The Influence of Attitudes towards Cycling and Walking on Travel Intentions and Actual Behavior
Round 1
Reviewer 1 Report
The subject of the article titled "The influence of attitudes toward cycling and walking on travel intentions and actual behavior" is very interesting. However, after reading the manuscript, the following comments and comments are made:
- what new article brings to the scientific area?
- is the research methodology presented in the article different from the others used and described in the literature? Does it contain any innovative element?
- the Methods section lacks a more detailed description of such issues as the selection of the sample size, the method of sampling of respondents, error assessment related to statistical surveys of the population based on the sample,
- The Measures section lacks a more detailed description of the measurement process,
- The Data analysis section - there is a lack of a more detailed description of EFA, CFA, Varimax,
- Table 2: It is not known what the Variable and Item describe.
- Fig. 1 - 9: unreadable (difficult to read), and lacking a more detailed description of the figures/schemes-diagrams in the text.
Author Response
Our responses to the Reviewer 1’s comments are in bold green:
The subject of the article titled "The influence of attitudes toward cycling and walking on travel intentions and actual behavior" is very interesting. However, after reading the manuscript, the following comments and comments are made:
- what new article brings to the scientific area?
The contributions of the article to the scientific area are twofold. First, the study adopts the three-component model of attitudes, which differentiates cognitive, affective and behavioral attitudes towards cycling and walking. To the best of our knowledge, this is the first study in the travel behavior field that adopts the three-component model of attitudes. And second, this study includes simultaneously the interrelationships among intentions and real use of four different travel modes: car, public transport, bicycle and walk. This approach allows us to identify how each attitudes’ component influences on both the intention and the actual use of complementary and competing travel modes. As far as we know, this is the first work in the travel behavior field that study the influence of attitudes on the intention and real use of four travel modes. And the first work in the travel behavior field that study how walking is affected by attitudes towards cycling and walking.
- is the research methodology presented in the article different from the others used and described in the literature? Does it contain any innovative element?
The main difference of the research methodology used in the article compared with other in the travel behavior literature is related to the detailed analysis carried out of both the instrument used and the data collected. Internal consistency of the items included in the questionnaire was checked using Cronbach’s Alpha. Bartlett's test of sphericity and Kaiser-Meyer-Olkin Measure of Sampling Adequacy were used to confirm that factor analysis technique is appropriate. Pearson’s correlation matrix was obtained to observe the correlations between the three latent variables and among individual items for each transport mode, which were considered later in the Structural Equation models (SEM). Then, exploratory factor analyses were conducted to define the latent variables. Considering the results obtained, confirmatory factor analyses were executed specifying the posited relationships of the observed indicators to the latent variables. Finally, SEM models were fit in order to study the interrelationships among cognitive, affective and behavioral attitudes towards walking and cycling, intentions and current use of car, public transport, bicycle and walk.
- the Methods section lacks a more detailed description of such issues as the selection of the sample size, the method of sampling of respondents, error assessment related to statistical surveys of the population
Sampling frame consisted of two parts: e-mail lists provided by different public institutions and private companies, and a customer research panel based on the sample. A survey response was considered valid when respondents completed information regarding demographic and socioeconomic information, attitudes and values data.
- The Measures section lacks a more detailed description of the measurement process,
A more detailed description of the measurement process is included in the new version of the paper. Below the additional descriptions are presented (in bold green).
“Information regarding intention and real use of each mode of transport were obtained using a one-hundred-point scale. This way, participants are asked to distribute 100 points between their intentions to travel with each mode (car, public transport, walking and cycling).Similarly, the same question is proposed for stating their actual use of each travel mode. Thus, both the actual modal split and the intention are obtained and measured with percentages of each mode compared to the total amount of travel. In the cases where the total percentage did not sum 100 points, a correction was applied in order to standardize the responses and distribute the sum among one hundred percent. We acknowledge that this subjective self-information data regarding real use of transport modes could include small difference compared with observed data. Nevertheless, considering that the information was provided in percentage of use of each travel mode, only small differences could exist between observed and stated travel mode use.
To evaluate attitudes towards cycling and walking modes, the three-component attitudes model already described (affective, cognitive and behavioral) are assessed through 5-point Likert scales, ranging from 1 ‘‘strongly disagree” to 5 ‘‘strongly agree,. For each transport mode, 16 items are included in the survey.
The construct which measures affective attitudes is composed by 5 items, such as “I like it” or “It’s relaxing”. For cognitive attitudes 6 items were used, e.g. “It suits my needs” or “It’s comfortable”. Last, behavioral attitudes are measured with 5 items, in this case, as these type of attitudes has a behavioral component, the questions we framed in a specific area of study: urban mobility. For instance, the items are formulated as follows “I choose this travel mode considering the urban structure and its convenience” or “For urban trips, I choose this travel mode considering other people’s influence and needs”.
For each of the items composing the constructs, several descriptive analysis were carried out. Table 2 and Table 3 include the name of the variable, description of the items, median, mode and Standard Deviation measures for attitudes toward cycling and attitudes toward walking respectively. Consequently, 3 latent variables are built based on Factor Analysis results, as it is shown in the next section.”
- The Data analysis section - there is a lack of a more detailed description of EFA, CFA, Varimax,
Exploratory factor analysis (EFA) was used to examine all the pairwise relationships between individual variables. EFA seeks to extract latent factors from the measured variables. To facilitate the interpretation of the results, the clusters of items are rotated so that they are more closely aligned with the axis lines. To this end, Varimax rotation method was used, which produce factors that are uncorrelated.
The confirmatory factor analysis (CFA) determines whether the hypothesized structure provides a good fit to the data, or in other words, that a relationship between the observed variables and their underlying latent, or unobserved, constructs exist. The CFA models were estimated using a robust maximum likelihood method.
- Table 2: It is not known what the Variable and Item describe.
The column Item includes the definition of each item, as they are presented to respondents.
The Variable is the acronym of each item: COG stands for Cognitive, AF stands for Affective, and BEH stands for Behavior.
- Fig. 1 - 9: unreadable (difficult to read), and lacking a more detailed description of the figures/schemes-diagrams in the text.
Figures 1-9 have been enlarged, and all acronyms have been explained in caption.
A more detailed description of the results in Figures 1-9 is included in the new version of the paper. All associations are analyzed, and the size of the estimated effects considering the estimated values of the coefficients are compared. Below the additional descriptions are presented (in bold green).
“Figure 1 shows direct effects of model 1. Significant relationships were found between cognitive attitudes toward cycling and the intention to cycle and walk. As expected, this type of attitudes are found negatively associated with the intention to use motorized transport modes: public transport and car, and positively associated to the intention to cycling and walking. According to the values of the estimated coefficients, the size of the effects are similar in all cases except for the intention to cycling, which is almost double than the others relationships. Affective attitudes toward cycling provided also a significant and positive relation with the intention to cycle, but surprisingly a negative association with the intention to walk. Behavioral attitude toward cycling is negatively related with the intention to use car. “
“Results also confirm the existence of several direct relationships between attitudes toward cycling and current use of each travel mode, although this is only observed for cognitive attitudes. Cognitive attitudes toward cycling are positively associated with the real use of cycling and walking, and negatively associated to the real use of the car. The size of the effect is very small on walking, and larger on cycling.”
“As expected, results indicate that the intentions to use each travel mode are positively associated with the real use of the same mode. The size of these effects are much larger than the relationships between attitudes and real use. Only the intention to walk is found to be significantly and negatively associated with real bike use, with a very small size effect.”
“Figure 2 shows the significant indirect effects found between attitudes toward cycling and the actual use of bicycle. There is a positive relationship between cognitive attitudes towards cycling and real bike use mediated through the intention to cycle. However, a negative relationship was found when the mediator is the intention to walk. Regarding affective attitudes towards cycling and bike use, both mediators (intention to cycle and intention to walk) were found to be positive. The size of the effect of attitudes toward cycling on the real use of bike mediated by the intention to walk is much lower than the other indirect effects.”
“Figure 3 shows the significant indirect effects found between attitudes toward cycling and the actual walking. In this case, the intention to walk mediates cognitive attitudes toward cycling positively through actual walking. In contrast, intention to walk mediates affective attitudes toward cycling negatively through actual walking. The latter effect is only somewhat higher. “
“Additionally, cognitive and behavioral attitudes toward cycling are indirectly associated to the real use of car through the intention to use car as a negative mediator (Figure 4). Both effects are found to be small, in particular the one caused by behavioral attitudes toward cycling.”
“A similar effect was found with cognitive attitudes toward cycling and the use of public transport, which is negatively mediated with the intention to use public transport (Figure 5). According to the estimated value of the coefficient, the size of this effect is low.”
“Figure 6 represents the results of the model. As it is shown, cognitive attitudes toward walking are positively associated with the intentions to use active transport (bike and walk). As expected, the association is negative between cognitive attitudes toward walking and the intention to use car. Similarly, affective attitudes toward walking are positively associated with the intention to walk, and negatively associated with the intention to use car. On the other hand, behavioral attitudes toward walking are found to be positively associated to the intention to cycle, but unexpectedly negatively associated to the intention to walk. In general, the size of the effects are small, with the lowest values associated to the intentions to active transport.”
“Again, this model shows several direct associations between attitudes toward walking and actual use of travel modes. Thereby, cognitive attitudes toward walking are positively associated with current walking and negatively related with the use of car. The size of the effect is only a bit higher on the current walking. The model did not provide any significant association between affective attitudes toward walking and the use of travel modes, but behavioral attitudes toward walking are positively associated with the use of bike, and negatively related to the use of car. According to the estimated value of the coefficients, both effects are quite small.”
“Logically, intentions and actual travel behavior are associated in a similar way than in the previous model. The intention to use each travel mode is positively associated to the actual use of the same travel mode. The size of these effects are much larger than the relationships between attitudes and real use.”
“Figure 7 shows the significant indirect effects found between attitudes toward walking and the actual walk. The intention to walk was found to positively mediate between cognitive and affective attitudes toward walking and the actual walk, but negatively mediates between behavioral attitude toward walking and the actual walk. The latter effect is much lower than those related to cognitive and affective attitudes toward walking.”
“Figure 8 shows the significant indirect effects found between attitudes toward walking and the actual use of car. The intention to use car was found to negatively mediate between cognitive and affective attitudes toward walking and the actual use of car. According to the estimated value of the coefficient, both effects are quite low.”
“Finally, Figure 9 shows the significant indirect effects found between attitudes toward walking and the actual use of bicycle. The intention to use bicycle was found to positively mediate between cognitive and behavioral attitudes toward walking and the actual use of bicycle. Again, according to the estimated value of the coefficients, both effects are quite low.”
Author Response File: 
Author Response.docx
Reviewer 2 Report
This paper is very interesting and potentially very meaningful with the use of a new methodology..
With respect to findings, I think that the low (somewhat low) R-squared values need a bit more discussion.
With respect to methods, I think the acknowledgement of the data estimation process should have been included in that section instead of just as the next to last paragraph of the paper.
I would like to know why the on-line survey did not occur in August. (I do not want to have to guess that it was because of vacations, for example). I also would like to know how the demographics of the sample compares to the study population as a whole.
Finally, there were very few grammatical issues -- mostly just awkwardness (Line 15: “to reduce”; Line 45: “but using”; Line 216: “The data collection step”; Line 269: “Exceptionally; Line 269 "AF_34" should be identified as a variable when being discussed as to why it was included).
Overall, I found this to be a good and valuable read.
Author Response
Our responses to the Reviewer 2’s comments are in bold green
This paper is very interesting and potentially very meaningful with the use of a new methodology.
With respect to findings, I think that the low (somewhat low) R-squared values need a bit more discussion.
Following the recommendation of Marsh, Balla, and Hau (1996), and of Jaccard and Wan (1996), in this study a range of goodness of fit indices from different classes are used, so that the limitations of each index can be overcome. Fit indexes used include Standardized Root Mean Residual (SRMR), the Comparative Fit Index (CFI), the Tucker Lewis index (TLI), and the Root Mean Square of Approximation (RMSEA). According to Hu and Bentler [53], and more recently by Newson (2018), the suggested values for each index are the following: CFI > 0.95,TLI > 0.95, RMSEA <.06, SRMR < .08.
The exploratory and confirmatory factor analyses (EFA, CFA) of attitudes toward cycling present fit indexes’ values that fulfill the aforementioned requirements. In the cases of the EFA and CFA of attitudes toward walking, only TLI is slightly lower than the cutoff values (0.932), so overall the goodness of fit indices support the validity of the constructed scales. The structural equation model (SEM) of attitudes toward cycling, intention and actual use of transport modes also present fit indexes’ values that fulfill the aforementioned requirements. Finally, only the TLI values of the SEM of attitudes toward walking, intention and actual use of transport modes is slightly lower than the cutoff values (0.941), so overall the goodness of fit indices support the validity of the constructed scales.
Newsom, J.T. Some Clarifications and Recommendations on Fit Indices. Psy 523/623 Structural Equation Modeling, Spring. 2018. Available on line: http://web.pdx.edu/~newsomj/semclass/syllabus_18.pdf (accessed on 19-04-2019).
Marsh, H. W., Balla, J. R., & Hau, K. T. (1996). An evaluation of incremental fit indexes: A clarification of mathematical and empirical properties. In G. A. Marcoulides & R. E. Schumacker (Eds.), Advanced structural equation modeling techniques (pp.315-353 . Mahwah , NJ : Lawrence Erlbaum.
Jaccard, J., & Wan, C. K. (1996). LISREL approaches to interaction effects in multiple regression. Thousand Oaks, CA: Sage Publications.
With respect to methods, I think the acknowledgement of the data estimation process should have been included in that section instead of just as the next to last paragraph of the paper.
The acknowledgement of the limitations of data estimation process has also been included in the Measures subsection of the Methods section.
We acknowledge that the information regarding real use of travel modes was estimated from subjective self-informed data. Real use of travel modes was assessed from stated information given by respondents. Nevertheless, considering that the information was provided in percentage of use of each travel mode, only small differences could exist between observed and stated travel mode use.
I would like to know why the on-line survey did not occur in August. (I do not want to have to guess that it was because of vacations, for example). I also would like to know how the demographics of the sample compares to the study population as a whole.
Considering that the methods used to recruit participants (contacting by e-mail, collecting data online) are characterized by a low response rate, it was decided to stop the data collection process during August, which is a vacations period in Spain. Otherwise, the response rate would have negatively affected.
The demographics of the sample compared to the study population is presented in Table 1. The sample is representing well the population in terms of gender and education level, but people over 50 years old is underrepresented in the sample. People who is not employed neither study is underrepresented in the sample as well.
Table 1. Sample distribution.
Sample | Valencia area | ||
Gender | Male | 46% (754) | 48% |
Female | 54% (887) | 52% | |
Age | <30 | 41% (679) | 30% |
30-50 | 41% (671) | 31% | |
>50 | 18% (291) | 39% | |
Occupation | Student | 24% (390) | 5% |
Employed | 54% (893) | 25% | |
Others | 22% (358) | 70% | |
Education level | University | 51% (833) | 58% |
No university | 49% (808) | 42% |
Finally, there were very few grammatical issues -- mostly just awkwardness (Line 15: “to reduce”; Line 45: “but using”; Line 216: “The data collection step”; Line 269: “Exceptionally; Line 269 "AF_34" should be identified as a variable when being discussed as to why it was included)
We want to thank Reviewer 2 for these remarks. We have fixed them and others detected in a new exhaustive revision of the text.
Overall, I found this to be a good and valuable read.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The clarification of the authors and the extension of the text of the manuscript is now satisfactory.
