Dear Dr. Li,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please carefully address the concerns of the reviewers.

Please submit your revised manuscript by Dec 07 2025 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org . When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

• A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

We look forward to receiving your revised manuscript.

Kind regards,

Zhou Yu, PhD

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information .

3. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

Reviewer #1: No

Reviewer #2: Partly

Reviewer #3: Yes

Reviewer #4: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously? -->?>

Reviewer #1: No

Reviewer #2: No

Reviewer #3: Yes

Reviewer #4: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available??>

The PLOS Data policy

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

Reviewer #4: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

Reviewer #4: Yes

**********

Reviewer #1: This study utilizes PSCUS (2016-2020) mixed-cross-section data to explore how individual, family, and school factors influence college students' urban preferences for employment. Understanding individual choice behavior from personal, familial, and organizational institutional perspectives holds significant value. Due to concerns regarding the data used and potential methodological shortcomings, the conclusions drawn in this paper are not evaluated at this time.

1.In this dataset, the proportion of students from Double First-Class (985) and 211 universities appears significantly higher than expected. This seems inconsistent with the actual distribution of students across Chinese universities. Could this discrepancy affect related findings, such as employment city preferences?

2. In the analysis of influencing factors, what is the theoretical basis for urban preference in employment choices? Why should academic performance, family background, and school characteristics be considered?

3. The pairwise comparison approach in Table 3's multiple regression seems wrong. We recommend including all three options in the model simultaneously.

4. Is there any basis for using bundled coefficient analysis in Table 4? Is this approach reasonable? Since this result is also derived from pairwise grouping, it appears problematic.

5. Given the pairwise comparison approach in multiple regression, the results in Figure 3 also require correction.

Reviewer #2: ***According to the topic "Where Talent Flows: Trends and Determinants of Chinese Graduates' City Preferences".

It is noted that no words are related to Talent Flows in the abstract and conclusion. Please revise the conclusion to include related content to the topic.

***Please ensure that Talent Flows, The Trends, and Determinants of City Preference are included in the literature review for definitions and related details, such as what the trends are, and which factors of components of determinants of city preference. Operational definitions are necessary.

***Line 177: Determinants of city choice (influencing factors) are academic performance, family background, and university characteristics. It is required for the operation definitions of these key terms.

***Line 83: After excluding cases with missing values, the final analytical sample includes 58136 respondents. Please check if this study included respondents or papers for analysis. If this study included respondents, it is required to include an IRB or ethical approval statement.

***Please add how to select the papers in this study for analysis, such as purposive sampling, and please add selection criteria.

***Please use "researchers" instead of "we".

***Please ensure that the research questions are addressed in the discussion section.

(1) How have Chinese graduates' city preferences evolved over time?

(2) How has the relative influence of academic performance, leadership experience, 68 family background, and institutional environments changed in shaping these decisions?

Reviewer #3: Dear Authors,

Your paper presents a valuable and technically competent analysis of Chinese graduates’ city preferences using nationally representative longitudinal data. The statistical design and execution are generally sound and consistent with the study’s aims. However, to strengthen the methodological transparency and align fully with PLOS ONE’s standards for analytical rigor, I recommend minor revisions to clarify several statistical procedures and assumptions.

Below are my detailed comments, supported by specific references to your manuscript.

1. Model appropriateness and assumptions

You have correctly employed a multinomial logistic regression model to predict graduates’ employment city preferences, which is suitable for your three-category outcome variable (first-tier, second-tier, smaller cities). The model is described clearly on page 8, lines 138–146:

“We use a multinomial logistic regression model [24] to examine the dependent variable graduates' employment city preferences (first-tier, second-tier, or smaller cities).”

However, I recommend that you explicitly state that the Independence of Irrelevant Alternatives (IIA) assumption was tested or at least considered. Because city choices are likely correlated (e.g., first- and second-tier cities may share similar attractions), this assumption is important for model validity. A brief note confirming the assumption holds, or a robustness check using a nested logit model, would strengthen confidence in your estimates.

2. Model diagnostics and fit reporting

While the regression tables (Table 3, pages 12–15) are comprehensive and well formatted, you do not report any model-fit indicators such as pseudo-R², log-likelihood, AIC/BIC, or likelihood-ratio statistics. These would allow readers to assess the explanatory power and comparative adequacy of your models.

I suggest adding a short paragraph (perhaps immediately after Table 3) reporting at least one measure of overall model fit, e.g.:

“Model 1 achieved a McFadden’s pseudo-R² of 0.21, indicating acceptable explanatory power for behavioral data.”

This addition would improve the analytical transparency expected in a PLOS ONE paper.

3. Multicollinearity and robustness checks

The analysis includes several correlated predictors (e.g., academic performance, university type, and university location), which may inflate standard errors. On page 9, lines 144–146, you mention comparing “grouped effects” but do not report multicollinearity diagnostics.

Please confirm whether variance inflation factors (VIFs) were checked, or note that multicollinearity was not problematic. This can be addressed briefly in the Methods section to demonstrate statistical rigor.

4. Bundled coefficient analysis

The bundled coefficients approach is innovative and appropriate for comparing domain-level influences. However, the description on page 15, lines 206–214 would benefit from clarification about how the bundled coefficients were computed and how standard errors were derived:

“This method aggregates the estimated coefficients of related predictors into a single summary measure…”

Please specify whether you used a bootstrapping procedure or analytical aggregation to obtain standard errors for the bundled coefficients. This small addition would make the method more transparent and reproducible for future researchers.

5. Missing data handling

In page 5, lines 81–83, you write:

“After excluding cases with missing values, the final analytical sample includes 58,136 respondents.”

While this approach is reasonable given the large sample size, please clarify whether the excluded cases were missing at random (MAR) or not. If feasible, you could note that results were robust to the exclusion of missing data, or that missingness was minimal. A brief justification will reassure readers that list-wise deletion did not bias your results.

6. Temporal trends and interaction effects

You note on page 17, lines 221–230 that:

“The impact of campus performance… gradually grew stronger, particularly in differentiating choices between first-tier and second-tier cities.”

This is an insightful observation, but it is not statistically tested. You might consider including interaction terms between key predictors (e.g., academic performance × year) or at least acknowledge in the Discussion (around page 18, line 253) that temporal trends were inferred descriptively rather than through formal interaction testing. This clarification would make your interpretation more rigorous without requiring additional analyses.

7. Presentation and interpretation

The interpretation of coefficients is largely accurate, and the reporting of standard errors and significance levels (Table 3) is clear. I appreciate that you avoid over-claiming causality. However, at a few points—such as page 19, lines 268–271, where you write:

“These patterns illustrate a layered decision-making process where graduates weigh symbolic validation, pragmatic incentives, and personal resources under conditions of uncertainty.”

consider softening causal phrasing to indicate association rather than causation, e.g., “the results suggest that graduates may weigh…”.

This will align your interpretation with the correlational nature of the model.

Overall, the statistical analysis is appropriate, competently executed, and largely rigorous.

Your models, sample, and analytical framework are technically sound and consistent with behavioral decision-making research. The requested revisions are minor clarifications aimed at improving transparency and meeting PLOS ONE’s high standards for statistical reporting.

Recommended actions:

Confirm or test the IIA assumption (Methods).

Add model-fit statistics (after Table 3).

Clarify bundled coefficient computation and missing-data handling.

Reframe temporal changes as descriptive unless formally tested.

Adjust phrasing to avoid implied causality.

These adjustments will strengthen the paper’s methodological credibility without altering your results or conclusions.

Reviewer #4: The manuscripts engages an important question for the discipline regarding higher-education graduates' migration decisions. It engages with these questions in an important research site (China) with relevant data (PSCUS) which advances existing research both in the detail level of the dependent variable and by being longitudinal, albeit for a fairly short period of 5 annual waves. The main conclusions are 1) the characteristics of educational institutions matter the most, 2) graduates of non-elite universities are responsive to de-centralization policies, 3) personal traits matter, and 4) family backgrounds matter less, but still matters somewhat. Based on these 4 basic findings the authors offer policy implications for decentralization interventions.

However, the policy implications are mostly based on the second finding, which is the least compelling from a methodological perspective. It relies mostly on annual changes between the first three years in the data and the last two years, which coincide with some new policies. Therefore, the authors see these differences as evidence for the efficacy of said reforms. However, this evidence is completely circumstantial, and it is not clear that the temporal trends cannot be better explained by other factors, most notably COVID19.

Furthermore, while the first finding is very strong, it is actually so strong that it casts doubt on everything else in the manuscript. That is, what the authors describe as a choice of city destination tear, which is very much influenced by "institution's characteristics", may actually be better describes in social rather than geographical terms. Rather than choosing cities, graduates may choose status mobility destinations. University location and type are the most important independent variables by a large margin. So much so that (to this reviewer) it makes little sense to continue the analysis without examining the interactions between university location and other variables. Does university type really matter as much as the analysis suggests, or is it the relative abundance of top-tier institutions in tip tier locations? similarly, do family background, or any other independent variables, have similar associations with graduates destination wishes when they study in top tier cities as in smaller cities? To me such questions seem to drive everything in the manuscript, they could be answered easily by the data, but are ignored in the current version.

**********

what does this mean? ). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy

Reviewer #1: Yes: Shunxu Peng

Reviewer #2: No

Reviewer #3: Yes: Holly Carter

Reviewer #4: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures

You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation.

NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications.

Dear Dr. Li,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Feb 06 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org . When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

• A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

We look forward to receiving your revised manuscript.

Kind regards,

Zhou Yu, PhD

Academic Editor

PLOS One

Journal Requirements:

If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.

Additional Editor Comments:

Please address the reviewer's concerns. Thanks.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #5: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions??>

Reviewer #5: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously? -->?>

Reviewer #5: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available??>

The PLOS Data policy

Reviewer #5: No

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English??>

Reviewer #5: Yes

**********

Reviewer #5: Comments to the Authors

Summary

This manuscript uses five waves (2016–2020) of data from the Panel Study of Chinese University Students (PSCUS) to examine trends and determinants of Chinese graduates’ employment city preferences. By classifying cities into first-tier, second-tier, and smaller cities, and estimating multinomial logit models supplemented by a “bundled coefficient” approach, the authors compare the relative roles of campus performance, family background, and university characteristics. The study finds that university type and location are the strongest predictors of city preferences, while the influence of academic performance and leadership experience has become more pronounced over time, and graduates from non-elite institutions are more responsive to talent policies in second-tier cities.

The topic is timely and relevant, the dataset is large and valuable, and the paper is generally clearly written. With some important clarifications and strengthening of the methods and interpretations, this study could make a useful contribution.

Below are some major and minor comments that I hope will help improve the manuscript.

Major Comments

Comments 1: Nature of the data: panel vs. repeated cross-sections

TThe manuscript describes PSCUS as a "nationally representative longitudinal survey" and uses five waves of data, but it is not clear whether the analysis is based on true panel data (the same individuals followed over time) or on repeated cross-sections (different individuals in each wave). This distinction has direct implications for the modelling strategy, because if individuals appear in multiple waves, the independence assumption of the multinomial logit model is violated and standard errors may be underestimated.

Suggestions:

Please clarify the exact structure of the analytic sample. In particular, you may wish to: (1) state explicitly whether individuals are re-interviewed across waves and, if so, how many unique individuals versus person-wave observations are included in the final N; (2) explain how you handle any within-individual correlation, for example by using panel or mixed-effects methods for multinomial outcomes, or by clustering standard errors at the individual level; (3) if instead the data are effectively repeated cross-sections, clarify this and briefly justify why panel methods are not required.

Comment 2: Use of survey design information and national representativeness

The paper characterises PSCUS as a nationally representative survey based on stratified sampling, yet the analysis does not indicate whether survey weights, stratification, or clustering are accounted for. If such design features exist but are not incorporated, this may weaken the claim of national representativeness and could affect the estimated associations.

Suggestions:

It would be helpful to explain more clearly how the survey design is treated in the analysis. For example: (1) indicate whether PSCUS provides sampling weights and design variables, and whether these were used in the descriptive statistics and regression models; (2) if weights or design variables are available but not used, provide a brief rationale and consider softening statements about national representativeness; (3) if no weights are available, please state this explicitly and acknowledge it as a limitation.

Comment 3: Dependent variable - preferences vs. actual choices

The key outcome variable is based on the survey question "Where would you most like to work after graduation?", which captures employment preferences or intentions rather than realised employment locations. However, in several places the manuscript uses terms such as "employment destinations", "city choices", or "job outcomes", which may give the impression that the analysis concerns actual post-graduation locations rather than intended ones.

Suggestions:

To improve conceptual precision, it would be useful to: (1) consistently describe the outcome as employment city preferences or intended employment locations, and make clear early on that this reflects intentions rather than realised behaviour; (2) in the Discussion and Conclusions, soften causal language regarding "flows of talent" or "job outcomes" and frame the findings more explicitly as patterns of preference under conditions of uncertainty; (3) add a sentence or short paragraph in the limitations section noting that intentions may not fully translate into actual behaviour due to labour market constraints, family expectations, or other factors.

Comment 4: Ethics statement and use of human data

In the submission form, the ethics statement is recorded as "N/A", and IRB approval and informed consent are listed as "Not applicable". At the same time, the study analyses individual-level survey data on university students, albeit in anonymized form and as secondary data from PSCUS. PLOS ONE typically expects a clear ethics statement for studies involving human data, even when formal approval is not required or the data are fully anonymized.

Suggestions:

I would encourage you to provide a more explicit ethics description. For example: (1) clarify whether the original PSCUS data collection received ethical approval from an institutional review board or ethics committee, and how informed consent was obtained from participants; (2) if your own institution determined that secondary analysis of anonymized PSCUS data is exempt from additional IRB review, state this explicitly using wording consistent with PLOS ONE guidance, such as "The analysis was based on fully anonymized secondary data; according to the guidelines of [institution], additional ethical approval was not required."; (3) update the Ethics Statement and Methods sections so that the treatment of ethics and consent is transparent to readers.

Comment 5: Data availability and access to PSCUS

There is a small inconsistency between the submission form and the manuscript regarding data availability. The submission form indicates that all data are within the manuscript and its Supporting Information files, whereas the manuscript states that the data are available from http://www.pscus.cn/.

Since PSCUS is an external dataset that typically requires registration or permission, readers may not be able to access the raw microdata directly.

Suggestions:

It would strengthen transparency to harmonise and specify the data availability information. In particular, you may wish to: (1) ensure that the Data Availability Statement in the manuscript and the text entered in the submission system are consistent; (2) clarify what exactly is contained in the Supporting Information (for example, summary tables, derived indicators, or analysis code) and what must be obtained directly from PSCUS (for example, the raw microdata); (3) indicate whether access to PSCUS is open or subject to application or approval, and, if the raw data cannot be shared publicly, follow the PLOS ONE template for restricted third-party data by naming the data owner and explaining how qualified researchers can request access.

Comment 6: Interpretation of multinomial logit coefficients

In the Results and Discussion, some effects are described using phrases such as "increasing the likelihood by X%". For instance, academic performance is said to increase the likelihood of choosing first-tier or second-tier cities by specific percentages. However, the coefficients reported in Table 3 are log-odds, and strictly speaking exp(beta) represents an odds ratio rather than a percentage change in probability. The term "likelihood" is therefore somewhat ambiguous.

Suggestions:

To avoid possible misunderstanding, you might consider: (1) when discussing regression results, either reporting odds ratios (exp(beta)) and explicitly referring to changes in odds, for example "the odds of preferring first-tier cities are X times higher"; (2) alternatively, computing average marginal effects and then describing changes in predicted probabilities, such as "academic performance is associated with a Y percentage point higher probability of preferring first-tier cities"; (3) avoiding expressions that can be interpreted as simple percentage changes in probability unless they are directly based on predicted probabilities from the model.

Comment 7: Bundled coefficient analysis - method and justification

The attempt to group predictors into domains (campus performance, family background, university characteristics) and to compare their relative contribution is conceptually appealing, but the current description of the "bundled coefficients" approach is quite brief. It is not entirely clear how the group-level effect values reported in Table 4 are calculated, whether raw coefficients or transformed values are used, whether they are weighted, and how their standard errors are obtained. The link to Frangioni (2002) also remains somewhat indirect for readers who are unfamiliar with bundle methods in optimisation.

Suggestions:

It would be helpful to elaborate on this method in the Statistical Analysis section. Possible steps include: (1) formally defining how each group’s effect value is constructed from the underlying regression coefficients, including whether absolute values, squared coefficients, or other transformations are used; (2) explaining why this metric is appropriate for comparing "relative importance" across domains and how the associated standard errors are derived; (3) considering references to methodological work that deals directly with the relative importance of groups of predictors, such as R-squared decomposition, dominance analysis, or Shapley value methods, and positioning your approach relative to these; (4) at a minimum, presenting the bundled coefficient results as an exploratory summary of group contributions rather than a precise measure of explanatory power, and acknowledging this in the discussion of limitations.

Comment 8: Causal language around policy interventions

The manuscript links changes after 2018 to intensified "talent war" policies in second-tier cities, including housing subsidies and hukou reforms. This interpretation is plausible and certainly interesting, but in the current design there are no explicit policy variables at the city level in the models, and the evidence relies on descriptive trends over time rather than a causal identification strategy.

Suggestions:

To keep the conclusions aligned with the analytical design, you might: (1) soften statements that imply a direct causal effect of specific talent policies and instead present them as possible explanations or interpretations consistent with the observed patterns, using phrases such as "may be related to" or "is consistent with the interpretation that"; (2) briefly note that identifying the causal impact of concrete policy measures would require a different research design, for example incorporating city-level policy timing in a difference-in-differences framework, and mention this as a promising direction for future research.

Minor Comments

Comments 9: Terminology consistency

The manuscript alternates between expressions such as "employment city choices", "preferences", "performances", and "flows". Using several different terms for the same concept can be slightly confusing, especially since "performances" is more commonly associated with academic outcomes than with location preferences.

Suggestions:

I suggest standardising the terminology as much as possible. For example, you could: (1) select one primary term, such as "employment city preferences" or "intended employment locations", and use it consistently throughout the text; (2) state clearly in the early part of the paper that these preferences refer to intended post-graduation locations rather than realised job placements.

Comment 10: Table and figure labelling

In Table 4, the phrase "employment city performances" appears, which appears to be a typographical slip and may cause confusion. More generally, some readers may benefit from slightly more detailed captions.

Suggestions:

You may wish to: (1) correct "performances" to "preferences" in Table 4; (2) check that all tables and figures are referenced in the main text in numerical order; (3) ensure that each caption is self-contained, briefly indicating the meaning of key variables and units so that readers can understand the content without returning to the main text.

Comment 11: Description of recoding city tiers

Section 2.2.1 notes that response options differed slightly across survey waves and that you recoded them into three tiers, but the details of this recoding are not fully documented. For replication and substantive interpretation, readers may wish to know more about how cities were classified.

Suggestions:

It would improve transparency to add a brief explanation of the recoding. For instance: (1) provide, either in the main text or in an appendix, a list or description of which specific cities fall into the first-tier, second-tier, and smaller city categories; (2) explain how borderline or changing categories were handled when harmonising response options across waves.

Comment 12: Language and style

The English writing is generally clear and readable, but there are a few minor grammatical issues, occasional repetition, and some long sentences that could be streamlined.

Suggestions:

After substantive revisions are complete, a careful language edit would be beneficial. In particular, you may wish to: (1) correct instances such as "employment city performances" to "employment city preferences"; (2) check for consistent singular-plural agreement in phrases like "university characteristics... were/was"; (3) shorten a few very long sentences in the Discussion to improve clarity and flow.

Comment 13: Limitations section

The paper already discusses some limitations, but a few important aspects could be made more explicit to give readers a fuller sense of the study’s scope.

Suggestions:

You might consider adding short statements noting that: (1) academic performance is self-reported rather than based on administrative grade records; (2) the analysis is based on preferences rather than actual employment locations, which may diverge when students enter the labour market; (3) any constraints related to survey weights, complex sampling design, or missing data handling, as discussed in earlier comments, may affect generalisability and should be kept in mind when interpreting the findings.

Overall, I find this study interesting and potentially valuable. Addressing the points above—especially regarding the nature of the data, survey design, ethics, and the interpretation of the models—would substantially strengthen the manuscript.

**********

what does this mean? ). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy

Reviewer #5: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures

You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation.

NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications.

Dear Dr. Li,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Mar 14 2026 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org . When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

• A letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.
• A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.
• An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols . Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols .

We look forward to receiving your revised manuscript.

Kind regards,

Zhou Yu, PhD

Academic Editor

PLOS One

Journal Requirements:

1. If the reviewer comments include a recommendation to cite specific previously published works, please review and evaluate these publications to determine whether they are relevant and should be cited. There is no requirement to cite these works unless the editor has indicated otherwise.

2. Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

Additional Editor Comments :

The revisions address many of the reviewers’ technical concerns, particularly with respect to data structure and terminology standardization.

To further strengthen the manuscript, however, the theoretical framework should be more explicitly anchored in established migration literature. Clarifying how your analysis extends existing debates—and where it departs from them—will help demonstrate the paper’s contribution more clearly.

Drawing on the broader migration and human capital literature, the hypotheses could be framed through various lenses. First, a Spatial Sorting Hypothesis, grounded in human capital theory, posits that high-achieving students are disproportionately attracted to first-tier cities to maximize returns on skills, with institutional prestige functioning as a dominant labor-market signal. Second, a Compensatory Resource Hypothesis, derived from social capital theory, suggests that family background—through economic and relational resources—buffers the high risks and costs of migration to competitive urban centers, particularly for graduates without elite institutional credentials. Third, a Policy-Induced Preference Hypothesis, informed by choice architecture literature, proposes that local interventions such as hukou reforms and housing subsidies reshape perceived opportunity structures, making second-tier cities more attractive alternatives for non-elite graduates. While these merely are some of the possibilities, it is crucial to link this research to the bigger migration literature.

The manuscript’s contributions should be articulated more explicitly. First, unlike the predominantly cross-sectional literature, this study leverages five waves of longitudinal data to show how urban preferences evolve in response to macro-level shocks, including the COVID-19 pandemic and shifting labor demand—an important and novel contribution. Second, it provides systematic evidence that the “Double First-Class” university designation produces a stratified migration hierarchy in which institutional signals often outweigh individual performance in access to top-tier urban labor markets. Third, the findings demonstrate that “talent war” policies have uneven effects across graduate groups: non-elite students are significantly more responsive to local incentives, offering a more nuanced account of how policy interventions shape regional talent allocation.

More clearly foregrounding these theoretical linkages and substantive contributions would substantially increase the manuscript’s impact.

[Note: HTML markup is below. Please do not edit.]

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

To ensure your figures meet our technical requirements, please review our figure guidelines: https://journals.plos.org/plosone/s/figures

You may also use PLOS’s free figure tool, NAAS, to help you prepare publication quality figures: https://journals.plos.org/plosone/s/figures#loc-tools-for-figure-preparation.

NAAS will assess whether your figures meet our technical requirements by comparing each figure against our figure specifications.

Where talent flows: Trends and determinants of Chinese students’ city preferences

PONE-D-25-52295R3

Dear Dr. Li,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice will be generated when your article is formally accepted. Please note, if your institution has a publishing partnership with PLOS and your article meets the relevant criteria, all or part of your publication costs will be covered. Please make sure your user information is up-to-date by logging into Editorial Manager at Editorial Manager®  and clicking the ‘Update My Information' link at the top of the page. For questions related to billing, please contact billing support .

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Zhou Yu, PhD

Academic Editor

PLOS One

Additional Editor Comments (optional):

The writing still needs polishing.

Reviewers' comments: