Multilevel Mixed-Effects Models to Identify Contributing Factors on Freight Vehicle Crash Severity
Round 1
Reviewer 1 Report
My comments are as follows:
1. This paper is based on the collision accident data, but lacks some overall proportion analysis, such as the proportion of small trucks in the whole road? Such analysis will make the data more objective.
2. The innovation needs to be further refined. The main differences between this paper and the existing research need to be reflected.
Author Response
The authors are thankful to the reviewer for their valuable comments that helped in revising and significantly improving the paper. Each comment/suggestion from the reviewer have been put in Italics and is followed by our response and action to revise the paper accordingly.
Comment 1
- This paper is based on the collision accident data, but lacks some overall proportion analysis, such as the proportion of small trucks in the whole road? Such analysis will make the data more objective..
Response 1
- Thank you for the comment. Statistics for all variables are reflected in table 1. The vehicle size in table 1 represents the size of the truck. This part has been clearly modified according to the reviewer's comment.
Comment 2
- The innovation needs to be further refined. The main differences between this paper and the existing research need to be reflected.
Response
- We reinforced the difference from the literature review. The content corresponds to lines 131 to 135.
“This study aims to construct a truck crash severity model using various models as well as the models used in previous studies. In addition to the ordered model that is basically used, various methodologies such as multinomial, random-effects, mixed-effects, and multilevel are used. In addition, evaluate the model constructed through performance comparison between these models.”
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The study aims to identify the contributing factors to freight vehicle crash severities. Unfortunately, the paper seems to me not to be publishable in its present state. Here are some specific comments below:
Abstract
From the statement below, is it the data that is expected to be used to enhance the safety or the findings of this study?
"Based on the results of this study, this data is expected to be used to enhance the safety of freight vehicles in expressways sections."
Introduction
The introduction is too shallow and poorly referenced. The reasons and advantages of using several models in this study were not justified.
Literature
Many parts of this section 2.1 are illogical and poorly written. Similarly, section 2.2 is grossly incoherent.
Methodology
In the abstracts, the authors claimed that they used 10 models which were not substantiated and justified in the introduction. And there is no pointer to those models in the literature review, with the relevance of these models to this present study.
Too many long sentences, for instance, lines130 to 133, the seems to be too long. There are too many unclear statements in this section, for example, in lines 134 - 135 " To improve these limitations, if the dependent 134 variable has an order, an ordered model is used.". Also, in line 138 what do you mean by "wibble distribution"?
The models were not discussed to capture the essence of deploying them for this study. Section 3.5, lines 205 to 208 should be properly referenced.
Results and Discussion
This section is too shallow and disappointing. The results are presented in the tables with little discussion on the findings with no depth in terms of arguments and interrogations of the findings with the relevant literature.
Author Response
The authors are thankful to the reviewer for their valuable comments that helped in revising and significantly improving the paper. Each comment/suggestion from the reviewer have been put in Italics and is followed by our response and action to revise the paper accordingly.
The study aims to identify the contributing factors to freight vehicle crash severities. Unfortunately, the paper seems to me not to be publishable in its present state. Here are some specific comments below:
Comment 1
Abstract
From the statement below, is it the data that is expected to be used to enhance the safety or the findings of this study?
"Based on the results of this study, this data is expected to be used to enhance the safety of freight vehicles in expressways sections."
Response
- Thank you for the good question. The results of the study were intended to show that the safety of freight vehicle on expressways was improved. The part was modified to clarify the content. The content corresponds to lines 21 to 23.
“It is expected that it can contribute to improving the safety of freight vehicles in the expressway section by utilizing factors related to the severity of crashes derived from this study.”
Comment 2
Introduction
The introduction is too shallow and poorly referenced. The reasons and advantages of using several models in this study were not justified.
Response
- The reviewer made a good point. The contents of the introduction were reinforced, and the reason for using the model was added to line 47-49.
“For analysis, based on ordered model, a model was developed that combines various additional methods such as multinomial, random parameter, mixed-effect, and multilevel.”
Comment 3
Literature
Many parts of this section 2.1 are illogical and poorly written. Similarly, section 2.2 is grossly incoherent.
Response
- Thank you for the good point. The contents were supplemented with the points you pointed out, and related contents were additionally written on lines 88-95 and lines 131-135.
“As a result of analyzing existing crash severity-related studies, it was confirmed that factors such as time, age, road alignment, and speed had a great influence on crash severity regardless of the type or region of the accident. In the study comparing passenger vehicle and freight vehicle, there have been many studies showing the results of the crash severity of freight vehicle being very high. In addition, in a study that set freight vehicle as the main analysis targets, factors due to vehicle characteristics and behavior characteristics of freight vehicle were found to increase the severity of accidents. The main factors derived were night, downhill section, and traffic conditions requiring rapid braking”
“This study aims to construct a truck crash severity model using various models as well as the models used in previous studies. In addition to the ordered model that is basically used, various methodologies such as multinomial, random-effects, mixed-effects, and multilevel are used. In addition, evaluate the model constructed through performance comparison between these models.”
Comment 4
Methodology
In the abstracts, the authors claimed that they used 10 models which were not substantiated and justified in the introduction. And there is no pointer to those models in the literature review, with the relevance of these models to this present study.
Response
- Thank you for the reviewer’s comment. The pointer to the 10 models are emphasized in the model description part of methodology.
Comment 5
Too many long sentences, for instance, lines130 to 133, the seems to be too long. There are too many unclear statements in this section, for example, in lines 134 - 135 " To improve these limitations, if the dependent 134 variable has an order, an ordered model is used.". Also, in line 138 what do you mean by "wibble distribution"?
Response
- Thank you for the good point. The sentence structure, unclear expressions, and typos in the description of the ordered model you pointed out(line 140 to 151) were corrected.
“In general, analysis can be performed using the probit model or the logit model in the case of dependent variables that do not have an order. But if the dependent variable is not binary, the analysis through the probit model or the logit model risks making errors. There is a limitation in making an error by analyzing the difference between Y=0, Y=1 such as difference between properly damage only and injury, and Y=1, Y=2 such as difference between injury and fatal. Ordered models are used to improve the limitation that binary models are not available when dependent variables have an order. The logit model and the probit model have the same basic characteristics, but are classified according to the probability distribution form of the error term. The logit model assumes that the probability distribution of the error term is the same variance and follows an independent logistic distribution, and the probit model assumes that the probability distribution of the error term has the same variance and the covariance is 0.”
Comment 6
The models were not discussed to capture the essence of deploying them for this study. Section 3.5, lines 205 to 208 should be properly referenced.
Response
- Thank you for the reviewer’s detailed review. The explanation of the marginal effect was added, and the corresponding part was modified so that it could be reflected in the interpretation of the results.
Comment 7
Results and Discussion
This section is too shallow and disappointing. The results are presented in the tables with little discussion on the findings with no depth in terms of arguments and interrogations of the findings with the relevant literature.
Response
- Thank you for the reviewer’s comment. We supplemented the results and discretion with a lot of content. For example, an explanation of the case where the research results and general common sense were reversed was added as the contents of line 419-433.
“Variables that lower the severity of the accident were toll gates, ramps. Toll gates and ramps are in a situation in which deceleration is required, and in many cases, concentration of the driver is required. Road fault, snowy, wet surface condition, and construction area were also analyzed as factors that lower the severity of the accident, but this should consider various situations. These four factors are situations in which drivers have to pass with very high concentration, and accordingly, the need for low-speed driving is high. However, as soon as a driver problem or other factors occur in the corresponding sections, a very dangerous situation may be reached. In the case of road fault, dangerous actions such as sudden changing lanes are often accompanied to avoid road fault sections. If there is no equipment such as a snow chain when it snows, there is a risk of driving on a very slippery road. Similarly, even if the road surface is wet, there is a risk that the road surface is slippery and the brake is not properly caught due to the water barrier. Finally, in the case of the construction area, there is a geometric problem in which the width of the road narrows and the alignment is poor. In addition, the presence of workers working at construction sites increases the risk of pedestrian accidents on expressways”
Author Response File: Author Response.pdf
Reviewer 3 Report
The study develops several Logit and Probit models to investigate the contributing factors of freight vehicle crash severity. Overall, the analysis is reasonable and interesting. However, the reviewer has several concerns.
1. what are the motivations and contributions of this study?
2. what are the research gaps? Why develop the multilevel mixed-effects models?
3. There are some typos in the text.
4. why this study employs so many models?
5. How were the developed models estimated?
6. What is the significance level used for this study?
7. All the model analysis should focus on the contributing factors of freight vehicle crash severity.
8. The effect of various factors on freight vehicle crash severity should be interpreted. As well, the corresponding practical implications are suggested to add.
9. Regarding the ramdom parameters (i.e., random coefficients and effects), more information is needed. For example, how to determine a parameter is random or fixed. Some insights can be found from the following references:
https://doi.org/10.1016/j.amar.2016.04.001
https://doi.org/10.1016/j.amar.2021.100204
10. The writing needs more work.
Author Response
The authors are thankful to the reviewer for their valuable comments that helped in revising and significantly improving the paper. Each comment/suggestion from the reviewer have been put in Italics and is followed by our response and action to revise the paper accordingly.
Comments and Suggestions for Authors
The study develops several Logit and Probit models to investigate the contributing factors of freight vehicle crash severity. Overall, the analysis is reasonable and interesting. However, the reviewer has several concerns.
Comment 1
what are the motivations and contributions of this study?
Response
- Thank you for the good question. Currently, the number of deaths from traffic accidents in Korea is on a steady decline. However, in the case of freight vehicle accident on the highway, it was found that it did not follow such a trend. So I started researching because I became interested in freight vehicle accidents. And, using the results derived from this study, we expect to identify the factors that increase the severity of the truck accident on the highway and reduce the severity of the accident by investing resources first for the main factors that increase the severity of the accident.
Comment 2
what are the research gaps? Why develop the multilevel mixed-effects models?
Response
- We reinforced the research gaps from the literature review. The content corresponds to lines 131 to 135. The reason for using the multilevel mixed-effects model was used in this study because it has the advantage of freely handling fixed and random parameters within one model.
“This study aims to construct a truck crash severity model using various models as well as the models used in previous studies. In addition to the ordered model that is basically used, various methodologies such as multinomial, random-effects, mixed-effects, and multilevel are used. In addition, evaluate the model constructed through performance comparison between these models.”
Comment 3
There are some typos in the text.
Response
- Thank you for the good point. We checked the typo and corrected it.
Comment 4
why this study employs so many models?
Response
- Thank you for the reviewer’s question. We starting with basic model analysis, we tried to see how the difference differs when additional options are reflected in the model one by one. In the process, multiple, random-effects, mixed-effects, random parameters, multilevel were combined. In addition, the difference between the logit model and the probit model was also confirmed. As a result, a total of 10 models were analyzed.
Comment 5
How were the developed models estimated?
Response
- Thank you for the comment. Maximum likelihood estimation was applied to model estimation. The content corresponds to lines 274 to 275.
“The maximum likelihood estimation method was used for each model estimation.”
Comment 6
What is the significance level used for this study?
Response
- Thank you for the good question. The statistical significance level was 95%, and the content was added to line 276.
“In addition, each model was constructed while removing variables until only 95% of the significant variables remained, and the final model construction results are as follows.”
Comment 7
All the model analysis should focus on the contributing factors of freight vehicle crash severity.
Response
- Thank you for the reviewer’s suggestion. In interpreting the analysis results according to the reviewer's comments, the content was reflected focusing on the contributing factors of crash severity.
Comment 8
The effect of various factors on freight vehicle crash severity should be interpreted. As well, the corresponding practical implications are suggested to add.
Response
- Thank you for the reviewer’s detailed review. A detailed description of each factor (variables) and the crash severity is written in the conclusion.
Comment 9
Regarding the ramdom parameters (i.e., random coefficients and effects), more information is needed. For example, how to determine a parameter is random or fixed. Some insights can be found from the following references:
https://doi.org/10.1016/j.amar.2016.04.001
https://doi.org/10.1016/j.amar.2021.100204
Response
- Thank you for recommending a good reference paper. According to the reviewer's advice, information on random parameters was added.
Comment 10
The writing needs more work.
Response
- Thank you for the reviewer’s suggestion. We revised and supplemented the sentence composition and grammar in the paper.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
The previous comments were not adequately addressed and as such in my own opinion, this manuscript is not publishable in its present state.
Author Response
Thank you for the reviewer’s detailed review. The grammatical expression and typos of the entire text corrected after reviewed by the proofreading team in university. In addition, accordingly, the contents of the previous review can be clearly reflected.
Author Response File: Author Response.pdf
Reviewer 3 Report
The reviewer comments have been addressed.
Author Response
Thank you for the reviewer’s detailed review. We revised the grammar and spelling of English again after being reviewed by the proofreading team in university. As a result, that the contents of the text could be explained more clearly accordingly.
Author Response File: Author Response.pdf
