Measuring Perceived Social Support After Disaster: Development and Validation of the Perceived Social Support in Disasters Scale (PSSDS-28)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript presents a well-conducted and methodologically rigorous effort to develop and validate the Perceived Social Support in Disasters Scale (PSSDS-28). The study addresses an important gap in disaster mental health research by proposing an instrument that simultaneously captures both the source and function of social support, a dimension that is not adequately represented in existing measures. The use of multiple psychometric approaches, including factor analysis, network psychometrics, item response theory, and measurement invariance testing, considerably strengthens the credibility of the proposed scale. Overall, the manuscript is well written, theoretically grounded, and makes a valuable contribution to disaster research and psychometric assessment. Nevertheless, a few issues could be addressed to further strengthen the paper and clarify the future utility of the instrument.

1. The authors appropriately acknowledge that the current scale does not fully operationalize the source-by-function matrix of social support. Given that this is a central conceptual limitation, the manuscript would benefit from a more detailed discussion of the practical consequences of this design choice. Specifically, it would be helpful to elaborate on the types of research questions for which the current hybrid structure is sufficient and those that would require a fully factorial instrument or supplementary items.

2. The absence of criterion-related validity evidence is correctly identified as a major limitation. Since the intended value of the PSSDS-28 lies partly in its ability to inform post-disaster interventions, the manuscript could further elaborate on how future studies linking the scale to outcomes such as PTSD, depression, functional recovery, and help-seeking behavior would establish its practical utility. A brief discussion of potential applications in disaster needs assessment and psychosocial service planning would also enhance the translational significance of the work.

3. The manuscript appropriately calls for cross-cultural replication and longitudinal studies. Given the likely international interest in this instrument, the authors may consider providing more specific recommendations regarding cross-cultural adaptation procedures and longitudinal implementation, including whether the factor structure and measurement invariance should be re-evaluated at different stages of disaster recovery. Such guidance would facilitate future adoption and validation efforts.

Author Response

Response to Reviewer 1 Comments

Manuscript ID: healthcare-4447903

Measuring Perceived Social Support After Disaster: Development and Validation of the Perceived Social Support in Disasters Scale (PSSDS-28)

Note on the revised files. A single revised manuscript is submitted, containing all changes made in response to all three reviewers and the Editor. Changes are marked with Word tracked changes, and each change carries a margin comment identifying the reviewer and comment number it addresses. Line numbers below refer to that file. The reference list has been re-sequenced into order of first appearance, so reference numbers differ from the previous version; this re-sequencing is a formatting operation and is not marked as a tracked change.

1. Summary

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions marked with tracked changes in the re-submitted file.

We are grateful for the reviewer’s careful and constructive assessment, and in particular for the observation that the conclusions could be strengthened. The three comments converge on a single issue: the manuscript stated its structural and evidentiary limitations accurately but did not tell readers what those limitations imply for use. Section 4.1 now specifies the research questions for which the hybrid architecture is and is not sufficient, supported by a new Table 8. Sections 4.4 and 4.5 set out the applied value of block-differentiated scoring and a concrete, falsifiable criterion-validity programme, including a power calculation and four new references. A new Section 4.6 provides a validation roadmap covering cross-cultural adaptation and longitudinal implementation, supported by a new Table 9, and answers explicitly whether structure and invariance should be re-evaluated across recovery phases. We have also acted on the comment regarding the English language. No claim has been strengthened as a result of these revisions; the additions are qualifications and usage guidance.

2. Questions for General Evaluation

Question	Reviewer’s Evaluation	Response and Revisions
Does the introduction provide sufficient background and include all relevant references?	Yes	No change requested.
Is the research design appropriate?	Yes	No change requested.
Are the methods adequately described?	Yes	No change requested.
Are the results clearly presented?	Yes	No change requested.
Are the conclusions supported by the results?	Can be improved	Addressed. The scope of the conclusions is now bounded explicitly: Section 4.1 states which question types the instrument can and cannot answer (new Table 8, line 721); Section 4.5 states what evidence would be required before predictive claims are made (line 902); Section 4.6 states the conditions under which the structure may be assumed to hold in other cultures and recovery phases (line 976, new Table 9).
Are all figures and tables clear and well-presented?	Yes	Two new tables (Tables 8 and 9) have been added in the format of the existing tables.

 

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: The authors appropriately acknowledge that the current scale does not fully operationalize the source-by-function matrix of social support. Given that this is a central conceptual limitation, the manuscript would benefit from a more detailed discussion of the practical consequences of this design choice. Specifically, it would be helpful to elaborate on the types of research questions for which the current hybrid structure is sufficient and those that would require a fully factorial instrument or supplementary items.

Response 1: We agree, and we thank the reviewer for identifying the point at which this limitation becomes actionable for readers rather than merely acknowledged. A paragraph has been added at the end of Section 4.1 (lines 704–719) reframing the hybrid architecture as a scope condition, together with a new Table 8 (line 721) listing eight question types with an explicit judgement on whether the PSSDS-28 can address each one and what would be required if it cannot. The organizing distinction is between marginal quantities—the standing of a channel or of a function in its own right, which the source-organized and function-organized blocks do estimate—and cell-level quantities, which they do not.

“The hybrid architecture is sufficient whenever the research question concerns the standing of a support channel or a support function in its own right… The architecture becomes insufficient as soon as a question requires a specific cell of the source × function matrix… Questions of this second type require either a fully factorial instrument or, more economically, a small set of supplementary items that cross the two dimensions explicitly for the cells of theoretical interest.”

Comments 2: The absence of criterion-related validity evidence is correctly identified as a major limitation. Since the intended value of the PSSDS-28 lies partly in its ability to inform post-disaster interventions, the manuscript could further elaborate on how future studies linking the scale to outcomes such as PTSD, depression, functional recovery, and help-seeking behavior would establish its practical utility. A brief discussion of potential applications in disaster needs assessment and psychosocial service planning would also enhance the translational significance of the work.

Response 2: We agree and have addressed both halves of the comment. A paragraph added to Section 4.4 (lines 804–818) works through two applied settings—post-disaster needs assessment and psychosocial service planning—and shows what block-differentiated scoring adds that a global score cannot, including use of the block-level SEM and MDC₉₅ values already reported in Section 3.7 as field decision rules. A paragraph added to Section 4.5 (lines 902–923) converts the acknowledged criterion-validity gap into a concrete, falsifiable research programme with named criterion domains, directional predictions derived from optimal-matching theory, a known-groups design, and a power calculation. Four new references were added accordingly (PCL-5 [58], PHQ-9 [59], WHODAS 2.0 [60], PTGI [61]). We have not weakened the limitation itself: the closing sentence still restricts block scores to descriptive use until the evidence exists.

“Both designs are feasible at sample sizes comparable to the present one: detecting an incremental ΔR² of .02 for the five blocks over the MSPSS with 80% power at α = .05 requires approximately 520 participants, assuming a total model R² of about .20. Until this evidence exists, block scores should be interpreted as descriptive indicators of perceived support standing rather than as predictors of clinical outcome.”

Comments 3: The manuscript appropriately calls for cross-cultural replication and longitudinal studies. Given the likely international interest in this instrument, the authors may consider providing more specific recommendations regarding cross-cultural adaptation procedures and longitudinal implementation, including whether the factor structure and measurement invariance should be re-evaluated at different stages of disaster recovery. Such guidance would facilitate future adoption and validation efforts.

Response 3: We agree and have added a new Section 4.6, “A Validation Roadmap for Prospective Users” (lines 976–1021), containing one paragraph on cross-cultural adaptation, one on longitudinal implementation, and a new Table 9 (line 1023) ordering the validation sequence by priority with the minimum design and decision criterion for each step. We answer the reviewer’s specific question explicitly rather than by implication: the factor structure and measurement invariance should be re-evaluated at different recovery phases, and the manuscript now states why. The cross-cultural paragraph goes beyond a general reference to the ITC guidelines by identifying three points specific to this instrument—re-specification of institutional referents where the response architecture differs or where the state is a party to the event, replacement and cognitive re-testing of the dated response anchor, and an independent dimensionality check before confirmatory testing—and by stating that partial scalar invariance is an acceptable and reportable outcome whereas block-level non-invariance is not.

4. Response to Comments on the Quality of English Language

Point 1: “The English could be improved to more clearly express the research.”

Response 1: We accept this. The manuscript has undergone a language revision carried out by the authors. Long multi-clause sentences joined by colons and semicolons have been divided throughout: the opening sentence of the Abstract, the instrument-comparison sentence in the Introduction (line 162), the sentences on the hybrid architecture and on the rationale for a targeted instrument, the observation-to-parameter sentence in Section 2.7, the ESEM sentence in Section 3.2, the test–retest benchmark sentence in Section 3.3, and the known-groups sentence in Section 4.4. Each edit divides a sentence without altering its meaning, and all are visible as tracked changes.

5. Additional clarifications

All changes made in response to this review are marked as tracked changes and carry a margin comment identifying the comment they address. No data were re-collected and no previously reported result has been altered in response to this review.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents the development and psychometric validation of the Perceived Social Support in Disasters Scale (PSSDS-28), a multidimensional instrument designed to assess perceived social support among disaster survivors by integrating both the source (family, social environment, institutions) and function (emotional and instrumental–informational support) of support. The study employs a rigorous scale development process involving systematic literature review, expert validation, cognitive interviews, pilot testing, exploratory and confirmatory factor analyses, measurement invariance testing, item response theory, network psychometrics, and reliability assessments using data collected from 607 survivors of the 2023 Kahramanmaraş earthquakes in Türkiye. The manuscript is well organized and addresses an important gap in disaster mental health research. The statistical analyses are comprehensive and the findings are generally convincing. However, several methodological and conceptual issues require clarification before the manuscript can be considered for publication. In particular, the authors should strengthen the justification for developing a new scale, discuss the limitations of the sampling strategy more thoroughly, provide stronger evidence for criterion validity, and improve the interpretation of some psychometric findings. Addressing the following comments will substantially improve the scientific quality and practical applicability of the manuscript.

1. The abstract can be reduced with more emphasis on important findings.
2. It might be useful to provide some basic information and connect that with the pa[er at the initial stage of writing introduction.
3. Lines from 76-97 does not cite any reference to support the statements.
4. A detailed report may be presented while doing the research literature collection.
5. How did the authors decide about the sample size in pilot study?
6. You did not discuss about the statistical significance of parameters in table 2.
7. The authors should explain how selection bias may have affected participants who remained available approximately three years after the disaster.
8. The manuscript would benefit from a comparison of PSSDS-28 with other recently developed disaster-specific psychosocial assessment tools rather than only traditional social support scales.
9. The manuscript should discuss whether the instrument is applicable across different disaster types (e.g., floods, hurricanes, pandemics, armed conflicts) beyond earthquakes.
10. Although institutional support is proposed as a unique dimension, further theoretical justification is needed regarding its distinction from instrumental support.
11. The manuscript should discuss whether the retrospective assessment of support approximately three years after the disaster may introduce recall bias.
12. The limitations may be reduced and kept either with conclusion or after conclusion.

 

Author Response

Response to Reviewer 2 Comments

Manuscript ID: healthcare-4447903

Measuring Perceived Social Support After Disaster: Development and Validation of the Perceived Social Support in Disasters Scale (PSSDS-28)

Note on the revised files. A single revised manuscript is submitted, containing all changes made in response to all three reviewers and the Editor. Changes are marked with Word tracked changes, and each change carries a margin comment identifying the reviewer and comment number it addresses. Line numbers below refer to that file. The reference list has been re-sequenced into order of first appearance, so reference numbers differ from the previous version; this re-sequencing is a formatting operation and is not marked as a tracked change.

1. Summary

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions marked with tracked changes in the re-submitted file.

We are grateful for a review that was unusually specific about where the manuscript fell short. Two of the reviewer’s evaluations—that the methods and the fit between findings and conclusions needed improvement—we accept without qualification, and they have driven the largest changes. The methods criticism was well founded: the systematic review was conducted and documented in far more detail than the manuscript conveyed, the pilot sample size was never justified, and the manuscript directed readers to search strings that were not where it said they were. All three have been corrected. The criticism of the conclusions has been addressed by adding the theoretical justification requested, by adding two limitations that were missing entirely (selection bias and disaster-type transferability), and by reorganizing the limitations so that they are stated once rather than twice. In the course of responding we also identified and corrected an inaccurate statement in the note to Table 2. The manuscript itself has been kept lean and the supporting detail placed in the Supplementary Material, to which the main text now points at each relevant location.

2. Questions for General Evaluation

Question	Reviewer’s Evaluation	Response and Revisions
Does the introduction provide sufficient background and include all relevant references?	Can be improved	Addressed. A new opening paragraph provides the event and epidemiological context (lines 54–61). The uncited passage identified in Comment 3 now carries citations (line 130). A new paragraph situates the instrument against disaster-focused instruments in adjacent domains (line 194).
Is the research design appropriate?	Can be improved	The design is unchanged, but its rationale is now reported: search strategy and screening (line 250), pilot sample size (line 295), and the basis for prioritizing the core evidence set.
Are the methods adequately described?	Must be improved	Addressed. Section 2.2 now reports the full PRISMA count chain, the exclusion taxonomy and the prioritization rubric, with the complete search strings and stage tables added to Supp. S2.1. The pilot sample size rationale has been added and an incorrect pointer to the search syntax corrected.
Are the results clearly presented?	Can be improved	The note to Table 2 contained an incorrect statement about chi-square reporting; it has been replaced, and the significance of all model chi-square values is now stated in the text (lines 475 and 492).
Are the conclusions supported by the results?	Must be improved	Addressed. Theoretical justification for the institutional block has been added (line 739); two previously missing limitations have been added (selection bias, line 870; disaster-type transferability, line 854); the Abstract now reports the block-profile finding; and the limitations are stated once, in Section 4.5, rather than repeated in the Conclusions.
Are all figures and tables clear and well-presented?	Can be improved	The note to Table 2 has been corrected. The comparison requested in Comment 8 is provided as Supp. Table S2.5 to keep the main text within scope.

 

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: The abstract can be reduced with more emphasis on important findings.

Response 1: Done. The Abstract has been reduced from 270 to 260 words, within the journal’s guidance of around 250 words for a structured abstract, and the balance shifted toward findings. Procedural detail was removed from the Methods section (record and prioritization counts, the content validity index range, the enumeration of analytic frameworks), and the space used to add the substantive result that was previously absent: that institutional and instrumental–informational support scored lowest, the finding with the most direct intervention relevance and the one a single global score would conceal (line 41).

“Institutional (M = 3.29) and instrumental–informational support (M = 2.99) scored lowest—a profile a global score would conceal.”

Comments 2: It might be useful to provide some basic information and connect that with the pa[er at the initial stage of writing introduction.

Response 2: Agreed. The Introduction previously opened on the theoretical literature without establishing the empirical setting. A new opening paragraph provides the event context and the post-event mental health burden, and connects both to the measurement problem the article addresses (lines 54–61). Two new references were added: one for the impact figures [1] and one for the epidemiological evidence [2]. Every figure given is attributable to a cited source.

“…affecting approximately 14 million people across eleven provinces. Roughly 50,800 people died in Türkiye, more than 500,000 buildings collapsed or were heavily damaged, and about three million people were still living in tents three months later [1]. Five months after the event, probable posttraumatic stress disorder was identified in 55.2% of 1,100 survivors sampled across the affected provinces [2].”

Comments 3: Lines from 76-97 does not cite any reference to support the statements.

Response 3: The reviewer is correct, and we thank them for the precision of the observation. The paragraph developing the source-by-function account carried no citations at all, and the preceding claim about institutional actors was likewise uncited. Citations have been added at three points: the institutional claim now cites the institutional-trust literature [15,16]; the structural account cites House’s provider–function distinction [17], optimal-matching theory [11], and relational regulation theory [12] (line 130); and the argument about item count and cognitive burden cites the scale-development literature [18]. All cited works were already in the reference list.

Comments 4: A detailed report may be presented while doing the research literature collection.

Response 4: Agreed, and this exposed a documentation error we are grateful to have caught. The review was conducted and archived in full, but the manuscript reported only the first and last numbers of the screening chain and directed readers to the Supplementary Material for search syntax that was not there. Section 2.2 now reports the complete chain—489 records identified (Web of Science 359; Scopus 130), 78 duplicates removed, 411 screened, 194 excluded under a prespecified five-category taxonomy, 217 eligible, 55 prioritized for extraction, of which 46 at full-text level—together with the reason for prioritizing rather than extracting all eligible records (lines 250–258). The four complete search strings, the search date, a stage-by-stage PRISMA count table, an exclusion-reason table and the prioritization rubric have been added to Supp. S2.1, and the complete set of 411 screened records has been deposited as a supplementary data file so that the retrieved evidence base can be audited directly. We kept this material in the Supplementary Material rather than the main text because the review served item generation rather than being the study’s object.

Comments 5: How did the authors decide about the sample size in pilot study?

Response 5: This was not stated and should have been. The pilot target was set by the classical minimum ratio of five respondents per item: with 31 candidate items this gives 155, which the realized pilot sample of 150 approximates. The pilot was intended for item screening and preliminary reliability estimation rather than definitive dimensionality testing, which is why the more demanding a priori estimation reported for the main study was not applied to it. Both points are now stated (line 295), with the main-study estimation retained in Supp. S4.3.

Comments 6: You did not discuss about the statistical significance of parameters in table 2.

Response 6: This comment led us to a factual error in our own manuscript, and we are grateful for it. The note to Table 2 stated that chi-square values for three of the four models "were not reported in the original output". That statement is untrue: the values appear in Table 2 itself and in Supp. Table S11.1, where all four models are significant at p < .001. The note has been replaced, and the text now states the significance explicitly while explaining why comparison rests on descriptive fit indices and BIC rather than chi-square differences (lines 475 and 492).

Comments 7: The authors should explain how selection bias may have affected participants who remained available approximately three years after the disaster.

Response 7: Agreed; this limitation was absent and has been added (lines 870–879). We have tried to do more than acknowledge it, by stating the likely direction of the bias. Survivors who remain reachable in a disaster region three years on are, on average, those with more intact local networks and better service access, so the constraint should bias interpersonal and institutional support estimates upward rather than downward. That matters for interpretation: it makes the low institutional (3.29) and instrumental–informational (2.99) means more notable rather than less. We also report what partially offsets the concern—multi-site recruitment across eight provinces, 42.2% of the sample not in permanent housing, 54.7% reporting a major loss—without claiming that it resolves it.

Comments 8: The manuscript would benefit from a comparison of PSSDS-28 with other recently developed disaster-specific psychosocial assessment tools rather than only traditional social support scales.

Response 8: We accepted this and ran a targeted supplementary search of PubMed (2016–2026) for disaster-focused psychosocial assessment instruments outside the perceived-support family. Five were identified and are now discussed: the CCRAM for community resilience [24], the WFRQ-9 for family resilience in disaster scenarios [25], the HPE-Q for household earthquake preparedness [26], the brPSCS-10 for personal social capital after flooding [27], and the DIS-HL for disaster health literacy [28]. A new paragraph in the Introduction states why none closes the present gap (line 194), and a structured comparison against the criteria used in Table 1 is provided as Supp. Table S2.5. We did not add these instruments to Table 1 itself, because that table compares instruments measuring the same construct and several of these measure community- or household-level constructs; including them would have implied a comparability that does not hold. We also report a finding from our own review that bears directly on the reviewer’s question: among the 25 disaster-context measurement studies retained in the Scale set (2015–2025), we identified no newly developed instrument for perceived social support in disaster survivors—the field consists predominantly of adaptations of general-purpose measures, most often the MSPSS [29]. The gap is therefore a property of the field rather than an artefact of our search.

Comments 9: The manuscript should discuss whether the instrument is applicable across different disaster types (e.g., floods, hurricanes, pandemics, armed conflicts) beyond earthquakes.

Response 9: The manuscript previously noted only that cross-disaster replication was needed. It now analyses transferability at the item level (lines 854–863). Items on emotional accompaniment, neighbourhood solidarity and institutional responsiveness are event-agnostic in wording; items on shelter provision and administrative or legal processes presuppose a sudden-onset event with property destruction and a formal relief architecture. We identify three transfers requiring content review rather than translation alone: epidemics, where restricted physical proximity changes the character of support delivery; armed conflict and forced displacement, where state institutions may be a threat rather than a support and the institutional block would invert in meaning; and slow-onset events such as drought, for which no dated event anchor can be specified.

Comments 10: Although institutional support is proposed as a unique dimension, further theoretical justification is needed regarding its distinction from instrumental support.

Response 10: Agreed. A dedicated paragraph has been added to Section 4.1 (lines 739–752) making both the theoretical and the empirical case. Theoretically, the distinction is one of level rather than content: in House’s typology the provider of support and the function it serves are logically independent, so institutional support is a provider category whose content is not exhausted by tangible provision. Appraisals such as being treated fairly by an agency, being recognized as eligible, or being able to trust that a service will persist are about an actor, not a resource, and the institutional-trust literature shows they carry mental-health associations not reducible to resources delivered [15,16]. Empirically, evidence previously scattered across the Results is now assembled in one place.

“The two blocks correlate at .611 in the full sample and .669 in the validation subsample, both below the .70 redundancy threshold, and their HTMT ratio is .631. Exploratory graph analysis separates them in every one of 500 bootstrap resamples… Two blocks this consistently separable are not one block measured twice.”

Comments 11: The manuscript should discuss whether the retrospective assessment of support approximately three years after the disaster may introduce recall bias.

Response 11: Agreed. This concern overlaps with a comment from another reviewer about the dual recall window, and we have addressed both. Section 2.6 now contains a paragraph explaining why the recall instruction combined a long, event-anchored window with a short window emphasizing the past month, and stating the cost of that combination (line 387). Section 4.5 now names recall bias explicitly and states its likely direction (line 965): reconstruction at 39 to 40 months is mood-congruent, so survivors in current distress may recall support as lower than it was. Because such a bias would inflate rather than attenuate observed support–distress associations, we note that the criterion-validity studies proposed in the Discussion should where possible separate the measurement occasions for predictor and outcome.

Comments 12: The limitations may be reduced and kept either with conclusion or after conclusion.

Response 12: Agreed, and this required care because two other reviewers asked for additions to the same section. We have therefore reduced repetition rather than content, and we have kept the limitations adjacent to the conclusion as suggested: Section 4.5 immediately precedes the Conclusions, and the journal template does not provide for a section after the Conclusions, so this is the position closest to the one the reviewer describes. Section 4.5 now opens by grouping the limitations into three families—structural, sampling and generalizability, and evidentiary—so that the section can be navigated (line 835), and the Conclusions no longer restate them, since they were previously listed in both places (line 1040). The Conclusions now point to Section 4.5 and retain only the forward-looking statement of what remains to be established. The substance of the limitations has not been reduced; two were in fact added in response to Comments 7 and 9.

4. Response to Comments on the Quality of English Language

Point 1: “The English is fine and does not require any improvement.”

Response 1: We thank the reviewer. Reviewer 1 suggested the English could be improved; a language revision has been carried out and the newly added passages were written to the same standard.

5. Additional clarifications

Responding to Comment 6 led us to a false statement in the note to Table 2, which we have corrected; we would not have found it without this review. No data were re-collected and no previously reported result has been altered. The changes are one corrected statement of fact, additional reporting of procedures already carried out, one supplementary literature search, and a set of clarifications and reorganizations, each marked as a tracked change and annotated in the margin.

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for the opportunity to review this timely and relevant manuscript. The paper is well written, well structured, and easy to follow. The methodology is rigorous and comprehensive, including a systematic review, expert consultation, Delphi rounds, cognitive interviews, pilot testing, and extensive psychometric evaluation. I particularly appreciated the transparency and level of detail with which the authors acknowledged the study limitations.

I have only a few comments that I believe could further improve conceptual clarity and facilitate the appropriate interpretation and future use of the scale.

Conceptual aspects

• The distinction between perceived and received social support could be further clarified. The authors define the construct as a "subjective retrospective appraisal of the support received and perceived as available", while the items focus primarily on actual post-disaster experiences. Since perceived availability and retrospectively perceived receipt of support are related but distinct constructs, I suggest clarifying earlier and more consistently what the PSSDS-28 specifically measures and providing a clearer operational definition of the "perceived–received interface."
• Data were collected approximately 39–40 months after the earthquake, while participants were asked to consider the period "since the 6 February 2023 earthquakes, with particular emphasis on the past month." This combines two substantially different recall periods. The authors should clarify how this may affect the interpretation of responses and whether the instrument is intended for use across all post-disaster phases or specifically during long-term recovery.

Methodological aspect

• Further clarification is needed regarding item 31. The Discussion recommends retaining this item as a standalone indicator rather than including it in the IIS composite score, whereas the validation analyses appear to include it within the seven-item IIS block. The authors should clarify the final scoring approach. If item 31 is excluded from the IIS composite score, the psychometric properties of the resulting six-item IIS score should ideally be reported.

Terminological and textual aspects

• In the Abstract, the expression "defensible assessment tool" might be reconsidered, given that criterion validity, cross-cultural validation, and cross-disaster replication remain to be established. A formulation such as "provides promising initial psychometric evidence" may better reflect the current stage of validation.
• The terms blocks, dimensions, factors, and subscales appear to be used somewhat interchangeably. Consistent terminology throughout the manuscript would improve clarity.

Author Response

Response to Reviewer 3 Comments

Manuscript ID: healthcare-4447903

Measuring Perceived Social Support After Disaster: Development and Validation of the Perceived Social Support in Disasters Scale (PSSDS-28)

Note on the revised files. A single revised manuscript is submitted, containing all changes made in response to all three reviewers and the Editor. Changes are marked with Word tracked changes, and each change carries a margin comment identifying the reviewer and comment number it addresses. Line numbers below refer to that file. The reference list has been re-sequenced into order of first appearance, so reference numbers differ from the previous version; this re-sequencing is a formatting operation and is not marked as a tracked change.

1. Summary

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions marked with tracked changes in the re-submitted file.

We are grateful for a review that was both generous and exacting. Three of the five comments identified genuine defects rather than matters of preference, and we have treated them as such. The construct definition was distributed across the article instead of being stated once and applied consistently; it is now stated in the first paragraph of the Introduction in a measures-X-not-Y form and repeated verbatim in the Methods. The dual recall window was never justified or bounded; the Methods now explain the design decision, state its cost, and declare the recovery phase for which the instrument is calibrated. Most importantly, the Discussion recommended a scoring rule for item 31 that the reported psychometrics did not support. We resolved this by running the sensitivity analysis the reviewer asked for, reporting it as a new Table 5, and stating a single unambiguous scoring rule. The analysis shows that removing item 31 makes the block worse on every criterion examined, which is the opposite of what the original wording implied; we corrected the wording accordingly rather than defending it. We have also softened the Abstract and Conclusions as suggested, and standardized the terminology.

2. Questions for General Evaluation

Question	Reviewer’s Evaluation	Response and Revisions
Does the introduction provide sufficient background and include all relevant references?	Yes	An operational definition of the construct has been added to the first paragraph (lines 80–90).
Is the research design appropriate?	Yes	No change requested.
Are the methods adequately described?	Can be improved	Addressed. Section 2.6 now contains a “Recall window” paragraph justifying the dual anchor and declaring the recovery phase for which the instrument is calibrated (line 387). Section 2.4 defines the terminology used throughout (line 327). Section 2.2 repeats the construct definition verbatim (line 272).
Are the results clearly presented?	Yes	A new Table 5 and an accompanying paragraph report the item 31 sensitivity analysis (lines 562–579). An inaccurate sentence about alpha-if-deleted has been corrected (line 530).
Are the conclusions supported by the results?	Yes	The Abstract and Conclusions have been re-worded to match the current stage of validation.
Are all figures and tables clear and well-presented?	Yes	One table added (Table 5).

 

3. Point-by-point response to Comments and Suggestions for Authors

Conceptual aspects

Comments 1: The distinction between perceived and received social support could be further clarified. The authors define the construct as a "subjective retrospective appraisal of the support received and perceived as available", while the items focus primarily on actual post-disaster experiences. Since perceived availability and retrospectively perceived receipt of support are related but distinct constructs, I suggest clarifying earlier and more consistently what the PSSDS-28 specifically measures and providing a clearer operational definition of the "perceived–received interface."

Response 1: We accept this and have restructured the presentation of the construct. The definition previously appeared in three places—the Introduction, the construct specification in Section 2.2, and the conceptual limitation in Section 4.5—in three different formulations, and none stated plainly what the instrument does not measure. An explicit operational statement has been added to the end of the first paragraph of the Introduction, in a measures-X-not-Y form, and the term perceived–received interface is now defined where it first appears rather than in the Discussion (lines 80–90). Section 2.2 states that this is the same definition and that the term carries no other meaning in the article (line 272). We also accept the substantive point that the items sit closer to the received pole than the label "perceived" alone would suggest, and the new text says so directly.

“The PSSDS-28 measures a survivor’s retrospective appraisal of the support they experienced as having been available and provided since a specified disaster… It does not measure prospective availability—the belief that support would be forthcoming should a future need arise—which is the target of instruments such as the MSPSS [7].”

Comments 2: Data were collected approximately 39–40 months after the earthquake, while participants were asked to consider the period "since the 6 February 2023 earthquakes, with particular emphasis on the past month." This combines two substantially different recall periods. The authors should clarify how this may affect the interpretation of responses and whether the instrument is intended for use across all post-disaster phases or specifically during long-term recovery.

Response 2: We agree that this was left unexplained. A new paragraph in Section 2.6 (lines 387–405) sets out why each window was used, what the combination costs, and what the instrument is calibrated for. The long, event-anchored window preserves the disaster-specific framing without which respondents would appraise support in general; the short window fixes the appraisal to a current, retrievable state, which the three-week retest design requires and which limits the reconstruction demand at 39–40 months. We state the cost plainly—respondents may have weighted the two windows differently from one another, adding measurement error the design cannot quantify—and we do not overclaim from the cognitive interviews, which establish comprehension but not identical weighting. On the reviewer’s direct question we now give a direct answer: the PSSDS-28 is calibrated for the long-term recovery phase, approximately two to four years post-event; use in the acute and intermediate phases is plausible on content grounds but is not established here. The same boundary condition has been added to Section 4.5 (line 960), and Section 4.6 specifies what phase-wise re-examination would require.

Methodological aspect

Comments 3: Further clarification is needed regarding item 31. The Discussion recommends retaining this item as a standalone indicator rather than including it in the IIS composite score, whereas the validation analyses appear to include it within the seven-item IIS block. The authors should clarify the final scoring approach. If item 31 is excluded from the IIS composite score, the psychometric properties of the resulting six-item IIS score should ideally be reported.

Response 3: The reviewer identified a real contradiction. Every psychometric property reported for the IIS block—internal consistency, composite reliability, AVE, test–retest stability, factor loadings, IRT parameters and block means—was estimated on the seven-item block including item 31, while the Discussion recommended a six-item composite for applied use. We were therefore recommending a scoring rule for which we had presented no evidence. First, we ran the analysis requested: a new paragraph and new Table 5 in Section 3.3 (lines 562–579) compare the seven-item and six-item versions on internal consistency, temporal stability, measurement precision, discriminant separation and score level. The result runs against our original recommendation: removing item 31 lowers ordinal alpha (.900 → .893) and omega total (.901 → .893), lowers test–retest stability (ICC = .892 → .870), widens the standard error of measurement and the minimal detectable change (SEM .251 → .291; MDC₉₅ .696 → .807), leaves discriminant separation from the institutional block essentially unchanged (HTMT .631 → .634), and raises the block mean by 0.09. The last effect is the most consequential: item 31 has the highest floor rate in the instrument (38.35%), so excluding it removes the clearest signal of unmet need from the composite. Second, Section 4.5 (line 936) now states a single scoring rule: item 31 contributes to the IIS composite in all applications, and standalone reporting is an addition to the composite rather than a substitute for it. The full analysis, with a standalone psychometric profile of item 31, has been added to the Supplementary Material as Section S18.6. In the course of this work we also corrected an inaccurate sentence in Section 3.3 (line 530): the manuscript had stated that no item’s alpha-if-deleted exceeded its block’s alpha, whereas Supplementary Table S9.2 records that item 31 marginally does so, by +.0005.

Terminological and textual aspects

Comments 4: In the Abstract, the expression "defensible assessment tool" might be reconsidered, given that criterion validity, cross-cultural validation, and cross-disaster replication remain to be established. A formulation such as "provides promising initial psychometric evidence" may better reflect the current stage of validation.

Response 4: We accept this without reservation and have adopted the reviewer’s own formulation. The Abstract conclusion now reads “provides promising initial psychometric evidence for a multidimensional post-disaster assessment instrument…” and adds an explicit sentence naming the three forms of evidence that remain outstanding (line 45). The opening sentence of the Conclusions, which contained the parallel phrase “an initial, psychometrically defensible tool”, has been revised in the same terms (line 1031).

Comments 5: The terms blocks, dimensions, factors, and subscales appear to be used somewhat interchangeably. Consistent terminology throughout the manuscript would improve clarity.

Response 5: Agreed. We audited every occurrence of the four terms and adopted a single convention, stated explicitly in Section 2.4 where the five units are first enumerated (line 327): block denotes one of the five scorable units of the PSSDS-28; factor denotes the corresponding latent variable in a factor-analytic model; dimension is reserved for the two theoretical organizing axes, source and function; and subscale is not used as a synonym for any of them. Retaining block as the primary term keeps the article consistent with its own tables and preserves factor for the statistical sense in which it is unavoidable. The audit found eight passages that departed from this convention; all have been corrected.

4. Response to Comments on the Quality of English Language

Point 1: “The English is fine and does not require any improvement.”

Response 1: We thank the reviewer for this assessment. Reviewer 1 suggested that the English could be improved, and a language revision has been carried out; the newly added passages were written to the same standard.

5. Additional clarifications

The six-item figures in Table 5 were produced from the archived dataset using the same estimation choices as the published analyses. As a check on the procedure, every seven-item value was re-estimated before the six-item values were computed, and the re-estimated values reproduced those already published to the third decimal place. The analysis script has been deposited in the OSF repository alongside the existing scripts. No data were re-collected and no previously reported result has been altered.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

It is fine for me now