Dear editor
We read with interest the study by Rage, which examined the relationships among predictive preparedness, adaptive governance, institutional fragmentation, and recurrent shock reduction in Somalia using partial least squares structural equation modelling (PLS-SEM) in a cross-sectional sample of 315 respondents.1 It is a study on an important policy question – can anticipatory preparedness play a role in the process of disaster-risk reduction in a context of fragmented governance? But a few points related to the interpretation of the statistics, moderated mediation, construct measurements and the conclusions that can be drawn need clarification.
One obvious worry is that there does not seem to be an agreement on how to interpret the conditional indirect effect at the highest level of institutional fragmentation. The abstract states that the effect was −0.10 (95% confidence interval [CI]: −0.18 to −0.02; p = 0.012), but it is reported in the abstract as non-significant. This same interpretation is found in the summary of hypothesis testing, while Table 6 classifies this effect as statistically significant. The interval does not contain zero and this result is reported with a p-value that is less than the conventional 0.05 value.1 The findings should thus not be termed “not significant” without any specific methodological explanation. This difference is especially relevant as it forms the foundation for the main finding in the article: that institutional fragmentation reduces the benefits from preparedness through the mechanism of governance.
Another interesting issue is the difference between a conditional indirect effect and evidence of moderated mediation. The study reports a positive indirect effect at low fragmentation (0.10; 95% CI: 0.04–0.16) and a negative effect at high fragmentation (−0.10; 95% CI: −0.18 to −0.02).1 If these estimates are correct, it appears that the direction of the indirect association may have reversed. This outcome should be given more attention before it is viewed as a diminishing of effect. No, a substantial conditional indirect effect at one value of the moderator and another at another does not prove that there is a significant difference in the conditional indirect effects between the two values. The formal test of the relationship between the moderated mediation and the moderator, typically a so-called index of moderated mediation and its confidence interval2 is required. The authors should specify whether this index was estimated and report the associated 95% bootstrap confidence interval for the index, and present relevant conditional-effect estimates and interaction plots. Also, the coding and direction of the moderator must be verified as well to be able to properly interpret the negative high-fragmentation estimate.
There’s one more concern with the measurement specification. In this article, institutional fragmentation is presented as a formative construct but the measurement-model evaluation seems to be reflective indicator-based. Formative and reflective specifications are not equivalent; they are appropriate for different types of relationships between the construct and the indicators, and can be misspecified, which can lead to incorrect interpretations of estimated relationships.3 Institutional Fragmentation: If the indicators represent institutional fragmentation, rather than a latent trait, then the rationale for the chosen specification and the assessment procedures should be explained. It would also be helpful to clarify if the measurement model was the same for all constructs and if the reported validity assessments are appropriate for each specification.
The cross-sectional, self-reported measures used in this study also warrant caution in drawing conclusions from the proposed mechanism. Predictive Preparedness, Adaptive Governance, institutional fragmentation and recurrent shock reduction were evaluated by the same respondents. Thus, the reported correlations might be due to shared measurement procedures, participants’ perceptions, and social-desirability bias.4 While the authors recognize that the information is subject to self-reported limitations and that it is cross-sectional, these are especially relevant when explaining how preparedness leads to better outcomes as a mediation process. Mediation estimates based on cross sections may not be valid indicators of processes over time.5 Furthermore, shock reduction as perceived by the population should not be interpreted as objective reduction of losses, displacement, service disruption, etc. related to a disaster. Whether or not the reported relationships are causal and reduce the impact of recurrent disasters could be determined by means of longitudinal observations and independent measures of disaster preparedness and impact.
Lastly, the purposive sample of disaster professionals, government officials, humanitarian workers and community leaders offers valuable insights from a sample of stakeholders that may not reflect all institutions and communities impacted within Somalia. The conclusions for generalisability need to be commensurate with the sampling process and the settings sampled. Further research should further build on the evidence by also assessing a more diverse range of stakeholders, multiple measurements, and independently recorded disaster outcomes.
The aforementioned concerns do not invalidate the study’s research question or the benefits of anticipatory preparedness. Instead, they emphasize the necessity of transparency in reporting the statistics and the need for sound matching between the measurement model, analysis of the statistics, and interpretation of the results. We invite the authors to make the significance statements clearer, to report the formal test of moderated mediation, and to provide an explanation for the specification of institutional fragmentation. This would enhance the interpretability of the results and enable better policy inferences on preparedness and governance in fragile contexts.
Funding Statement
No funding was received for this work.
Data Sharing Statement
Data sharing does not apply to this article as no new data were created or analysed in this study.
Author Contributions
All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.
Disclosure
The authors reported no potential conflict of interest in this communication.
References
- 1.Rage AN. Predictive preparedness and recurrent shock reduction in Somalia: the roles of adaptive governance and institutional fragmentation. Risk Manag Healthc Policy. 2026;19:629815. doi: 10.2147/RMHP.S629815 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Hayes AF. An index and test of linear moderated mediation. Multivariate Behav Res. 2015;50(1):1–3. doi: 10.1080/00273171.2014.962683 [DOI] [PubMed] [Google Scholar]
- 3.Jarvis CB, MacKenzie SB, Podsakoff PM. A critical review of construct indicators and measurement model misspecification in marketing and consumer research. J Consum Res. 2003;30(2):199–218. doi: 10.1086/376806 [DOI] [Google Scholar]
- 4.Podsakoff PM, MacKenzie SB, Lee JY, Podsakoff NP. Common method biases in behavioral research: a critical review of the literature and recommended remedies. J Appl Psychol. 2003;88(5):879–903. doi: 10.1037/0021-9010.88.5.879 [DOI] [PubMed] [Google Scholar]
- 5.Podsakoff PM, MacKenzie SB, Lee JY, Podsakoff NP. Common method biases in behavioral research: a critical review of the literature and recommended remedies. J Appl Psychol. 2003;88(5):879. doi: 10.1037/0021-9010.88.5.879 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data sharing does not apply to this article as no new data were created or analysed in this study.
