuscript; it is explicitly stated there.

“Fourth, based on the sampling approach adopted by this study, nurses were nested within 238 hospitals. The small number of respondents per hospital and unmeasured hospital-level factors, such as hospital type, workload, and staffing levels, may have influenced the results; staffing levels, in particular, were not measured and could not be controlled. Thus, a multi-level design with sufficient participants per hospital is recommended”

From page 21 of the revised manuscript.

Here are examples of recent nursing studies that have adopted a similar approach. These studies are published in reputable journals:

Liu, D., Zhou, Q., Ma, S., Zhang, S., Liu, Y., Zhang, H., ... & Chen, C. (2026). The mediating and moderating roles of surface acting and perceived organizational support in the relationship between perceived overqualification and silence behavior among young nurses: A cross-sectional study. Plos one, 21(8), e0356282.
Guo, Y.-F., Fan, J.-Y., Lam, L., Plummer, V., Cross, W., Ma, Y.-Z., Wang, Y.-F., & Jia, Y.-N. (2022). Associations between perceived overqualification, transformational leadership and burnout in nurses from intensive care units: A multicentre survey. Journal of Nursing Management, 30(7), 3330–3339. https://doi.org/10.1111/jonm.13774

Furthermore, hospital identifiers were not retained to protect the anonymity of the participating hospitals, as agreed with hospital management; therefore, multilevel models or cluster-robust standard errors cannot be estimated retrospectively.

Comment 5

Institutional approval from the authors' university is documented, but the applicability of local Nigerian research ethics requirements and the nature of hospital-level authorizations require clearer statement.

Response

We thank the reviewer for the comment. We have revised the method section of the manuscript to distinguish more clearly between institutional ethical approval and site-level administrative authorization. Ethical approval for the study protocol was obtained from the Ethics Committee of the University of Mediterranean Karpasia, the authors' affiliated institution. Before data collection, authorization to access and recruit nurses was also obtained from the management of the participating hospitals. These hospital-level authorizations constituted administrative permissions for data collection rather than separate ethical approvals.

The changes are highlighted in yellow on pages 8 & 9 of the revised manuscript.

Comment 6

The Conclusions contain contradictory statements regarding self-leadership moderation. Several statistical entries and interpretations also require further verification.

Response

We thank the reviewer for pointing this out. The contradictory sentence in the Conclusions has been corrected: it now states that self-leadership did not moderate the association between perceived overqualification and self-control depletion, but that the direct association and the association between self-control depletion and negligence behavior were weaker at higher self-leadership. We also re-verified the statistical entries and made the following corrections. (i) In Table 6, the outcome-equation coefficients for gender and age were corrected (0.001 and 0.009), so that they agree with their t-values and confidence intervals

We have revised the conclusion section of the manuscript based on the reviewer’s comment. We highlighted the changes in yellow on page 19 of the revised manuscript.

Overall assessment

Although the research question is relevant, these limitations substantially affect the credibility and clinical interpretation of the findings. I cannot recommend publication in the current form.

Response

We appreciate the reviewer’s recognition that the research question is relevant, and we are grateful for the constructive guidance throughout the review process. We have addressed each concern by narrowing our claims to what the design and measures can support.

We thank the Editor and the Reviewer once again for their time and valuable guidance aimed at improving our manuscript

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for your comprehensive and detailed response to my comments, as well as the effort put into revising the manuscript.

You have satisfactorily addressed all my points, particularly regarding the conceptual definition of negligence behaviour, the theoretical integration of COR and SCT, the clarification of the statistical model (Model 59), and the methodological details. The revisions made throughout the text have significantly strengthened the overall quality and clarity of your work.

I have no further comments or requests. Congratulations on this excellent revision.

Author Response

We are grateful to the reviewer acknowledging our efforts in revising the manuscript. We thank Reviewer 2 once again for their time and valuable guidance aimed at improving our manuscript

Reviewer 3 Report

Comments and Suggestions for Authors

I would like to thank the authors for carefully considering my previous comments and for their efforts to improve the manuscript.

The revised manuscript shows substantial improvement, particularly in clarifying the PROCESS model specification, correcting inconsistencies in statistical reporting, and improving the presentation and interpretation of the moderation and moderated mediation analyses.

I therefore accept the manuscript in its current form.

Author Response

We are grateful to the reviewer for acknowledging our efforts in revising the manuscript. We thank Reviewer 2 again for their time and valuable guidance in improving our manuscript.

Thank you for recommending acceptance of our manuscript.