Exploring Associations between Multimodality and Built Environment Characteristics in the U.S

Round 1

Reviewer 1 Report

1. This study investigated the association between multimodality and built environment characteristics across United States at CBG level. It can make a contribution to current research and planning. I recommend a major revision.

2. I recommend the text quality of this paper should be improved and polished. For instance, 1) The first sentence in Abstract could make ambiguity that what are pollution and sprawl, do you mean environmental pollution, and urban sprawl? 3) In line 18, is it street network density? and so on…, e.g., line 46: 2015-2019 5-year estimates…; in line 62, what is the mode of active transportation? In line 82, what is population centrality, and is it represent accessibility?

 

3. The literature review should be improved by providing context for the current studies related to the reviewed topic. For example, section 2.1 is not enough on current multimodality researches; and how existing studies measured multimodality; In line 74-75, a better way to list the related factors with the references should be one by one, and how these factors related with multimodality.

 

4. The research gap is not clear. As the author stated, the role and importance of the built environment characteristics on multimodality have not been adequately explored, from what perspectives? In addition, this study stated that it has demonstrated a link, is the link a relationship or causality? As the title shows, it explored the association. It should be more clearly towards the role and importance of the built environment, and the research gap.

 

5. The dependent variable of multimodality is measured by commuter trips, are there other trip purposes? or other travelers?

 

6. The result of multimodality in 3.2.1 is better to move to result section, as well as independent variables of Table 2 and Figure 3.

 

7. In line 226, since the CBG is a relatively small scale of boundary, there could exist spatial autocorrelation. Two regional job accessibility variables (X8 and X9 are still measured at CBG level). Statistical models such as spatial regression (lag model or error model), or mixed model (e.g., taken State as a mixed effect) should be a better complement with OLS.

 

8. The nine socioeconomic factors are covariates, not control variables. Is certain factor a control variable, there should exist control group and experimental group (i.e., low and high). Any overstatement is not appropriate. Also in the results section, e.g., line 296, the variables can determine multimodality may be overstated. There is no causality found, just relationships of OLS.

 

9. Does the GBDT model could disentangle causality, or take the SES covariates as control variables? I think GBDT just find the relationship contribution (or prediction importance), and non-linear relations.

 

10. What are the differences between impurity-based and permutation-based rules. Why you use permutation-based results in section 4.2?

 

11. The discussion on the results, especially built environment factors, should be tailored, and discussed more thoroughly, and why.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

This article presents a case study to explore the impact of built environment characteristics in the United States of America.

The article is well-structured, states its research questions clearly, and acknowledges the limitations of the approach in the conclusions section.

The study itself seems to fill a missing gap between built environments and multimodality, which has not yet been sufficiently addressed, and uses publicly available information to conduct the study.

The choice of metrics and the choice of indicators for the anticipated study goals seems appropriate to me. However, I wonder which of these indicators are US-specific. Even though the authors acknowledge at the end of the article that the results may not be applicable to other countries, because of differing travel behaviors, it might be interesting to know for which countries the results might be similar or applicable. For example, could the results be significant for Canada as well?

The authors also remark that travel patterns of non-commute trips were not investigated. While it is a reasonable assumption to exclude those, could they state an estimate on the importance of non-commute trips in comparison to commute trips for their research questions?

Then, the authors state that vital key findings of the study were predictors of multimodality, in particular walkability index, and network density, among others. Could the authors elaborate on problems that were actually identified by those metrics? Could you give some concrete examples of the cities you investigated that showed e.g. a low walkability index and how one could improve the situation? 

Finally, now that we have learned from this study that the built environment impacts the multimodality choices of users to a significant extent, could the authors set these results into perspective with multimodality studies that have been previously conducted? How important is the built environment with respect to other indicators identified by science before?

Apart from the aforementioned questions that I would like the authors to elaborate on more, I think the study is worth publishing and I would recommend it with only minor revisions.

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

I recomment to accept.