Exploring the Contribution of Road Infrastructure and Environmental Factors to Crash Severity at Intersections in Mixed Traffic Settings

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The submitted manuscript presents a well-structured literature review, clearly defines the methodology, and adequately addresses the results. However, I would like to ask the authors to place more emphasis on environmental factors contributing to crashes. Given the title of the article, this aspect is not sufficiently  addressed to the conclusions. Nevertheless, I consider the manuscript to be of high quality with a clear scientific contribution. After addressing this point, I recommend it for publication.

Author Response

Please find all the responses to your comments in the attached file.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

        This paper presents an empirical investigation into the impact of road infrastructure and environmental factors on crash severity at intersections in mixed-traffic settings, using data from Douala and Yaounde, Cameroon. The application of both binary probit and logistic regression models offers complementary insights into key risk factors such as lane configuration, temporal variables, and midblock design. The topic is highly relevant, especially within the context of low- and middle-income countries where mixed traffic conditions prevail and infrastructure-related safety research remains limited. The study employs a substantial dataset and clearly outlines its methodological approach. However, several aspects could be further clarified and enhanced to strengthen the paper’s contribution. The following points may help the authors in refining the manuscript:

（1）The literature review covers a wide range of factors and previous modelling approaches, but it could benefit from a more critical synthesis of how existing studies: particularly those from high-income countries apply (or fail to apply) to mixed-traffic, LMIC contexts. Deeper discussion of the transferability and limitations of prior findings would better contextualize the study's novelty.

（2）While the practical implications of the findings are discussed, the specific contributions to theory, policy, and practice could be more clearly and quantitatively articulated. For instance, how might the estimated marginal effects directly translate into safety interventions or design guidelines?

（3）A conceptual or analytical framework diagram illustrating the hypothesised relationships between variables, the modelling structure, and how different factors interact could greatly help readers visualise the study’s approach and logic.

（4）The selection of binary models is well-justified, but a brief comparison with other commonly used methods in crash severity analysis (e.g., ordinal models, machine learning techniques) could better highlight the rationale behind the chosen methodology and its relative strengths/limitations.

（5）The models exhibit relatively modest goodness-of-fit (e.g., Nagelkerke R² = 0.122). The authors appropriately acknowledge this and attribute it to data imbalance and unobserved factors, but further discussion on how this affects interpretation and potential strategies for improving model performance in future work would be valuable.

Author Response

Please find all the responses to your comments in the attached file.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The article addresses a interesting road safety research topic, however the article requires improvements. I have following comments for the authors:

1. The abstract requires significant improvement. This includes correcting typo errors, enhancing the technical writing quality, and refining the content to ensure that only relevant information is included, the discussion on future research directions may not be needed in the abstract.
2. The literature review needs a major revision. Since the research focuses on low- to middle-income (developing) countries, most of the cited literature should also come from studies conducted in similar contexts. Currently, much of the referenced work pertains to developed countries. It is strongly recommended to either (1) replace or supplement the current references with studies from developing countries, or (2) divide the literature review into two sections, one discussing research from developed countries and another focusing on developing countries. This distinction is important because the factors influencing crash occurrence and injury severity can vary significantly across different geographical, temporal, and environmental contexts.
3. It is recommended to develop a simplified model by excluding all statistically insignificant parameters. Doing so will help reduce noise and improve prediction accuracy. If the authors choose to retain insignificant parameters, appropriate references should be provided to justify this modeling approach. Specifically, was there a hypothesis being tested that required keeping these parameters? Otherwise, the model should be re-estimated excluding all insignificant variables and using a advance modeling approach. Specifically, given the potential heterogeneity in the data, the authors should consider using a random parameter modeling approach (e.g., mixed logit model) to capture unobserved variations across observations. There are articles published with limited data set but utilizing advance modeling approaches. 

Author Response

Please find all the responses to your comments in the attached file.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

Authors have addressed the comments in the revised manuscript and I do not have any further comments.