Influencing factors analysis of helmet wearing for electric bicycle riders based on ordinal multinomial logistic model

Abstract

1 Introduction

As a green transportation tool, electric bicycles can meet the diversified travel needs of urban and rural residents in China. However, with poor riding habits such as irregular riding and wearing no helmets, electric bicycle riders have caused many traffic accidents and have had seriously negative social impacts on society. In 2019, China's electric bicycle production was 27,077 million and social ownership reached almost 300 million, which came first in the global market, based on data from the Ministry of Industry and Information Technology of the People's Republic of China [1]. A large number of electric bicycles are mostly distributed or sold in second- and third-tier cities. The lack of safety awareness and bad riding habits of electric bicyclists, which have not yet been effectively regulated and strictly monitored by the management department, result in great uncertainty of potential collision risks. Yuan et al. [2] studied the relationship among occurrence time, vehicle type, riding behaviour, accidental rider's age and degree of injury, which provided a strong reference for standardizing riding habits. Further examination and analysis of the influencing factors of electric bicyclist helmet wearing has great research significance.

Electric bicycle crash data are important for exploring the causes of accidents. Gu et al. [3] conducted an empirical study on the development of traditional bicycles, electric bicycles and motorcycles in China from 1985 to 2019. It was found that due to the increasing demand for flexible transportation modes of short-distance and last-mile travel services, the use of electric bicycles has become more frequent. However, due to the increase in the proportion of electric bicycle trips and the rise in riding speed limits, safety issues have gradually become prominent. In 2018, the number of deaths in traffic accidents caused by riding electric bicycles in China reached 7469, occupying 11.71% of the total deaths, and 31,447 were injured, covering 15.00% of the total injuries. He et al. [4] collected alarming information from 2010 to 2017 in a city of China, and extracted information from accidents involving electric bicycles. The findings showed that accidents and injuries involving electric bicycles increased annually. In addition, Hertach et al. [5] surveyed 3658 e-bike users in Switzerland in 2016 and found that 17% of them had suffered at least one accident. Fyhri et al. [6] conducted a bicycle safety survey in Norway and found that the overall accident risk of electric bicycles is higher than that of traditional bicycles. King et al. [7] performed a retrospective study on the data of all injured patients related to the use of personal mobile devices (PMD) and electric bicycles in Singapore National Trauma Registry (SNTR) in 2016, and the results indicated that electric bicycles caused more serious injuries. Research on relevant data has shown that the proportion of electric bicycle accidents and their severity are at a high level.

In traffic accidents involving electric bicycles, the head is one of the main injured parts of casualties, which has a strong correlation with the severity of an accident. Du et al. [8] collected hospitalization information on electric bicyclists involved in road traffic accidents from hospital records in Suzhou, China. The results indicated that the number of injuries of electric bicyclists accounted for 57.2% among the total hospitalized people injured in road traffic accidents, and in these findings head injuries were common. Moreover, the chance of an electric bicycle collision at night causing a craniocerebral injury to the rider greatly increased. Zhou et al. [9] conducted a statistical study towards electric bicycle-related cases in 2008–2011 treated by Zhejiang Provincial People's Hospital in China and found that 32.7% of traffic accidents were related to electric bicycle cases, most of which had head injuries. Huang et al. [10] investigated the factors affecting the risk of a head injury in a two-wheeled vehicle collision, and the analysis results revealed that the risk of head injury increases with faster vehicle speed. When colliding with an SUV, due to the higher structural rigidity, the risk would increase. Yang et al. [11] collected data on electric bicycle-related injuries that occurred in a trauma centre from 2014 to 2016 and compared the injuries of adults to children (less than 18 years old) by electric bicycles. The analysis results showed that children were more vulnerable to head and face injuries. Therefore, it can be found that head injuries are more frequent and fatal in e-bike accidents. Because helmets are used to protect the head of riders, research on helmet wearing will motivate more people to wear helmets while riding and reduce the accidents of injuries and deaths.

The helmet is an important protection accessory for an electric bicycle rider, which can protect the head from direct external impact. However, due to the lack of self-safety protection awareness of riders and the incomplete laws and regulations, it is very common to ride electric bicycles without helmets. Xing et al. [12] employed observation methods to find that the helmet-wearing rate of electric bicycle riders in Anhui Province, China was only 5.90%. In October 2016, Chen et al. [13] randomly selected 1244 non-motorized bicyclists from seven non-central urban areas in Shanghai as research subjects. As the result showed, 71.8% of the total number were electric bicyclists and among them, 43.6% had never worn a helmet while riding, which was the dangerous behaviour with the highest incidence. Papoutsi et al. [14] reviewed the injured in electric bicycle accidents treated in the emergency room of a Swiss hospital from April 2012 to September 2013 and compared with the situation in China. It was found that the proportion of electric bicycle riders wearing helmets in Switzerland was about 75.0%, while in China it was just 9.0%. Yang et al. [11] conducted a road observation study in a rural area of Suzhou, China in September 2012 and found that only 2.2% of the observed 20,647 electric bicycles were wearing helmets. Hu et al. [15] discussed the risk factors related to injuries caused by electric bicycle collisions in Hefei, Anhui Province, China, and found that only 4.8% of all injured patients wore helmets. Capua et al. [16] conducted an observational study on patients with injuries related to electric bicycles and traditional bicycles who were in the emergency room from December 2014 to November 2015 and found that less than one-third of the patients had helmet use records. Besides, compared with bicycle riders (48%), there were significantly fewer electric bicycle riders (19%) who used helmets. In addition to electric bicycles, scholars have also investigated the casualties and helmet-wearing behaviour for other types of two-wheeled vehicles. Li et al. [17] collected bicycle injury information from 58 hospitals in Shanghai, as well as the causes of illnesses and deaths of residents in the city and related demographic data. The findings revealed that from 1992 to 2007, the death rate of bicycle injuries increased by 79.6%. Among all the casualties of bicycle accidents, the death rate of head injuries was 71.9%, and none of the riders wore a helmet while riding. There were great differences regarding helmet-wearing by two-wheeler riders in different countries and regions. Son et al.[18] analysed the rate of helmet use by Korean cyclists in 2013–2014 as well as demographic factors independently related to helmet use. The study found that of 4103 cyclists, 782 (19.1%) wore helmets. Among them, 21.1% of male riders used helmets, while the proportion of female riders using helmets was 15.5% . Debnath et al. [19] recorded videos of cyclists in 17 locations in Queensland, Australia and accurately obtained the information of riders using helmets. The results showed that the level of compliance with the law by riders in this area was very high. Among more than 27,000 riders, 98.3% wore helmets. The conclusion shows that the helmet is very important for the safety of riders, whether taking electric bicycles or bicycles.

Besides, scholars have analysed the helmet-wearing situation and related factors of electric bicycle riders. Yuan and Chen [20] compared the collision and injury characteristics as well as influencing factors of pedestrian, bicycle and electric bicycle accidents. Li et al. [21] conducted a study on 2044 accidents involving electric bicycle collisions and found that fewer riders were wearing helmets at the age of 60 or above at the time of accidents. Weber et al. [22] analysed road traffic accidents involving 504 e-bike cyclists in Switzerland in 2011 and 2012, and by comparing accidents in different areas between rural and urban environments, they found that helmet usage in rural areas was more than in urban areas. Wang et al. [23] completed 16,859 field surveys in nine regions (four urban areas and five rural areas) from 2015 to 2017, including 794 electric bike users from field surveys and 4426 users through face-to-face interviews and online surveys. The results showed that 74.2% of electric bicycle users deemed it necessary to wear helmets. Compared with other road users, electric bicycle users have a lower rate in correctly understanding electric bicycle violations. Ma et al. [24] reviewed the research results on the dangerous riding behaviours of electric bicycles and believed that improving the riding environment, safety awareness and training were effective measures to prevent electric bicycle accidents.

In summary, for ensuring the safety of electric bicyclists, wearing helmets is of great significance to reducing the probability of injuries and deaths of electric bicycle riders in traffic accidents. Through questionnaire research, data analysis and processing, as well as the construction of an ordinal multinomial logistic model, this paper analyses the current e-bike riders’ requirements and experience of wearing helmets, proposes the different feelings of the relevant crowd about helmet wearing and obtains valuable conclusions through discussion.

2 Methods

2.1 Data source

In this study, data are obtained through a questionnaire survey and used to analyse the influencing factors of helmet wearing of electric bicycle riders. Questionnaire survey is widely employed in the study of traffic safety such as riding behaviour, safety attitude and risk perception.

This study adopts a questionnaire structure combining non-scale questions and scale questions (shown in the Appendix). The non-scale questions include background information and characteristic behaviour questions. The characteristic behaviour questions include a survey on the riding habits of electric bicycle riders and their sentiments and preferences for helmets. The scale questions include 20 items about the actual scene and personal preference related to helmet wearing. Through this questionnaire, the background information, riding habits, sentiments related to helmet wearing and other key information are collected comprehensively and systematically.

In the process of data collection, the sample service of a questionnaire platform is adopted to collect the sample information of the relevant population online, which can ensure the reliability of the data. In order to screen valid samples, we add the brand of electric bicycle in the questionnaire. A total of 1000 samples are obtained through data collection procedure. Then according to the answers to the brand and other subjective questions, the invalid samples with irrelevant answers and no electric bicycles are removed. Finally, 976 valid samples are obtained.

2.2 Statistical methods

2.3 Model construction

In the scale questions of this study, dependent variables are divided into five levels, namely ‘strongly agree’, ‘agree’, ‘not agree and not disagree’, ‘disagree’ and ‘strongly disagree’, with an obvious ranking order. Therefore, the ordinal multiple logistic regression model is used for modelling analysis. The odds ratio obtained by modelling can intuitively and accurately acquire the influence of different levels of independent variables on a dependent variable.

3 Results

3.1 Descriptive statistical analysis

In this study, the background information of the sample mainly includes living area, gender, age, marital status and education level. These five items are also selected as independent variables for the chi-square analysis and correlation analysis. After statistical analysis, the total number of effective samples is 976. The statistical results of background information are shown in Table 1.

Therefore, the characteristics of the sample population can be obtained: (1) Urban population accounts for the majority (78.6%); (2) The gender distribution is relatively balanced, with 55.1% for males and 44.9% for females; (3) The number of people aged between 26 and 35 is the largest (34.9%), then decreases to both sides; (4) There are more married persons in the sample (58.2%); (5) In terms of educational level, the number of people with a bachelor's degree is the largest (42.2%), followed by junior college degree (23.7%) and high school degree (16.2%) and the number of people with lower and higher educational background is less.

In the non-scale questions, we mainly investigated the cycling habits of electric bicycle riders and their feelings about helmets. According to the research content, the following three questions are basically analysed: (1) Whether to wear a helmet while riding an electric bicycle; (2) Inconvenience of wearing the helmet (subjective); (3) Requirements for helmets while riding.

In question 1, 70.7% people choose to wear the helmet while 29.3% choose not to. Therefore, it can be confirmed that most people in the crowd of electric bicycle riders have awareness of wearing the helmet and will actively wear it while riding.

Question 2 investigates the inconvenience of wearing a helmet while riding. By frequency analysis, the keywords with the highest frequency consist of ‘line of sight’ and ‘field of vision’, followed by ‘uncomfortable’, ‘hot’ and ‘trouble’, etc. Therefore, it can be determined that the most noticeable inconvenience of the helmet lies in its impact on vision during cycling, followed by the sultry discomfort caused by wearing it.

Question 3 examines people's requirements for wearing helmets while riding, the options contain safe and standard, strong and durable, beautiful in appearance, affordable, high-end in quality, cool and ventilated, easy-to-carry and so on. The research results are shown in Table 2. Of the sample crowd, 87.2% choose safe and standard, accounting for the largest proportion, followed by strong and durable (70.3%), easy-to-carry (45.4%), affordable (44.8%), beautiful in appearance (43.6%), cool and ventilated (40.0%) and high-end in quality (28.8%). Therefore, it can be found that when choosing an electric bicycle helmet, people will pay more attention to whether the helmet is safe, strong and cheap, not to mention whether it is high-end.

Then, to explore the influence of the characteristic information on the helmet-wearing requirement during cycling, chi-square analysis is conducted on question 3 in terms of living area, gender, age, marital status and education level, and the results are shown in Table 3.

Age has a significant effect on the choice of strong and durable, cool and ventilated, easy-to-carry at the level of 0.01, while on the choice of beautiful in appearance at the level of 0.05. Similarly, marital status has a significant influence on the choice of affordable, high-end in quality at the level of 0.01, while on the choice of beautiful in appearance, easy-to-carry at the level of 0.05. Also, education level has a significant effect on the choice of strong and durable, beautiful in appearance at the level of 0.01, while on the choice of cool and ventilated at the level of 0.05.

Based on the chi-square analysis results, the following conclusions can be drawn: (1) With the increase of age, one's requirements on whether the helmet is strong and durable, beautiful in appearance, cool and ventilated and easy-to-carry all show a downward trend; (2) Compared with a married person, an unmarried person pays more attention to whether the helmet is beautiful, affordable, high-end in quality and easy-to-carry; (3) People with high education level have higher requirements on whether the helmet is strong and durable, beautiful in appearance, cool and ventilated than those with low education level.

Also, some questions related to the riding behavior are investigated in the non-scale problems, as follows. (4) How often do you ride your e-bike every day (frequency). (5) How long do you ride your e-bike every time (length). The statistical results of these questions are shown in Table 4.

We find that most of the riders ride e-bikes one to two times a day, and the length of each ride is about 10–30 minutes. Besides, as we can see, these two questions investigate the riding habits of e-bike riders in the sample population. Therefore, in the modelling analysis, these two questions together with the background information of the sample are modelled as independent variables to analyse the influencing factors of e-bike helmet wearing.

3.2 Model analysis

The ordinal multiple logistic regression models are established for scale questions. By using the reliability analysis to verify the reliability of the data, the scale is first simplified and dimensionless through factor analysis, which cannot only obtain intuitive and meaningful factors but also reduce the workload. Then, through correlation analysis, the basic correlation between factors and independent variables is determined. Finally, models are established according to the correlation results.

3.2.1 Factor analysis

Exploratory factor analysis is carried out on the whole scale, and four common factors are extracted, as shown in Table 5.

From Table 5, it can be obtained that the eigenvalues of the four common factors, from large to small, are 6.132, 3.661, 1.382 and 0.965. The eigenvalues of common factors after rotation are 3.896, 3.587, 3.050 and 1.607, and the variance explained rates are 19.482, 17.934, 15.248 and 8.036. The cumulative variance explained rates are 19.482, 37.416, 52.663 and 60.700. After rotation, the variance explained rate of each common factor does not differ too much, and the cumulative variance explained rate reaches 60%. Therefore, it can be considered that the common factor obtained by factor analysis can be used for replacement analysis of the original scale.

Table 6 shows the corresponding relationship between the original questions and common factors. Questions 1−6, 7−12, 13−17, 18−20 correspond to factors 1, 2, 3 and 4, respectively. According to the semantic analysis of the questions, the practical significance of each factor can be determined. The questions corresponding to factor 1 contain keywords such as danger and protection. Hence, factor 1 can be defined as the safety perception for the electric bicycle helmet of the sample. The questions corresponding to factor 2 focus on the investigation of the situation without the helmet. By flipping the dependent variable level, factor 2 can be defined as the practical perception for the helmet. The questions corresponding to factor 3 investigate habits and type preferences. It can be seen that the higher the sample crowd agree with the questions, the more they will know about the current situation of electric bicycle helmets, and they are more willing to choose to wear helmets. Therefore, we think that this represents one's overall cognition of the current situation of e-bike helmets in China. The higher the evaluation is, the higher the overall satisfaction is. Accordingly, factor 3 corresponds to the satisfaction perception for the helmet. Finally, the questions corresponding to factor 4 investigate helmet use in dangerous situations, which can be defined as the emergency perception for the helmet.

The four common factors, namely, the four dimensions of electric bicycle riders’ perception towards the helmet, are quantified and counted. According to the results of factor analysis, scores are divided into five levels of 1−5 for statistics. The statistical results of mean value and standard deviation are shown in Table 7.

According to the statistical results, the overall feeling of the sample population on the helmet can be obtained. Among the four dimensions, the average safety perception is 4 points, indicating satisfaction. The average practical perception reaches 3.5 points, which is between general and satisfaction, and has a large standard deviation. The average satisfaction perception is about 3 points, which is at the middle level. The average emergency perception is also about 3.5 points, between general and satisfaction.

3.2.2 Correlation analysis

Correlation analysis is conducted between the four factors and the characteristic information of the sample. The results are shown in Table 8.

After sorting out the data analysis results of Table 8, the following conclusions can be drawn. (1) Safety perception is significantly correlated with age and marital status at the level of 0.01. (2) The practical perception is significantly correlated with gender at the level of 0.01. (3) Satisfaction perception is significantly correlated with gender, age and marital status at the level of 0.01, and correlated with education level and frequency of riding at the level of 0.05. Among them, it is negatively correlated with education level and positively correlated with age frequency of riding. (4) Emergency perception is significantly correlated with age at the level of 0.01, and marital status and frequency of riding at the level of 0.05.

It suggests that differences in gender, age, marital status, education level and frequency of riding affect one's perception for helmets, and more accurate conclusions need to be confirmed by logistic regression models.

3.2.3 Logistic regression model analysis

In this section, logistic regression models are established for the above four perception dimensions based on correlation analysis results.

Test of parallel lines is performed before modelling to ensure the validity of the models. In the process of testing, it is found that only the modelling of practical perception with gender as an independent variable cannot pass the test. Hence, education level is added according to the correlation analysis results to assist the modelling. Finally, all four groups pass the test of parallel lines. The test results are shown in Table 9.

The P-value greater than 0.05 indicates passing the test of parallel lines.

Analysis of the modelling results:

1. Safety perception

    The result is shown in Table 10. According to the P-value of Wald significance, it can be determined that the independent variable age has a significant impact on one's safety perception for the helmet; marital status has no statistically significant effect on safety perception.

    Compared with people aged 56 and above, people aged 18−25, 26−35, 36−45 and 46−55 are 1.3 times (P = 0.475), 1.8 times (P = 0.100), 2.0 times (P = 0.55) and 2.3 times (P = 0.26), respectively, more likely to have at least a grade higher safety perception for the current helmet.

2. Practical perception

    The result is shown in Table 11. The independent variable gender has a significant impact on practicability, while the education level has no statistical significance on practicability.

    While other conditions are equal, the probability that the male population is at least one grade higher than the female population on the practical perception for the current helmet is 0.77 times (P = 0.026).

3. Satisfaction perception

    The result is shown in Table 12. Gender and education level have significant influence on one's satisfaction perception for e-bike helmet. Age, marital status and frequency of riding have no statistically significant effect on satisfaction perception.

    Therefore, one's satisfaction perception for the helmet is related to gender and education level. While other things are equal, the probability that the male population is at least one grade higher than the female population on the satisfaction perception for the current helmet is 1.48 times (P = 0.001). Compared with people with a graduate degree or above, people with education levels of primary school and below, junior high school, senior high school, junior college and bachelor's degrees are 1.60 times (P = 0.418), 2.39 times (P = 0.003), 1.47 times (P = 0.139), 1.67 times (P = 0.040) and 1.52 times (P = 0.073), respectively, more likely to have at least a grade higher satisfaction perception for the current helmet.

4. Emergency perception

    The result is shown in Table 13. The P values of the three independent variables are all at a higher level, indicating that age, marital status and frequency of riding have no significant influence on emergency perception in this model. Therefore, in emergency perception, no influencing factors are found.

Fig. 1 shows the relationship between the influencing factors involved in this paper and the four dimensions of one's feelings towards electric bicycle helmets.

Fig. 1.

Relationship between the factors and the feelings for helmets.

Open in new tabDownload slide

4 Discussion

In the previous section, the questionnaire data are analysed and modelled to obtain the overall situation and influencing factors of helmet wearing preference of electric bicycle riders.

In the analysis of non-scale questions, it is understood that the main inconvenience of wearing helmets lies in vision and the feelings of head warmness and uncomfortableness. Therefore, it can be speculated that the common problems of electric bicycle helmets at present are the influence of vision and head warmness.

Then, in the survey on the requirements of the helmet, 87.2% of sample crowd choose safe and standard and 70.3% choose strong and durable, which are the two most important items. This indicates that riders mostly focus on safety and protection by wearing helmets, which is in line with the concept of helmet design. According to the chi-square analysis, one's age, marital status and education level all affect requirements for the helmet. Therefore, the helmet can be designed with different characteristics, in a way of meeting demands of all kinds of people. For example, with the age growth, an older crowd will have lower requirements as compared to the young crowd that it must be strong and durable, beautiful in appearance, high-end in quality, cool and ventilated and easy-to-carry. Therefore, it only needs to be designed according to the common requirements, and the safety and durable helmet can meet the needs of the older crowd.

In the analysis of the scale questions, four aspects have been classified: namely, safety perception, practical perception, satisfaction perception and emergency perception. According to the statistical results, people are satisfied with the safety of the helmet, slightly satisfied with practicability and emergency, but feel general about the overall satisfaction. This indicates that the electric bicycle helmet for riding is generally attractive at the current stage. According to the modelling results of the ordinal multiple logistic models, the influencing factors of people wearing helmets can be determined.

Safety perception is affected by age. People over 56 years old have the worst safety perception for the helmet, followed by people aged 18−25 years old (1.3 times than people over 56), 26−35 years old (1.8 times), 36−45 years old (2.0 times) and 46−55 years old (2.3 times). It can be found that the influence of age on safety perception exists with a large gap between people aged 46−55 years old and people aged over 56 years old. With the increase of age, one's safety perception for the helmet is gradually growing, but when they enter the old age stage, their safety perception gradually deteriorates.

The practical perception is influenced by gender. In the previous data analysis, men have lower satisfaction than women (0.77 times), which may be related to the psychological expectations of men and women. Men are more confident about riding electric bicycles, so they have a lower practical perception for helmets.

Satisfaction perception is influenced by gender and educational level. Men have higher levels (1.48 times) of satisfaction perception, as opposed to practical perception. Regarding education level, people with junior middle school education background are the most satisfied (2.39 times), followed by people with a junior college (1.67 times), primary school (1.60 times), bachelor (1.52 times) and high school (1.47 times) education. People with graduate education are the most dissatisfied. Generally speaking, the higher the education level, the lower the satisfaction level. The reason behind this previous argument is that higher education level will enhance one's requirements for helmets, leading to a decrease in satisfaction.

5 Conclusions

As a green means of transportation to meet the diversified travel needs of urban and rural residents in China, the electric bicycle brings convenience but also huge traffic risks for people. The head, as the main injured part in electric bicycle accidents, needs to be protected with effective and convenient measures. Therefore, it is of great significance to investigate helmet wearing and analyse the influencing factors. Through a questionnaire survey, this paper proposes four dimensions of riders' intuitive feelings towards the helmet, establishes an ordinal multiple logistic regression model and analyses the influencing factors at each dimension. At the same time, the paper also carries out statistical analysis and influence factor analysis on the inconvenience of wearing helmets and the requirements for helmets, and finally presents some valuable conclusions.

First, through the analysis of non-scale questions, the conclusions are as follows.

People think that the main inconvenience of wearing a helmet when riding an electric bicycle is affected vision, and the second is that they feel warm and uncomfortable.

The most essential part of the rider's requirements for electric bicycle helmets is safe and standard, followed by strong and durable and so on. It is affected by one's age, marital status and education level.

Then, the following conclusions are drawn by modelling the scale questions.

About the safety perception of helmets, people are satisfied with it at present, which is affected by age. People aged 46−55 are more satisfied, people aged 56 and over are less satisfied. In detail, people aged 18−25, 26−35, 36−45, and 46−55 are 1.3 times (P = 0.475), 1.8 times (P = 0.100), 2.0 times (P = 0.55) and 2.3 times (P = 0.26), respectively, more likely to have at least a grade higher safety perception for current helmets than those aged 56 and over.

About the practical perception of helmets, people's perception is between general and satisfaction, which is influenced by gender. Women are more satisfied with the practical perception than men. In detail, men are 0.77 times (P = 0.026) more likely than women to feel at least one grade higher about the practical perception.

About the satisfaction perception of helmets, people's perception is general, which is influenced by gender and education level. Men are more satisfied than women this time. With the improvement of education level, people's satisfaction perception shows a downward trend. In detail, compared with women, men are 1.48 times (P = 0.001) more likely to feel at least one grade higher about the satisfaction perception. While compared to people with master's degree, people with primary school, junior high school, senior high school, junior college and bachelor's degree education are 1.60 times (P = 0.418), 2.39 times (P = 0.003), 1.47 times (P = 0.139), 1.67 times (P = 0.040) and 1.52 times (P = 0.073), respectively, more likely to feel at least one grade higher about the satisfaction perception.

People's emergency perception for helmets is between general and satisfaction. The five influencing factors mentioned in this paper do not have an impact on this dimension.

ACKNOWLEDGEMENTS

This study is jointly supported by The National Natural Science Foundation of China (Grant No. 52072214 and Grant No. 71871123), Global Road Safety Partnership (GRSP) (Grant No. CHNXX-RD16-1188). Besides, the authors would like to acknowledge the support of all the participants in the survey.

Conflict of Interest

The authors declare that they have no conflicts of interest.

References

Appendices:

Survey on Helmet Wearing by Riders of Electric Bicycles

(Note: This questionnaire is completely anonymous. All survey data are for research purposes only and are strictly confidential. Thank you very much for your participation!)

Questions about basic Information: