Defining Psychological Factors of Cycling in Tehran City

Abstract

Studying active transportation (walking or cycling) is widespread in American and European research. Studies which include latent variables (LV) are growing to identify the exact results of determining the strategies to increase the utility of active transportation (AT). LVs help us conduct more accurate research. LVs are defined as psychological factors such as feeling safe while you ride at night, and thus they are not subjective and hard to understand, but very important to consider in order to increase the utility of using AT modes. In the present paper, most of the previous studies on cycling were reviewed. Different variables, including subjective and LVs, were included to maximize using the bicycle utility and introduced to have better sight for future researchers to deal with modeling AT mode choice. This study applied the latent class analysis to a sample of 345 survey respondents in Tehran, the capital city of Iran, exploring the variables affecting cycling behavior and a confirmatory factor analysis, and a structural equation modeling (SEM) was developed. Results show the importance of having a ‘will’ for using a bicycle, especially in difficult situations, and in view of cultural barriers that affect women cyclists.

1. Introduction

Cycling is a form of healthy exercise that provides a way to partake in physical activity as individuals. Riding to school or work is one of the most time-efficient ways to integrate exercising into a daily routine. On the other hand, fuel consumption in the year 2021 was around 85 million liters per day in Iran, which subsequently made Tehran the most polluted city in the world for 2 days in 2022; the city mostly faced polluted days during winter. These are reasons for policymakers to think of a solution. Countries such as the Netherlands have learned to reduce these problems by investigating active transportation (walking or cycling [1]). Active transportation (AT) helps provide substantial health benefits to societies [2,3], reduces the need for new parking lots and roadways, and reduces congestion.

Much research was conducted on ways to increase the use of AT. For this purpose, lots of variables were recognized and different categories constructed, some of them more complete [4,5], including individual characteristics (variables such as age, gender, etc. [6]), household characteristics, trip characteristics, built environment (variables such as density, bus stops, etc. [7]), seasonal and weather characteristics, and work conditions. Most research has divided the variables that affect using AT into objective and subjective variables [8]. In recent decades, many studies have led to the study of latent variables (LV) to be more precise in the results of findings; however, they have not all led to consistent results because of varied LV data sources and methodologies. A latent variable is a variable that is not directly measured [9]. The field of AT has recently become one of the attractive fields of study but still needs to be examined. Earlier studies show that the cycling behavior of regular riders (private bicycle riders) is influenced by environmental factors, including population density, land use mix, green space, cycling facilities, and safety [10]. In addition, the Tehran transportation master plan in 2008 was nearly the first governmental study with a section on active transportation. After that, some studies were conducted on AT and sustainability. The Tehran Municipality’s most recent research on green transportation has a solid vision for active transportation. According to the study’s findings, reducing traffic is the key to boosting cycling rates; the study also affirms that the municipality should act quickly to provide infrastructure for utilizing e-bikes and e-scooters.

This study, therefore, aims to (1) identify the LVs with the most effects on cycling behavior to understand different ways of increasing cycling rate, mostly those which have not been identified in previous studies, and (2) providing policies and suggestions for transportation planners. The number of women workers in Iran has increased from 10% of all workers in 2011 to 16% in 2021, which indicates an increase in women’s home-based trips. For this reason, in this article, (3) we conducted an investigation into the practical hidden factors of “limitations from the family” and “concern about dignity” for the rate of cycling. For this purpose, we adopt a 5-scale Likert analysis for different LVs by a survey and estimate parameters and validate them using confirmatory factor analysis (CFA) and create a structural equation model (SEM).

However, no previous study attempted to determine latent variables that relate to family limitations or get the unpleasant feeling of cycling from other people factors that we have evaluated. This study aims to answer such questions. In the following sections, some of the important issues reported by the literature are reviewed.

This paper is organized as follows. Section 2 describes the previous studies in variables affecting cycling, and Section 3 describes the methodologies used. Section 4 analyzes the data, and Section 5 represents the empirical results. Section 6 offers a discussion about the results and compares them with previous study results. Section 7 concludes the paper and identifies future research directions.

2. Literature Review

2.1. Socio-Economic Characteristics

Age and gender are two of the variables most associated with bicycling. Age and bicycle travel have been the subject of several studies, with varying results. Although some research has identified a negative relationship between youth and cycling [11], most investigations have shown that youth is a factor positively influencing the chance of cycling [12,13,14,15]. Individuals under the age of 18 do not use private cars, which is one of the important reasons for increasing the use of bicycles to reach their destination. Furthermore, young individuals usually use a bicycle for recreational purposes. Bicycling rates are also affected by gender. Many studies determined that males are more likely to use bicycles than females [11,12,13,14,15,16,17,18,19], but some studies contradict this finding [20,21]. This issue cannot be only affected by gender. For example, one of the most important reasons why women do not use bicycles is the possibility of bathing at the destination, which is not very important for men [22]. Some studies determined that in most cases, those cycling to work are male [23,24]. Some studies indicate income as a positive factor to increase the rate of cycling, where the higher the income, the higher the cycling rate [25,26]. However, another study points out that a higher level of education would result in a higher salary, and people with higher income are less likely to bike [27]. At the same time, other studies assert the opposite, namely, that individuals with a higher level of education cycle more [24,28].

Emond and Handy [29] determined that male students are much more likely to a bicycle than female students. The other factor that prevents individuals from cycling is obesity.

2.2. Trip Characteristics

Many studies found a negative effect of long travel time on the likelihood of cycling [30,31,32,33,34]. This research came to the conclusion that cycling is less competitive than other means of transportation due to high trip distances and times. According to the proportion of employees that commute by bicycle as their main mode of transportation, Zahran et al. selected the top 25 counties in the United States. The predicted number of bike commuters declined by 5.8% for every extra minute of the average route time, based on the study’s findings [31]. Another study found that reducing the travel time of bicycling by increasing the provision of route facilities causes more likelihood of bicycling [35]. According to previous research, another important factor in the probability of choosing AT is trip distance. Most studies found that the increase in travel distance reduces the likelihood of bicycling [4,13,18,23,33,34,36,37]. Further, the research found that short distances are an advantage of using a bicycle, and will reduce the total vehicle/kilometer in Belgium [27]. Some other studies acclaimed that the recreational and social purpose of bicycling has a significant effect on the likelihood of bicycling [13,38], but other studies may not see this way [21]. Few studies have found the effect of low-cost transportation in increasing the cycling rate [33,39].

2.3. Built/Natural Environment

Some studies found positive signs of the effect of high population density on the likelihood of cycling [37,40,41,42]. Some studies show the employment accessibility-density that has a positive effect on cycling in business districts [28,43], but some other studies disagree with these results [23]. Proximity to services/recreational areas could be beneficial and make cycling more attractive [28]. Traffic lights are another reason to prevent the individuals from biking [39]. Some other studies noticed low bicycle usage in areas having high slopes [44,45].

According to the findings, rainy and windy weather has a great negative effect on cycling [35,46]. In contrast, sunny days, especially in summer, attract individuals to bike [33,37]. There are many studies representing the positive effect of cycling infrastructure on increasing the cycling rate [30,47,48].

2.4. Work Conditions

According to certain studies, the chance of riding a bicycle is significantly influenced by one’s work position, suggesting that students and the unemployed would be more likely to utilize bicycles for trips [6,47,49,50]. Another variable is hours of work, which would decrease using a bicycle as the work hours raise [48,51]. Another study shows that informal workers cycle more frequently [26].

2.5. Latent Variables

There are many other subjective variables called latent variables (LV) that were not studied that we identify and analyze. LVs cannot be directly measured and need an indicator to estimate them. Many studies, especially in the last decade, worked on these variables [52]. Another research study introduced bicycle security and cost importance as new LVs affect the utility of bicycle use [22]. Another study talks about the positive effect of convenience [53,54]. Social norms [55] and cycling ability [12,56] are other variables involved in bicycling. Researchers have spoken about how cyclists may perceive safety and comfort in a variety of scenarios that might be stressful for them and make riding less likely [57,58]. Studies on family interactions and psychological issues are few. For instance, we could only find one paper about the positive effect of family relationships on cycling [20]. This paper finds and analyzes other psychological factors affecting the cycling rate.

Table 1. Key variables examined in previous studies.

Reference Number	Socio-Economic								Trip				Built/Natural										Work Condition	Latent
	Age	Gender	Student	Ability to Bath	Cycling to Work	Obesity	Income	Education	Travel Time	Travel Distance	Cost	Recreational/Social	Density	Air\Noise Pollution	Bus Stop	Infrastructure	Season	Population Density	Slope	Traffic Light	Wind	Accessibility Density	Employment Status	Convenience	Dependent to Car	Social Norms	Cycling at Night	Bicycle Security	Perception of Safety	Family Relationship	Cycling Alone	Cycling Ability
4	✓	✓
5	✓	✓
6	✓	✓																					✓
7													✓		✓
11	✓	✓
12	✓											✓										✓
13	✓	✓																														✓
14	✓
15	✓
16		✓
17		✓								✓		✓
18		✓
19		✓
20		✓
21		✓
22		✓										✓
23		✓
24		✓																												✓
25				✓
26					✓					✓												✓
27					✓			✓
28							✓
29							✓
30							✓	✓		✓
31					✓			✓														✓
32		✓	✓
33									✓							✓
34									✓
35									✓
36									✓	✓	✓						✓
37									✓	✓
38									✓												✓
39										✓
40										✓							✓	✓
41												✓
42											✓									✓
43																		✓
44																		✓
45																		✓
46																			✓
47																					✓
48																✓							✓
49																✓
50																							✓
51																							✓
54																								✓
55																								✓
56																										✓
57																																✓
58																													✓
60																													✓
64						✓
65																									✓
66																											✓
67																											✓
68																															✓
69													✓	✓
70													✓	✓

lists the key variables that were studied in previous research.

3. Methodology

3.1. Area of Study

Currently, in Iran, cycling as a means of urban transportation has a relatively low social image. In Iran, apparently, non-recreational use of bicycles has decreased. Fewer workers are seen arriving at their place of employment on bicycles than in prior years, which suggests a negative shift in their utilization. In recent years, a 24-km dedicated cycling route was built, and Tehran Municipality, as the trustee of non-motorized transportation infrastructure development, is developing it, although the share of cycling among other modes of transportation in Tehran is less than 1%.

The study area is Tehran, the capital of Iran, which is one of the most populated cities in the Middle East. The industrial growth of this city means that, based on official meteorological data, there were only 26 days of clean air last year. Another big problem of this city is its heavy traffic, which despite creating BRT lines and new metro stations has not been solved yet.

As mentioned, the study area is in the city of Tehran, although a smaller area in the center of the city is considered for face-to-face interviews with the interviewees, which can be seen as the red circle in Figure 1. The reason to choose this range is that it includes different levels of society, including employees and students. This range includes a complete sample of public transportation systems in the country, which makes the statistical sample more accurate.

In the next sections, we will examine the model, whose methodological steps are presented in Figure 2.

3.2. The Survey

Peer reviewers with expertise in active transportation systems and one psychologist conducted the survey. Approximately six more latent variables were left out of further research because of their repetition and difficulty in establishing indicators. The survey contains different sections: individual characteristics, aim and subjective variables, and latent variables. Regarding a confidence interval of 5 and a confidence level of 95 percent, the sample size for the city of Iran with a population of about 8 million is 384 individuals; after collecting the answered surveys and removing incomplete questionnaires, 345 questionnaires could be used for the analysis, which is close to this sample size. This amount of sample size is near the suggested Cochran’s formula, which is a good way of choosing a sample size [60]. The questionnaire was created via an online questionnaire site and published in Persian and English. The link to this questionnaire was sent to 2128 individuals via text messages and phone calls that were made via an advertising company to people who had the most traffic (no matter which mode of transportation) in the study area in the last month, or their workplaces or homes were in that area. The people’s mobile phone service provider (MTN and IRANCELL), which supplied the location data, sent this information to the individuals. On the other hand, 221 people were asked to complete the survey face-to-face in the CBD (central business district) as mentioned in the previous section over two days, on 28–29 April 2021, between 8:00 a.m. and 11:00 a.m. and 3:00 p.m. and 6:00 p.m. Questionnaires were provided to the interviewees online and through tablets and mobile phones.

4. Analysis

The purpose of data analysis and statistical processes is to answer research questions, hypotheses, or objectives. In this section, the obtained data are described and analyzed concerning each question, aim, or hypothesis. Large, complicated, and often even unintelligible data sets must be transformed into understandable units, patterns, and indications in order for an analysis to be effective. The collected data is presented and analyzed as a table or graph.

The statistical results are presented in this section, and two descriptive and inferential sections. In the descriptive part, demographic and main variables were described using frequency and frequency percentage, mean and standard deviation, and in the inferential part, the validity and reliability of the questionnaire were checked with confirmatory factor analysis, the relationships among the variables with the Pearson correlation test, and the research model. It was tested using the structural equation modeling technique. Data analysis was done using SPSS and Amos software. The maximum level of alpha error for hypothesis testing was determined as 0.05 (p < 0.05).

4.1. Individual Characteristics (Objective Variables)

Table 2. Demographics of obtained data.

		Number	Percent (%)
Gender	Male	189	45.2
	Female	156	54.8
Marital status	Single	176	51
	Married	169	49
Employment status	Employed	149	43.2
	Student	74	21.4
	Workingstudent	79	22.9
	Unemployed	43	12.5
Education	Under-diploma	39	11.3
	Diploma	68	19.7
	Bachelor	98	28.5
	Master	114	33
	PhD	26	7.5
Income	No income	15	4.3
	Less than 75$	53	15.4
	75–200$	128	37.1
	200–300$	112	32.5
	More than 300$	37	10.7
Car ownership	Do not have	116	33.6
	Having 1 car	72	20.9
	Having 2 car	98	28.4
	3 cars or more	59	17.1
Purpose of trip	Work	70	20.3
	Education	63	18.3
	Leisure	59	17.1
	Socializing	78	22.6
	Shopping and weekend	75	21.7
BMI	Less than 20	28	8.1
	20–25	126	36.5
	25–30	124	35.9
	More than 30	67	19.5

lists many of the individual characteristics that were provided. According to the results, 45.2% of respondents were men and 54.8% were women. This finding is consistent with Tehran municipal data from 2018, which showed that 50% of the population is made up of men and 50% of women. This finding confirms that the sample used to represent the population is good because it closely matches the statistics from the municipal data.

4.2. Latent Variables (Subjective Variables)

According to

Table 3. Defining main variables with Skewness and kurtosis values.

Main Factor	Standard Deviation	Average	Skewness	Kurtosis	Definition
Cycling	2.55	0.79	0.085	−0.227	Cycling behavior
Cost	2.19	0.81	0.595	−0.037	The cost of cycling
Built/natural environment	3.52	0.98	−0.813	0.134	Variables such as air/noise pollution that affect cycling rate
Will	3.59	0.88	−0.65	0.456	Influence of will on desire for cycling
Dignity	3.4	0.87	−0.518	0.061	Influence of dignity on desire for cycling
Limit by family	3.53	1.10	−0.605	−0.409	How the restrictions imposed by the family affect the cycling of family members
Unpleasant feeling	2.45	1.11	0.667	−0.137	How your feelings influence on cycling rate
Safety at night	2.37	0.92	0.488	−0.438	Influence of safety issues on cycling rate

, the main variables were described using mean and standard deviation statistics. The average range of all variables is from a minimum of 1 (strongly agree) to a maximum of 5 (strongly disagree) based on a 5-Likert scale of answers.

Moreover, the lowest average was the cost factor with a value of 2.19, and the highest average was the will factor with an average of 3.59.

Skewness and kurtosis values were used to determine the state of data distribution (normality). The results are reported in Table 3. Regarding the skewness and kurtosis, if the values of these statistics are between −2 and +2, it indicates the normality of the univariate distribution [61].

The results of Kolmogorov–Smirnov test showed that all research variables have a normal distribution. Examining the values of skewness and kurtosis shows that according to the fact that the values of kurtosis and the values of kurtosis of all variables were obtained in the range from +2 to −2, it can be concluded that all the variables have a normal or close to a normal distribution, and it can be tested used parametric methods (Pearson’s correlation and structural equation modeling with Amos software or covariance-based methods).

5. Result

The validity and reliability of the questionnaire as well as the test of relationships among the variables and the test of hypotheses were investigated using the confirmatory factor analysis test, Pearson correlation test, and structural equation modeling.

5.1. Confirmatory Factor Analysis

The validity of survey was studied using factor load indices, t-value, combined reliability, and mean-variance extracted, and the results are shown in

Table 4. The results of confirmatory factor analysis: checking the validity and reliability of questionnaire variables.

Main Factor	Item	F.L	AVE	C.R	Cronbach’s a
Cycling	Travel time by bicycle per trip	0.53	0.34	0.71	0.61
	The distance you cycle	0.64
Cost	The cost of cycling (buying a personal bike) is one of my main reasons for not using a bike.	0.66	0.52	0.73	0.68
	The cost of cycling (the cost of using a shared bike) is one of my main reasons for not using a bike	0.78
Built/natural environment	In secluded places, I am more interested in cycling	0.81	0.53	0.79	0.77
	I also use a bicycle on days when the weather is polluted	0.78
	I also use a bicycle in places where there is a lot of noise pollution	0.58
Will	If there is something I don’t like, I will still deal with it	0.52	0.44	0.72	0.68
	If I want to understand something, I will definitely try to understand it	0.78
	If I start something, I will definitely finish it	0.67
Dignity	I always appear outside the house in formal clothes	0.82	0.47	0.81	0.8
	I never leave the house in sports clothes, even to exercise	0.79
	I think cycling is for those who don’t have a car	0.83
	I think that cycling is reserved for the wealthy classes of society with lower job stress	0.43
	I think cycling is for those who are in their teens or younger	0.44
Limit by family	My family does not pay me to buy a bicycle	0.89	0.77	0.91	0.91
	I feel that my family is upset about my cycling	0.87
	I feel that my family will not allow me to ride a bicycle	0.87
Unpleasant feeling	Other people pay more attention to me when I ride a bike (in a bad way)	0.85	0.84	0.95	0.96
	While cycling, other people make fun of my clothing	0.92
	When cycling, other people object to my clothing	0.97
	While cycling, other people make fun of me because of my gender	0.93
Safety at night	Accidents are more likely to occur while cycling at night	0.89	0.61	0.92	89/0
	There is a higher chance of harassment while cycling at night	0.96
	There are few police at night in the city	0.93
	More thefts occur from cyclists at night	0.54
	I have more anxiety while cycling at night.	0.46

C.R = composite reliability, AVE: average variance extracted, FL: factor loading.

. Factor load values, and other indicators are taken from Figure 3 or the research model test. Factor loadings are computing by calculating the correlation value of the indicators of a structure with that structure. If this value is equal to or greater than 0.40, it indicates that the variance between the structure and its indicators is greater than the variance of the measurement error of that structure, and the validity of that measurement model is acceptable. If the researcher encounters values less than 0.40 after calculating the factor loadings between the structure and its indicators, they should modify those indicators (questionnaire questions) or remove them from the research model. The minimum value of factor load was considered to be 0.40.

Cronbach’s alpha and composite reliability were used to measure reliability. Because Cronbach’s alpha is a traditional criterion for determining construct reliability, the partial least squares method employs a more modern criterion called composite reliability rather than alpha. The superiority of composite reliability over Cronbach’s alpha is that it calculates the reliability of constructs not in an absolute way but according to the correlation of their constructs with each other. If the composite reliability value is greater than 0.7, it indicates the appropriate internal stability for the measurement models [62]. However, some sources consider a range of 0.5 to 0.7 as acceptable for exploratory research, if the sample size is small or the number of items on the scale is low [63].

Average Variance Extracted Index (AVE) was used to check the convergent validity. This index measures the amount of variance that a latent variable obtains from its indicators. The indicators of a certain construct should converge or share a significant fraction of the shared variance, according to the concept of convergent validity. Higher values of this index reflect the convergence validity of the intended structure. The extracted mean-variance index has a value between 0 and 1.

Based on the number of factor loadings obtained for all questions, which is more than 0.40 and at a significance level of less than 0.05 (p < 0.05) (all t values are greater than 1.96 done), we conclude that the validity of all questions in the questionnaire is confirmed. All questions have a factor load greater than 0.40, which is significant (p > 0.05), and the validity of all questions is confirmed. The questions No. 4, 5, and 6 of the built/natural environment variable were removed from the survey and analysis because of the factor loading of less than 0.40 and weak validity.

The survey’s dependability is statistically supported by the composite reliability value of more than 0.70, which is the adequate and acceptable value. Cycling is the primary component with the lowest combined reliability of 0.71, while the unpleasant feeling variable is the one with the greatest combined reliability of 0.95. Moreover, the value of Cronbach’s alpha ranges from a minimum of 0.61 for the variable of bicycle use to a maximum of 0.96 for the variable of unpleasant feeling, which is an acceptable range, and demonstrates that the reliability test using the internal correlation method (Cronbach’s alpha) validates the questionnaire’s reliability. Three variables of cycling, the cost factor, and the will factor have Cronbach’s alpha value of less than 0.70, but because the number of questions of these three variables was small and included only two or three questions (the value of Cronbach’s alpha coefficient is affected by the number questions are a variable), and because the questionnaire was made by the researcher, the reliability of these three variables was also confirmed.

The average variance extracted, which measures the convergent validity of the variables, was obtained from a minimum of 0.34 for the cycling variable to a maximum of 0.84 for the unpleasant feeling factor. The results show that the convergent validity values of the three variables of cycling, the will factor, and the dignity factor is of moderate value, and the convergent validity of other variables is of high value. In general, the results show the validity and reliability of the research survey.

5.2. Correlation of Variables

We used Pearson’s correlation test to investigate the correlation among the research variables. The test showed that the dependent variable cycling has a significant relationship with all the independent variables in

Table 5. Pearson correlation matrix between main variables and divergent validity.

Variables	Cycling	Cost	Built/Natural Environment	Will	Dignity	Limit by Family	Unpleasant Feeling	Safety at Night
Cycling	0.58
Cost	−0.34 **	0.72
Built/natural environment	−0.49 **	0.21 **	0.73
Will	0.57 **	−0.23 **	−0.31 **	0.66
Dignity	−0.45 **	0.06	0.27 **	0.18 **	0.66
Limit by family	−0.39 **	0.33 **	0.45 **	−0.30 **	0.31 **	0.88
Unpleasant feeling	−0.19 **	0.24 **	0.05	0.08	0.33 **	0.48 **	0.92
Safety at night	−0.25 **	0.07	0.25 **	0.07	0.24 **	0.37 **	0.16 **	0.78

** = p ≤ 0.01.

. The significance level is showed by p < 0.05, meaning that the results are statistically significant. The findings showed that the direction of the relationship between the factor of will and cycling is positive and it shows that by strengthening and improving will power, the amount of cycling increases. Except for will, the cycling has a negative correlation with all other factors. Examining the strength of correlations revealed that cycling had the largest relationships with the factors of will (0.57), built/natural environment (0.49), and will (−0.49). Examining the intensity of the correlation among independent variables showed that the correlation between independent variables is moderate or weak and there is no correlation greater than 0.60 between independent variables. Consequently, it can be concluded that there is no problem of multiple collinearities among independent variables and it can be used with multivariate methods, such as structural equation modeling.

5.3. Examining Fit Indices

The fit indices of model are reviewed in

Table 6. The fit indices of the research model.

Interpretation	Result	Range Acceptable	Indicator
Acceptable fit	0.91	>0.90	GFI (index goodness-of-fit)
Acceptable fit	0.072	<0.80 (Smaller than 0.80)	RMSEA (root mean square error of approximation)
Acceptable fit	0.93	>0.90 (Greater than 0.90)	CFI (fit index comparative)
Acceptable fit	0.92	>0.90 (Greater than 0.90)	NFI (fit index Normed)
Acceptable fit	0.89	>0.90 (Greater than 0.90)	IFI (fit index Incremental)
Acceptable fit	0.71	>0.50 (Greater than 0.50)	AGFI (goodness-of-fit index Adjusted)
Acceptable fit	0.56	>0.50 (Greater than 0.50)	PGFI (goodness-of-fit index Parsimonious)
Acceptable fit	2.56	5 ≥ Indicator ≥ 1 (between 1–5)	chi-square/DF (Chi-square ratio on the degree of freedom)

. After estimating the parameters of the model, the question that arises is to what extent the model is compatible with the relevant data. Only by looking at the model’s fit is it possible to provide an answer to this query. As a result, when analyzing structural equations, the researcher must first estimate the parameters and then interpret them before determining whether the model is appropriate. An important point that should be considered to interpret the fit indices is that the fit of the model should be evaluated via different methods and criteria to check its fit from different dimensions.

In general, by evaluating all fit indices in Table 6, it can be concluded that obtained fit indices show an acceptable and appropriate fit of the data with the model, and the model can be fitted according to the obtained fit indices considered acceptable.

6. Discussion

Results shown in

Table 7. SEM test results (table of coefficients).

Type of Relationship			Standardized Beta	Un−Standardized Beta	Standard Error	T Value	p Value
Cycling	←	Built/natural environment	−0.53	−0.31	0.045	6.82	<0.001
Cycling	←	Cost	−0.25	−0.2	0.066	2.98	0.003
Cycling	←	Will	0.71	0.49	0.067	7.33	<0.001
cycling	←	Dignity	−0.23	−0.21	0.063	3.41	<0.001
Cycling	←	Limit by family	−0.34	−0.18	0.034	5.35	<0.001
Cycling	←	Unpleasant feeling	−0.19	−0.1	0.029	3.35	<0.001
Cycling	←	Safety at night	−0.26	−0.29	0.073	3.93	<0.001

show that the will factor has a great influence on the cycling rate (Beta = 0.71). This seems logical because in some hard situations such as rainy/snowy weather, the will factor plays a significant role to use a bicycle instead of all other modes of transportation such as private cars with higher comfortability. Moreover, the cycling rate among obese individuals is lower [64] which might mean they have a lower will for cycling because they are not able to lose weight. Will is the only direct factor that positively affects cycling. Another issue that may affect the cycling will is dependent on the availability of a car, as it is not easy to distance yourself from the comfortability of private car [65].

The model estimates showed that feeling safe while driving at night is a significant factor for increasing the cycling rate. This factor shows that fear of accident or incidents as one the most significant issues for individuals to cycle at night which is in line with the finding in some previous studies [66,67]. Previous research has stated that cycling alone at night can be unsafe and suggests that group cycling can help increase the desire to cycle [68]. The results showed that the lack of police and a fear of theft at night deter people from cycling, and group cycling can be one of the solutions.

The built/natural environment factor has the most negative effect on cycling. It means that air/noise pollution and crowded areas cause a lower cycling rate, which is in line with the results of previous studies [69,70]. However, cycling is a strategy for emission and road traffic reduction.

Other estimated variables like unpleasant feeling and dignity still need to be studied more in future.

7. Conclusions

We studied the factors affecting the rate of cycling. In past articles, variables such as age and gender were discussed and few hidden variables were studied. In addition to variables, such as gender, income, and education, we tried to address variables that are less known but have a high impact on the amount of bicycle use, especially in countries such as Iran with specific cultures.

The results showed that having will has a directly positive effect on the cycling rate. This variable has 3 indicators, the most effective of which shows people’s effort to learn (If I want to understand something, I will definitely try to understand it). On the other hand, the cost factor is one of the inhibiting factors that indicate cycling both for people who intend to buy a bicycle and for people who use a shared bike.

Concern for the dignity of those who believe that riding is inappropriate for them given their social level is another aspect. Wearing formal clothing outside the home, which is thought to discourage cycling, is one of its signs. Another factor is the misconception that riding is only for those without a personal car or those with limited financial resources; however, a few people also believe that cycling is just for those with great incomes and fewer obligations.

Limitations from the family are another thing that we examined. One of the important indicators is that the family does not spend money to buy bicycles for their children, which is mostly mentioned by teenagers.

Another indicator is that the family does not allow their children to ride bicycles, some families indirectly (I feel that my family is upset about my cycling/I feel that my family will not allow me to ride a bicycle (and some directly (My family does not pay me to buy a bicycle) prevent their children (of any age) from riding bicycles. This case was raised more among women, both in the teenage years who live with their parents and in the older ages who live with their husbands.

The factor of safety at night also reduces the desire to ride a bicycle, and the fear of theft, the lack of sufficient police in the city, and the possibility of more accidents at night are deterrent indicators of this variable.

We tried to examine different variables, of which the cost variable was one of the most important. If we want to investigate the study from an economic angle, the cost to the household of using bicycles as a method of transportation was looked at, and it revealed that one of the things that makes it more appealing is the decrease in cost. On the other hand, by implementing the following policy proposals, we can expect to increase the attractiveness of cycling. The cost of health issues, infrastructure, repair, and maintenance of non-motorized transportation is far less than other modes of transportation [71]. By reducing travel time and travel distance, it helps reduce household expenses.

Variables, such as safety at night and built/natural environment, by reducing the stress of current cyclists and increasing attractiveness for others, helps individuals to choose this mode of transportation to improve the mental health of the community and experience a higher level of social well-being. The limit by family variable was also investigated, which is one of the problems in the area of this study that even women are facing such limitations.

Based on the issues mentioned above, we tried to take a step towards the sustainability of the city’s transportation as cycling is considered as one of the sustainable modes of transportation [72].

7.1. Policy Implications

Among the most important policies based on the results obtained to increase the attractiveness of cycling, the following can be mentioned:

• Increasing the safety of cycling routes, especially in areas where the demand for cycling by women is higher. It gradually creates a positive mentality among the people of the society, which ultimately increases the attractiveness of this mode of transportation. On the other hand, it is likely that the family’s prohibition on cycling stems from a lack of awareness of societal security, and enhancing security and safety can make the family happy with the cycling of its children or spouse. The main body responsible for developing cycling infrastructure is the deputy transportation department of Tehran Municipality. Its main beneficiary is the only company providing shared bicycles (BDOOD company) that can help Tehran Municipality in investing for developing the infrastructure.
• Allocation of government subsidies to buy or apply discounts to the users of shared bicycles, which will reduce traffic and air pollution. The body responsible for this issue is the transportation deputy of Tehran Municipality.
• Developing behavior and acculturalization of bicycle use in society via advertisements and using it by the senior transportation managers of the countries. All decisions that are made in the field of transportation of Tehran city are made by the deputy transportation department of Tehran Municipality which cooperates with the Deputy of Social and Cultural Affairs of Tehran Municipality in the field of cultural and behavioral development.

7.2. Limitations

We had some limitations during this study as follow:

• The information received from the variables in Table 1 is for comparing with 2018 census data to estimate the representativeness of the data collected by the surveys [73,74]. Except for the gender of the individuals, the correct data of other variables from this census were not available. The age of respondents to the questionnaire was not correctly answered in the online survey, which was deleted due to incomplete information. However, obtained data can be used in other research.
• We were limited of face-to-face questionnaire because of the COVID-19 pandemic and that is why we had online surveys as well.
• There was no accurate data about income, car ownership, and other variables (except gender) in Table 1 about the individuals.

This article does not discuss other significant topics, such as developing the non-motorized transportation infrastructure or other subjective factors, although there may be more latent variables which may be examined in further studies.