Patterns and Analysis of Traffic Accidents in New York City between 2013 and 2023

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper primarily analysis of accident trends in New York City from 2013 to 2023, examining various dimensions including accident types, severity, causes, and locations. accidents.Upon reviewing the paper, I found that the paper has a relatively complete and in-depth research process, and the overall structure of the article is fairly complete and clear.However, there are areas for improvement.

Firstly, The abstract contains too much background information: The opening section provides detailed descriptions of New York City's population density and traffic environment. While this background information aids in understanding, it may take up too much space, making the abstract less concise.

Secondly, The conclusion is lengthy and the information is too scattered: While the conclusion is rich in content, it presents too many dispersed points, making it difficult to highlight the key findings, and preventing readers from quickly grasping the core conclusions.

Thirdly,The background information is too broad: While the introduction offers a description of traffic accidents in New York City, it may be overly general and does not adequately focus on the specific motivation for the study or the reasons for selecting this particular time period.

Finally,The connection between the discussion and introduction is insufficient: The introduction fails to establish a strong foundation for the analysis and interpretation in the discussion section, lacking a deeper connection between the background information and the research findings.

The final recommendation for the paper is a major overhaul. 

 

 

Comments on the Quality of English Language

The English language in the paper is generally clear and conveys the main ideas effectively, though there are some areas where conciseness could be improved. Additionally, the conclusion would benefit from more focused and structured presentation to enhance readability and highlight key findings more clearly.

Author Response

We would like to thank the editors and reviewers for their feedback.  Your recommendations were taken seriously, and we performed a major revision to this paper. The new analysis is much more detailed and scientifically sound.  We feel that the re-submitted paper is a substantial improvement over the initial submission, thanks to your feedback.

Below, we address each of the reviewer’s comments individually.  The reviews are in black, with our responses in red.

REVIEW 1:

The paper primarily analysis of accident trends in New York City from 2013 to 2023, examining various dimensions including accident types, severity, causes, and locations. accidents.Upon reviewing the paper, I found that the paper has a relatively complete and in-depth research process, and the overall structure of the article is fairly complete and clear.However, there are areas for improvement.

Firstly, The abstract contains too much background information: The opening section provides detailed descriptions of New York City's population density and traffic environment. While this background information aids in understanding, it may take up too much space, making the abstract less concise.

Concur.  We took out the unnecessary background information that was in the abstract to make room for more detailed discussion of our analysis and results.

 

Secondly, The conclusion is lengthy and the information is too scattered: While the conclusion is rich in content, it presents too many dispersed points, making it difficult to highlight the key findings, and preventing readers from quickly grasping the core conclusions.

Concur.  We rewrote the conclusion to ensure that it is more organized.

 

Thirdly,The background information is too broad: While the introduction offers a description of traffic accidents in New York City, it may be overly general and does not adequately focus on the specific motivation for the study or the reasons for selecting this particular time period.

Concur.  We removed non-pertinent information from the background section and made it more focused to our study.

 

Finally,The connection between the discussion and introduction is insufficient: The introduction fails to establish a strong foundation for the analysis and interpretation in the discussion section, lacking a deeper connection between the background information and the research findings.

Concur.  We significantly modified the paper so that the introduction provides a stronger foundation for this analysis.  We introduced more datasets to help us better analyze the data and present the results.

 

The final recommendation for the paper is a major overhaul. 

Concur.  We overhauled the entire paper!

Reviewer 2 Report

Comments and Suggestions for Authors

The paper effectively highlights the significance of the study by emphasizing the high population density and the impact of various factors like the COVID-19 pandemic and vehicle electrification on traffic accidents. The inclusion of the COVID-19 pandemic period is particularly interesting and relevant, as it represents a unique scenario that could have affected traffic patterns and accident rates. The paper can be accepted after minor revision. Here are the questions,

1. How confident can we be that the observed trends are consistent across different areas of New York City, and not just localized to certain neighborhoods or boroughs?

2. Are there any specific factors or behaviors that can be attributed to the lasting changes in commuting patterns and traffic volumes post-pandemic?

3. How do the conclusions address the safety of pedestrians and cyclists, who are often more vulnerable in traffic accidents?

4. Based on the conclusions, what areas of research would the authors recommend for further exploration to enhance road safety?

Author Response

We would like to thank the editors and reviewers for their feedback.  Your recommendations were taken seriously, and we performed a major revision to this paper. The new analysis is much more detailed and scientifically sound.  We feel that the re-submitted paper is a substantial improvement over the initial submission, thanks to your feedback.

Below, we address each of the reviewer’s comments individually.  The reviews are in black, with our responses in red.

 

The paper effectively highlights the significance of the study by emphasizing the high population density and the impact of various factors like the COVID-19 pandemic and vehicle electrification on traffic accidents. The inclusion of the COVID-19 pandemic period is particularly interesting and relevant, as it represents a unique scenario that could have affected traffic patterns and accident rates. The paper can be accepted after minor revision. Here are the questions,

1. How confident can we be that the observed trends are consistent across different areas of New York City, and not just localized to certain neighborhoods or boroughs?

With our updated analysis, we were able to look at the accidents trends from the different boroughs and found that the results were not localized to an individual borough.  What we saw in terms of the different boroughs is a reduction in Manhattan relative to the other boroughs.  This coincides with what we found with the increase in FHV services, which impacts Manhattan significantly more than the other boroughs.  We integrated discussion about the boroughs throughout the paper.

2. Are there any specific factors or behaviors that can be attributed to the lasting changes in commuting patterns and traffic volumes post-pandemic?

We decided to dive into this topic significantly in the revision.  We pulled datasets to look at different causes.  Sections 4 and 5 includes discussions on this topic.

3. How do the conclusions address the safety of pedestrians and cyclists, who are often more vulnerable in traffic accidents?

We added additional discussion to discuss the safety of pedestrians and cyclists. This is in Section 3.4 as well as our Section 5.3. 

4. Based on the conclusions, what areas of research would the authors recommend for further exploration to enhance road safety?

We added discussion in Section 5 to address areas of research that would be relevant to further explore road safety.

Reviewer 3 Report

Comments and Suggestions for Authors

This study investigated the traffic accidents in New York City during the past decade. Although some statistical results and findings were provided, the contribution of this study is limited due to the following reasons.

(1) The methodology part is missing. In the study, there is no detail about the methodology. From the following experiments, we can guess that the method they used is primarily the descriptive statistics. Considering the simplification of the methodology, the contribution of this study is limited.

(2) The findings from the experimental results were drawn from existing studies and did not provide new insights for researchers and practitioners. In this regard, the contribution of this study is limited.

(3) My major suggestion is that the authors could use multi-source data and a few advanced techniques (e.g., machine learning) to mine more insightful patterns and knowledge for accident analysis and managment.

Author Response

We would like to thank the editors and reviewers for their feedback.  Your recommendations were taken seriously, and we performed a major revision to this paper. The new analysis is much more detailed and scientifically sound.  We feel that the re-submitted paper is a substantial improvement over the initial submission, thanks to your feedback.

Below, we address each of Reviewer 3’s comments individually.  The reviews are in black, with our responses in red.

REVIEW 3:

This study investigated the traffic accidents in New York City during the past decade. Although some statistical results and findings were provided, the contribution of this study is limited due to the following reasons.

(1) The methodology part is missing. In the study, there is no detail about the methodology. From the following experiments, we can guess that the method they used is primarily the descriptive statistics. Considering the simplification of the methodology, the contribution of this study is limited.

We modified our methodology significantly.  The new methodology is discussed in Section 2.2.  Section 4 and 5 are effectively new sections to move away from simply being descriptive statistics.

(2) The findings from the experimental results were drawn from existing studies and did not provide new insights for researchers and practitioners. In this regard, the contribution of this study is limited.

Concur!  With our new analysis, we were able to pull new insights from the data.

(3) My major suggestion is that the authors could use multi-source data and a few advanced techniques (e.g., machine learning) to mine more insightful patterns and knowledge for accident analysis and managment.

Concur.  We made these modifications into our revision.  We pulled data from a number of different sources.  While we did not use machine learning to analyze the data set, we did build a linear model to understand the correlations between vehicle accidents and factors that effected them.

Reviewer 4 Report

Comments and Suggestions for Authors

General comments: Overall, this is a well-written and clearly analyzed article describing important changes in crash rates in recent in years in New York City, largely affected by the pandemic and health-related restrictions and voluntary changes in driving patterns. I have relatively minor comments below that should be addressed; most importantly, to consider alternative explanations for changes in severity patterns—specifically, that the pandemic changed the population of drivers (by reducing the number of drivers who obeyed restrictions or voluntarily chose to reduce exposure), potentially leaving the driving population with a higher-than-average amount of risky drivers, which could yield continued rates of severe crashes despite overall crash rates and mileage dropping.

 

Minor edits / typos:

 

Line 225: “dd” should be “did”

 

Specific comments:

 

Table 1: “Vehicle Type” is very vague here. Does it mean make, model, model year, trim-level, or a higher level class of vehicle (“SUV”, “light truck”)? 

 

Table 2: Any theories about why Manhattan’s share of accidents dropped so much between 2013 (26.7%) and 2020 (14.7%)? Additional data showing population for the boroughs (or vehicle registration data, if there’s a source for that), would be extremely useful to evaluate these sizable changes. A lot of popular press reports had indicated NYC public transit, which is likely heavily utilized in Manhattan, had suffered significantly during and after the pandemic, and yet these numbers would indicate that people traveling in Manhattan are avoiding driving. Could it be related to an increase in telecommuting for people who work in Manhattan, and, as a result, fewer people driving or getting taxis or rideshares to offices in Manhattan? If that’s the case, then some figure that measures how many people are in an area for work-related reasons, as opposed to primary residence, would be useful.

 

Line 194: Can you describe how these were identified? You stated that there is a “Contributing Factor” field in the reports, but also noted that it was free entry. Did you have to search for specific strings, and, if so, what strings did you search for?

 

Line 213: Similar to above, how were these crashes identified? Were there specific strings you searched for, and what were they?

 

Line 222: The steadiness of the contribution of distraction to crashes fits with what NHTSA has reported recently, and the percentage you computed (25%) is close to a recent NHTSA report (29%), which I think is interesting and worth mentioning:

Blincoe, L., Miller, T., Wang, J.-S., Swedler, D., Coughlin, T., Lawrence, B., Guo, F., Klauer, S., & Dingus, T. (2023, February). The economic and societal impact of motor vehicle crashes, 2019 (Revised) (Report No. DOT HS 813 403). National Highway Traffic Safety Administration.

 

Line 259: This paragraph is very speculative (it could be argued that people telecommuting instead of driving into offices reduces situations in which drivers’ minds are preoccupied by thinking about work, because they are mostly at home); I wouldn’t remove it, but I would qualify the language as being more speculative than definitive.

 

Line 274: This paragraph is a bit strange—it’s introducing new data in the discussion, but also linking active safety systems to vehicle powertrain (here, electric)—these safety features are increasingly common on many cars, including internal combustion engine vehicles, and I’m not sure they’re more common on electric vehicles (although certainly Teslas frequently are equipped with Autopilot, which, depending on how it’s being evaluated, can be seen as having a safety benefit or presenting a new risk). It could be argued that the increasing prevalence of electric vehicles in the fleet of NYC vehicles is emblematic of fleet turnover as a whole; as drivers increasingly buy new vehicles, those vehicles are typically more likely to be equipped with new safety features. As a counterpoint, there have been arguments that new vehicles may also be bigger and heavier, and present unique risks to vulnerable road users (pedestrians, pedalcyclists), especially in urban settings.

 

Line 302: Where is this data from? Is this using the “vehicle type” field from the main dataset?

 

Line 350: One possibility that you’ve left out is the composition of the population of drivers. Many of the Covid-related safety precaution measures were voluntary (or not heavily enforced), and reductions in driving that accompanied reductions in exposure to the virus were also, largely, voluntary. Thus, the resulting pool of drivers who remained on the road contained a mix of both essential workers but also, critically, people who may be ignoring voluntary guidelines or flouting actual rules—they may be people who are also associated with risky driving. It’s possible that the driving reductions that were seen during Covid mostly removed lower risk drivers from roads, but higher risk drivers remained, which would explain the consistently high severe crash and fatal crash rates. One way to check would be to look to see if countries that had either higher overall levels of pandemic restriction compliance or greater penalties and/or enforcement for restriction violation saw significantly lower serious crash rates.

Author Response

We would like to thank the editors and reviewers for their feedback.  Your recommendations were taken seriously, and we performed a major revision to this paper. The new analysis is much more detailed and scientifically sound.  We feel that the re-submitted paper is a substantial improvement over the initial submission, thanks to your feedback.

Below, we address Reviewer 4’s comments individually.  The reviews are in black, with our responses in red.

General comments: Overall, this is a well-written and clearly analyzed article describing important changes in crash rates in recent in years in New York City, largely affected by the pandemic and health-related restrictions and voluntary changes in driving patterns. I have relatively minor comments below that should be addressed; most importantly, to consider alternative explanations for changes in severity patterns—specifically, that the pandemic changed the population of drivers (by reducing the number of drivers who obeyed restrictions or voluntarily chose to reduce exposure), potentially leaving the driving population with a higher-than-average amount of risky drivers, which could yield continued rates of severe crashes despite overall crash rates and mileage dropping.

 Minor edits / typos:

 Line 225: “dd” should be “did”

Corrected.

Specific comments:

 Table 1: “Vehicle Type” is very vague here. Does it mean make, model, model year, trim-level, or a higher level class of vehicle (“SUV”, “light truck”)? 

Corrected. We modified the description in Table 1 to reflect this.

Table 2: Any theories about why Manhattan’s share of accidents dropped so much between 2013 (26.7%) and 2020 (14.7%)? Additional data showing population for the boroughs (or vehicle registration data, if there’s a source for that), would be extremely useful to evaluate these sizable changes. A lot of popular press reports had indicated NYC public transit, which is likely heavily utilized in Manhattan, had suffered significantly during and after the pandemic, and yet these numbers would indicate that people traveling in Manhattan are avoiding driving. Could it be related to an increase in telecommuting for people who work in Manhattan, and, as a result, fewer people driving or getting taxis or rideshares to offices in Manhattan? If that’s the case, then some figure that measures how many people are in an area for work-related reasons, as opposed to primary residence, would be useful.

Great point!  Digging through the data, we identified that a primary cause is due to the rise of for-hire vehicles, as opposed to taxi cabs.  We added additional discussion and analysis to show this.

 Line 194: Can you describe how these were identified? You stated that there is a “Contributing Factor” field in the reports, but also noted that it was free entry. Did you have to search for specific strings, and, if so, what strings did you search for?

We added a paragraph at the start of Section 3.4 to address this.  We reviewed how that data was collected, and it was from a selection of 60 different options, with the opportunity for a free-response. However, in practice, officers only used the free response six times, and they appear to have been typos.

 Line 213: Similar to above, how were these crashes identified? Were there specific strings you searched for, and what were they?

We added a paragraph at the start of Section 3.4 to address this

Line 222: The steadiness of the contribution of distraction to crashes fits with what NHTSA has reported recently, and the percentage you computed (25%) is close to a recent NHTSA report (29%), which I think is interesting and worth mentioning:

Blincoe, L., Miller, T., Wang, J.-S., Swedler, D., Coughlin, T., Lawrence, B., Guo, F., Klauer, S., & Dingus, T. (2023, February). The economic and societal impact of motor vehicle crashes, 2019 (Revised) (Report No. DOT HS 813 403). National Highway Traffic Safety Administration.

THANK YOU!!!!  WE INCLUDED THAT REFERENCE.

Line 259: This paragraph is very speculative (it could be argued that people telecommuting instead of driving into offices reduces situations in which drivers’ minds are preoccupied by thinking about work, because they are mostly at home); I wouldn’t remove it, but I would qualify the language as being more speculative than definitive.

You are correct!  We changed the language to make it sound less definitive.

Line 274: This paragraph is a bit strange—it’s introducing new data in the discussion, but also linking active safety systems to vehicle powertrain (here, electric)—these safety features are increasingly common on many cars, including internal combustion engine vehicles, and I’m not sure they’re more common on electric vehicles (although certainly Teslas frequently are equipped with Autopilot, which, depending on how it’s being evaluated, can be seen as having a safety benefit or presenting a new risk). It could be argued that the increasing prevalence of electric vehicles in the fleet of NYC vehicles is emblematic of fleet turnover as a whole; as drivers increasingly buy new vehicles, those vehicles are typically more likely to be equipped with new safety features. As a counterpoint, there have been arguments that new vehicles may also be bigger and heavier, and present unique risks to vulnerable road users (pedestrians, pedalcyclists), especially in urban settings.

Great point!  We adjusted our paper to discuss these safety features more generally.

 Line 302: Where is this data from? Is this using the “vehicle type” field from the main dataset?

With modifications requested from the other reviewers, this line was removed.

 Line 350: One possibility that you’ve left out is the composition of the population of drivers. Many of the Covid-related safety precaution measures were voluntary (or not heavily enforced), and reductions in driving that accompanied reductions in exposure to the virus were also, largely, voluntary. Thus, the resulting pool of drivers who remained on the road contained a mix of both essential workers but also, critically, people who may be ignoring voluntary guidelines or flouting actual rules—they may be people who are also associated with risky driving. It’s possible that the driving reductions that were seen during Covid mostly removed lower risk drivers from roads, but higher risk drivers remained, which would explain the consistently high severe crash and fatal crash rates. One way to check would be to look to see if countries that had either higher overall levels of pandemic restriction compliance or greater penalties and/or enforcement for restriction violation saw significantly lower serious crash rates.

Our updated model attempted to capture the impact (quantitatively) for a number of different changes in NYC over the past decade. However, it is difficult to capture the change in driver mindsets!  We added a specific paragraph discussing this in our limitations / future work.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I have no other opinions. I believe that the paper has met the publication requirements of your journal.

Author Response

Reviewer's Comments:

I have no other opinions. I believe that the paper has met the publication requirements of your journal.

Authors Response:

Thank you!!!!  

Reviewer 3 Report

Comments and Suggestions for Authors

I acknowledge the great effort of the authors on the paper revision. However, the methodology is still weak. Please use mathematical description in your paper.

Author Response

Comment 1: I acknowledge the great effort of the authors on the paper revision.

Response: Thank you!  The feedback from your reviews have really helped us shape this paper into something much stronger.

Comment 2: However, the methodology is still weak. Please use mathematical description in your paper.

Response: You were correct; this is something that we should have included in our paper.  We added a paragraph and a figure to the methodology section to provide a mathematical description of what we did.

Round 3

Reviewer 3 Report

Comments and Suggestions for Authors

I am pleased to accept this paper.