Variations in Mode Choice of Residents Prior and during COVID-19: An Empirical Evidence from Johannesburg, South Africa
Round 1
Reviewer 1 Report
Lines 61-64: please avoid using statements like "some of the changes". Please explicitly explain which changes were investigated in this study. Please improve this sentence by including the research gap, aim, and objectives in advance.
Please add more about your novelty.
Subtitle 2.1 is not necessary because there was no 2.2. Please merge the sections and make a full story for the method.
A subtitle should not start with a table. Please correct section 3.2
185 respondents may not be enough for a survey study. Please calculate the minimum sample size required using the population of the case study area.
Please correct grammar and typo mistakes.
You have applied some statistical tests and models. Why did you choose these tests? Did you meet their assumptions? Please improve your method and result in sections.
The method of this study was not novel. These types of studies have been carried out. One of your biggest novelty was your study area. Therefore you need to discuss your results by comparing other former relevant studies. A separate discussion section is required. Please, improve your discussion section.
Author Response
please see the attachment in the box
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Referee Report: Variations in Mode Choice of Residents Prior and During COVID-19: An Empirical Evidence from Johannesburg, South Africa
1. It is an interesting and relevant aim to assess how travel patterns were modified during the pandemic in African countries, and in South Africa as well.
2. Although the referee appreciates the fact that the authors might not be native English speakers, the paper still needs a thorough proofreading (syntax, etc.). The way this paper is written is rather confusing at times. The approximations that are accumulating throughout the paper confuse the reader and weakens the points that the authors are trying to make. This is true from the Abstract onwards. In the Abstract, the datasets used are mentioned at different times and places but the choice of these databases is not really motivated. Moreover, the period analysed is never mentioned, so the readers cannot appreciate the time frame nor the granularity of the data used which can make or break the results’ robustness. I would not advise this paper should be published before a thorough proofreading. Many mistakes throughout (e.g., “Walking and Cycling is seen as safe and low-carbon means of transport” instead of “are”, p.3 l.97), mention of “peak-fattening” instead of “flattening” (l.101-102)…
3. Several studies of the impact of Covid on travel in South Africa are not referenced (some examples: https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0245886; https://www.sciencedirect.com/science/article/pii/S235234092031341X among others). They would be useful to compare the authors’ results with those from the current state of the art literature.
4. The analysis lacks rigor overall. The graphs are very imprecise (Table 2 shows a distribution of chosen modes without the number of observations, Figure 3 . Authors keep referring to “before the pandemic” / “after the pandemic” then “during the first wave” / “during the second wave” of the pandemic but it’s not clear (1) what’s the frequency of the data used (monthly? Daily? Else?) nor (2) what are the periods precisely under study.
5. The identification strategy is not very convincing either: when introducing the multinomial logit specification, authors say “The independent variables represent the gender (male [1], female[0]), Age, household size, income, travel time, and travel cost”. However, there is a large risk of correlation between (1) gender and income, (2) age and travel time (amongst others) that are likely to bias the final results.
6. On the basis of the contents of comments #3 and #4 above, implying causal effects under such premises is simply not possible. At best, these effects are associations/correlations. Tables 7 and 8 remain very difficult to read (no clear flagging of the significance of coefficient using “*”, the number of observations is not available…).
7. Finally, regarding the specification, it would make more sense to pool together all information (pre/post Covid) and create a pre-post Covid dummy variables that would indicate us whether there’s been a significant change in mode choice after the pandemic. Therefore, tables 7 and 8 miss the point of the research objective as it is.
Author Response
please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
â‘ There is no introduction of related research in this paper. How to reflect the innovation of the article?
â‘¡ The questionnaire data used in the article is not sufficient.
â‘¢ Is the research method used in the article applicable?
â‘£ The discussion part of the results of the article is not closely integrated with reality.
Author Response
please see the attachment
Author Response File: Author Response.pdf
Reviewer 4 Report
The manuscript presents an analysis of pre and during COVID-19 pandemic travel behavior in Johannesburg. However, the theoretical soundness is still questionable, based on the content being presented in the present form. In addition, the manuscript writing is still lacking detail in several sections, i.e., literature review, data, methods. As many parts still need to be improved, I would recommend a major revision.
-
It will be informative if the COVID-19 infection cases for the whole South Africa and the city of Johannesburg be reported in the first section as a good reference.
-
For the online survey, who were the target respondents? Whom were the emails sent to? How were the samples obtained from such social networks as Facebook validated to be a good representative of the population? Although the profile of the samples was briefly reported in Section 3.2, it is important to evaluate how they were a good representative for the population. For instance, those middle-age and highly educated groups were the major travelers on the roads of the city, etc.
-
The design or outline of the questionnaire was not reported. Such important details as example questions and answers, scale of the measurements were missing. This is important to validate the data used in the study.
-
Summary statistics of the data was missing.
-
Taro Yamane’s formula is common. No need to put much detail here. But rather provide relevant reference.
-
Figure 1 was not referred to in the text, although it was not very informative. The basemap might have a copyright concern, please confirm.
-
Figure 2 was referred to TomTom 2021. Again a copyright issue might arise.
-
Section 3, line 241s, although the congestion of these roads may be common on weekdays, discussing in this context might need specific time, at least specifying year, pre or during the pandemic.
-
Figure 3 was referred to as congestion level, however in traffic engineering this is the V/C ratio. The highest value was 38% far under the capacity, which is generally not considered to be congested. On the other hand, congestion is commonly shown by speed, or density in a more technical sense. This section needs a revision.
-
The manuscript lacks a literature review on the analysis part, e.g., similar study or findings in the other cities must be discussed and referenced in terms of the analysis method adopted.
-
The analysis methods were not adequately described in Section 2 to be consistent with the results being presented in Section 3.3 to 3.6.
-
Why were the degrees of freedom of the before and during the COVID different?
-
The multinomial logistic regression took three modes of transport (private, public, and active modes). The background of the mode choice behavior in the city was needed, such as the average trip length, trip travel time, trip purpose of each mode. Without such information, it is difficult to judge the validity of the results in the following parts.
-
The multinomial logistic regression model needs description of the explanatory variables, i.e., how they were calculated with respect to the reference mode (private car), especially the travel cost and travel time. The discussion might need to be based on behavioral model or utility maximization theory, rather than just a statistical model based on logistic regression.
-
The df in Table 7 and 8 may not be very informative.
-
The values shown in Table 7 and 8 included the statistically insignificant variables, making the reading very confusing.
-
The discussion in Section 4 was not majorly led by the findings of the results having been presented, e.g. change of trip purposes, etc.
-
Section 5 is presently superficial, so needs more substantial supports from the finding of the study.
Author Response
please see the attachment
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Dear Authors, Many thanks for considering my comments and improving the manuscript. As you calculated in the method section, min sample size (i.e. number of responses) is around 400 for your case study area. An analysis with 185 participants gives biased results in your models. I recommend you improve the number of samples and run the analysis one more time in order to obtain a good model. Please immediately improve your data and resubmit the manuscript.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 3 Report
The revision is fine.
Author Response
Please see the attachment
Author Response File: Author Response.pdf
Reviewer 4 Report
The revision was significant where most of the comments were responded. This has made the paper more comprehensive and can be recommended for publication in Sustainability. However, please recheck the caption of Table 2 and some remaining typo.
Author Response
Dear Reviewer,
Thank you for your comments. As suggested, we have changed the caption for Table 2 and we have also edited the article for typographical and grammatical errors.
Thank you for your time.
Round 3
Reviewer 1 Report
The number of survey participants is not enough. Therefore, the results are biased. Please improve your data and revise your manuscript.
Author Response
Please see attachment
Author Response File: Author Response.pdf
