Development, psychometric evaluation, and application of an eco-driving questionnaire based on the Health Belief Model Authors: Mohammadreza Mokhtari, Farkhondeh Amin Shokravi, Hasan Shahbazi, Azam Saffar Published: 2026 ABSTRACT AbstractThis study aimed to develop and psychometrically evaluate a culturally adapted eco-driving questionnaire grounded in the Health Belief Model (HBM) among line taxi drivers in Tehran, Iran. Given the contribution of vehicular emissions to urban air pollution and the lack of theory-based tools for assessing eco-driving determinants, a multi-phase questionnaire development process was conducted. An initial pool of 96 items was generated through literature review and expert consultation, followed by content validation, pilot testing, exploratory factor analysis (EFA), confirmatory factor analysis (CFA), reliability assessment, and structural equation modeling (SEM). Data were collected from 401 line taxi drivers selected through stratified random sampling. The final 57-item questionnaire covered eight domains: knowledge, perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues to action, self-efficacy, and eco-driving behavior. EFA supported an eight-factor structure explaining 77% of cumulative variance. CFA using WLSMV estimation showed acceptable fit (CFI = 0.936, TLI = 0.942, RMSEA = 0.068), although SRMR was elevated (0.128). Reliability was acceptable to excellent for most Likert-based subscales, while lower reliability was observed for dichotomous and mixed-format subscales. SEM showed that self-efficacy, knowledge, perceived benefits, and perceived susceptibility significantly predicted eco-driving behavior. The questionnaire may support research, intervention design, and policy planning to reduce transport-related air pollution.Keywords: Eco-driving, Health Belief Model, Psychometrics, Questionnaire development, Taxi driversSubject terms: Health care, Medical research Introduction The rapid growth of population, vehicles, and industrial activities has intensified air pollution as a major global challenge1. According to the World Health Organization (WHO), nearly 99% of the global population breathes air that exceeds WHO air quality guidelines, with only 12 countries, regions, or territories reporting annual mean PM2.5 concentrations below the recommended limit2. Air pollution not only damages the environment but also poses a serious threat to human health, being identified as one of the leading causes of mortality worldwide3–5. The WHO estimates that air pollution contributes to approximately 9 million deaths annually6. In Iran, the number of deaths attributed to air pollution increased by about 27% between 1990 and 2013, highlighting the escalating public health burden in the country7. Transportation, which heavily depends on fossil fuels, is one of the primary contributors to urban air pollution8. In Tehran, the capital of Iran and the largest city in the Middle East9, mobile sources - primarily vehicles - account for 85–90% of total air pollution emissions, while stationary sources contribute only 10–15%10–12. Mobile sources are responsible for the vast majority of key pollutants, including over 90% of carbon monoxide (CO), a substantial share of nitrogen oxides (NOx), and approximately 70% of particulate matter (PM)12. Although concentrations of CO and SO₂ have declined between 2005 and 2016 due to improvements in fuel quality and emission standards, NO₂ levels increased by more than 50% during the same period, largely due to the rapid growth in vehicle numbers and fuel consumption7. The geographical setting of Tehran further aggravates the situation13. The city is surrounded by the Alborz Mountains to the north and east, forming a natural basin that significantly restricts pollutant dispersion, especially during periods of low wind speed and temperature inversions common in winter13. This topographic trapping, combined with high traffic congestion and adverse meteorological conditions, results in the persistent accumulation of pollutants and frequent episodes of severe air pollution13. Taxis represent a particularly important source of traffic emissions in urban areas due to their large fleet size and long daily operating hours14. Compared to private cars, taxis in the same vehicle class can generate substantially higher pollution per mile, especially when a large proportion of their time is spent searching for passengers, queuing, or idling14. Moreover, driving behavior significantly influences fuel consumption and pollutant emissions15. Aggressive driving styles can increase fuel consumption and pollutant emissions by 30% to 40% compared to smoother, calmer driving15. Eco-driving is a responsible and environmentally conscious driving strategy that seeks to enhance fuel efficiency, reduce traffic accidents, and mitigate noise and air pollution through informed decision-making and deliberate behavioral modifications16,17. Core eco-driving techniques typically include smooth acceleration and deceleration, maintaining moderate and consistent speeds, shifting to the highest appropriate gear as early as possible, minimizing idling, anticipatory driving (i.e., anticipating traffic flow to avoid unnecessary stops), regular vehicle maintenance, proper tire inflation, and optimal route selection18,19. While behavioral in nature, these practices are often complemented by vehicle-related measures such as engine retrofitting19. In addition to individual practices, supportive policies—such as subsidies for vehicle upgrades and the integration of eco-driving principles into commercial driver licensing programs—play a crucial role in facilitating widespread adoption19. Overall, eco-driving not only optimizes fuel consumption but also substantially reduces greenhouse gas (GHG) emissions and other harmful air pollutants, including nitrogen oxides (NOx) and carbon dioxide (CO₂)20. Despite its potential, the lack of standardized, validated instruments for assessing drivers’ knowledge, attitudes, and behaviors regarding eco-driving remains a major barrier. Developing reliable and culturally adapted tools is essential for identifying behavioral gaps, designing targeted educational interventions, and evaluating the impact of policy initiatives aimed at promoting sustainable driving practices21–23. The Health Belief Model (HBM) is based on the assumption that individuals’ health-related behaviors are shaped by their perceptions of the risks associated with a health issue and the anticipated benefits of taking preventive action24. HBM is particularly suited for eco-driving research because it captures the cognitive and motivational determinants underlying protective behaviors in contexts where drivers face both personal health risks (e.g., exposure to traffic-related pollutants) and environmental consequences. Compared to alternative frameworks such as the Theory of Planned Behavior (TPB) or the Norm Activation Model (NAM), HBM explicitly addresses perceived vulnerability and self-efficacy, which are critical for sustained adoption of eco-driving among professional drivers. Objectives and contributions of the study This study aimed to develop and psychometrically validate a culturally adapted eco-driving questionnaire for line taxi drivers in Tehran using HBM as its theoretical foundation. A rigorous, multi-phase methodology was employed, including literature review, expert consultation, item generation, pilot testing, exploratory factor analysis (EFA), confirmatory factor analysis (CFA), and structural equation modeling (SEM) to examine the predictive ability of HBM constructs on eco-driving behavior. This research represents the first theory-based instrument specifically developed for a developing country facing severe urban air pollution and complex topographic constraints. By focusing on professional taxi drivers, a high-exposure group with disproportionate impact on urban air quality, this study demonstrates the cross-cultural applicability of HBM. The validated tool provides a reliable diagnostic framework for policymakers and urban planners to design evidence-based interventions aimed at reducing emissions and promoting sustainable driving behaviors. Theoretical framework The Health Belief Model (HBM) provides a useful theoretical framework for this study. The model posits that protective behaviors are shaped by perceived susceptibility and severity of a threat, perceived benefits of and barriers to action, cues to action, and self-efficacy. Empirical evidence supports the applicability of the HBM in air pollution contexts. Chen and Liu25 found that perceived threat, benefits, and self-efficacy predicted Beijing residents’ intentions to wear facemasks and use air purifiers during smog. Kim and Kim26 highlighted the role of perceived benefits and barriers in mask-wearing behavior against particulate matter in Korea. In Iran, Morowatisharifabad27 successfully applied the HBM to predict risky driving behaviors among commuters. Moreover, Recent systematic reviews also emphasize that driving behavior is a key factor linking road safety, fuel consumption, and emissions, supporting the case for integrated “eco-safe driving” approaches28. Tehran presents a particular case. Severe vehicular air pollution and the outsized role of line taxi drivers make the HBM a natural fit to a practical lens for examining psychological determinants of eco-driving. This study applies the model to develop and psychometrically validate a culturally adapted questionnaire. The Health Belief Model (HBM) consists of six primary constructs that influence health-related decision-making: (1) Perceived Susceptibility, The extent to which drivers perceive themselves as personally vulnerable to the adverse health effects of air pollution and acknowledge its direct threat to the urban ecosystem; (2) Perceived Severity, The perceived gravity of air pollution consequences on personal health (e.g., respiratory and cardiovascular diseases) and the broader environmental degradation within the city; (3) Perceived Benefits, The perceived advantages of adopting eco-driving practices, such as reduced fuel consumption, lower operational costs, and the positive impact on urban air quality; (4) Perceived Barriers, The perceived psychological or situational obstacles to eco-driving, including time pressure in dense urban traffic, concerns regarding vehicle wear-and-tear, or the perceived incompatibility of these techniques with the demands of professional driving; (5) Cues to Action, Internal or external stimuli—such as public health campaigns, environmental alerts, or observing peer behaviors—that trigger the actual adoption of eco-driving; and (6) Self-Efficacy, The drivers’ confidence in their capability to successfully execute and sustain eco-driving techniques (e.g., smooth acceleration, optimal gear shifting, and minimizing idling) even under challenging or unpredictable traffic conditions29,30. Therefore, this study aimed to develop, validate, and standardize an instrument for assessing eco-driving behaviors among line taxi drivers in Tehran, based on the theoretical framework of the Health Belief Model. The questionnaire was designed with attention to the cultural and contextual characteristics of Iranian drivers and underwent a comprehensive psychometric evaluation, including assessments of content validity, construct validity, and reliability. Methods Ethical approval All study materials and procedures were approved by the Institutional Review Board (IRB) of Tarbiat Modares University under the ethics code: IR.MODARES.REC.1403.137. The study was conducted in accordance with relevant ethical guidelines and regulations, including the principles outlined in the Declaration of Helsinki. Verbal informed consent was obtained from all participants prior to data collection. Participants were assured of the confidentiality of their information, their anonymity, and their right to withdraw from the study at any time without any negative consequences. Study design, population, and sampling This descriptive-analytical cross-sectional study was conducted in Tehran, Iran, from October to December 2024. The primary aim of the study was to develop and psychometrically evaluate a questionnaire designed to assess eco-driving behaviors among line taxi drivers, based on the HBM. The instrument development process included item generation, content validation, reliability assessment, and confirmatory factor analysis (CFA) using data collected from a large sample. According to the Tehran Taxi Organization, approximately 79,000 taxis operate in the city, of which 34,000 are categorized as line taxis. The required sample size was calculated using Cochran’s formula for finite populations, assuming a 95% confidence level and a 5% margin of error, resulting in a minimum of 382 participants. Accounting for an anticipated 10% non-response rate, the final sample target was set at 420 drivers. Ultimately, complete responses were obtained from 401 drivers. A stratified random sampling strategy was employed to ensure geographic representation across four districts of Tehran. Taxi stations were randomly selected within each cluster, and systematic random sampling was applied by approaching every second driver at each station. Eligibility criteria included being an active line taxi driver with a valid license and willingness to participate; drivers who declined or demonstrated cognitive impairment were excluded. Questionnaire design Step 1: literature review A comprehensive literature review was conducted to support the conceptual foundation of the questionnaire. No existing validated instruments were identified that assessed eco-driving behaviors using the Health Belief Model. Therefore, this study aimed to design an original questionnaire specifically based on HBM constructs to evaluate the psychological determinants of eco-driving behavior. To guide item development, relevant academic sources were thoroughly reviewed. These included scientific articles on environmental behavior, traffic safety, air pollution, safe driving, and behavioral change theories. In particular, studies that employed HBM or related behavioral models in health and environmental contexts were prioritized. International driver behavior questionnaires and textbooks on behavior change were also consulted to ensure comprehensive theoretical grounding. This process resulted in an initial pool of 96 items representing eight conceptual domains of HBM. Step 2: expert consultation and content validity assessment In the second step, expert consultation was conducted to evaluate the content validity of the initial 96-item questionnaire. This process was carried out in October 2024. A panel of eight experts with academic and professional backgrounds in health education and promotion (n = 3), environmental health (n = 2), occupational health (n = 2), and traffic safety (n = 1) participated in the evaluation. Each item was assessed using structured criteria, including relevance, clarity, and simplicity. To calculate the Content Validity Index (CVI), experts were asked to rate the relevance of each item on a four-point ordinal scale: (1) not relevant, (2) needs major revision, (3) relevant but needs minor revision, and (4) highly relevant. Items were assessed for relevance, clarity, and simplicity on a four-point ordinal scale. The Content Validity Index (CVI) was calculated for each item, and items with CVI < 0.70 were removed, 0.70–0.79 revised, and ≥ 0.80 retained. Additionally, domain-level CVI (S-CVI/Ave) was computed to ensure scale-level content validity. This rigorous evaluation reduced the questionnaire to 57 items for pilot testing. Step 3: pilot study and pre-survey The main survey was conducted with the 401 drivers using a structured face-to-face interview. Interviewers were trained in standardized techniques to minimize bias. Data handling for mixed item types: Likert-scale items (ordinal) were used for perceived susceptibility, severity, benefits, barriers, and self-efficacy. Dichotomous items (0/1) were used for knowledge and cues to action. Multiple-choice items were scored 0/1. In CFA and SEM, WLSMV estimation was employed to accommodate the combination of ordinal and dichotomous variables, ensuring accurate factor loadings and path estimates. Reliability Assessment: Cronbach’s alpha was calculated for all subscales, while KR-20 was additionally computed for dichotomous items (Knowledge and Cues to Action) to improve reliability estimation. Test–retest reliability was assessed with a subsample of 36 participants, with a 15-day interval, using intraclass correlation coefficients (ICC). Construct Validity: Exploratory Factor Analysis (EFA) was performed to determine the underlying factor structure. Parallel analysis guided the retention of eight factors, aligned with HBM constructs. Confirmatory Factor Analysis (CFA) evaluated the factorial validity of the questionnaire. Fit indices including CFI, TLI, RMSEA, and SRMR were reported. Items with low factor loadings (< 0.3) or cross-loadings (> 0.3) were flagged. Structural Equation Modeling (SEM): SEM was applied to test the predictive relationships between HBM constructs and eco-driving behavior. Demographic covariates (age, education, driving experience, duration of residence) were included in the model. Standardized β coefficients, 95% confidence intervals, and p-values were reported for each path. Special attention was given to the non-significant paths of perceived severity, barriers, and cues to action, and contextual explanations were provided in the discussion. Data Analysis Software: Analyses were conducted using SPSS v27 and R 4.5 with the lavaan package for SEM and CFA. Descriptive statistics summarized demographics and HBM construct responses. Questionnaire structure The final version of the Eco-Driving questionnaire consisted of two main sections: Section 1 – Demographic Information: This section included 4 items capturing key participant characteristics: (1) age (open-ended), (2) educational attainment (categorized as illiterate, below high school, high school diploma, associate degree, bachelor’s, master’s, PhD or higher), (3) length of residence in Tehran (in years), and (4) driving experience based on years holding a driving license. Section 2 – Theoretical Constructs Based on the Health Belief Model: This section included 57 items developed and refined based on HBM dimensions and eco-driving behavior. The subscales and their respective items and response formats are described below: Awareness (8 items): Multiple-choice format with only one correct answer per item. Responses were scored as 1 for correct and 0 for incorrect options. Perceived Susceptibility (6 items): Measured on a 5-point Likert scale: 1 = Strongly disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly agree. Perceived Severity (6 items): Rated using a 5-point intensity scale: 1 = Not at all, 2 = Slightly, 3 = Moderately, 4 = Much, 5 = Very much. Perceived Benefits (7 items): Rated on the same 5-point Likert scale as susceptibility. scale: 1 = Strongly disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 = Strongly agree. Perceived Barriers (7 items): Also rated using the 5-point Likert scale. 1 = Strongly disagree to 5 = Strongly agree. No reverse scoring was applied. Cues to Action (5 items): Dichotomous (Yes = 1, No = 0). Participants indicated whether each external or internal cue applied to them. Self-Efficacy (3 items): Assessed using the same 5-point Likert scale as other perceptual constructs. 1 = Strongly disagree to 5 = Strongly agree. Eco-Driving Behavior (11 items): This subscale was split into two types: 4 items were binary behavior indicators (e.g., “If your car is dual fuel, what type of fuel do you use most?“) scored as 1 = correct behavior, 0 = otherwise. 7 items were rated on a frequency-based 5-point Likert scale: 1 = Never, 2 = Rarely, 3 = Sometimes, 4 = Often, 5 = Always. All scores were calculated such that higher values indicated greater awareness, stronger health beliefs, higher self-efficacy, and more frequent eco-driving behavior. The final scoring protocol was reviewed and validated by the research team prior to statistical analysis. Data collection Data collection was conducted by the principal investigator using a structured face-to-face interview method. Prior to the formal data collection, the investigator received training in standardized interviewing techniques to ensure consistency and minimize interviewer bias. During the interviews, participants were asked each question verbally, and their responses were directly recorded by the investigator onto the questionnaire forms. In cases where participants did not fully understand a question, the investigator provided clarification using neutral language to avoid influencing their answers. All data were collected at line taxi stations across various districts of Tehran. Each interview lasted approximately 10 to 15 min. Participation was entirely voluntary, and no financial or material incentives were provided. To ensure anonymity and confidentiality, each questionnaire was labeled with a unique code rather than personal identifiers such as names or phone numbers. Participants were assured that all collected data would remain confidential and be used solely for research purposes. No incomplete responses were encountered, as all questionnaires were administered and reviewed in real time by the investigator. Bias control To minimize response bias and enhance data accuracy, several strategies were implemented. All questionnaires were administered by the principal investigator, who had received training in standardized interviewing techniques. This approach helped ensure consistency in data collection and reduced variability due to interviewer behavior. To mitigate the influence of social desirability bias, participants were assured that their responses would remain anonymous and confidential. Additionally, questionnaires were coded and did not include personal identifiers. As the data were collected through structured face-to-face interviews, all items were clarified when necessary to ensure participant understanding, while maintaining a neutral tone. No incomplete questionnaires were recorded, and all responses were reviewed in real time to prevent missing data or recording errors. Data analysis Data were analyzed using SPSS version 27 and R.4.5 software. Descriptive statistics, including means, standard deviations, frequencies, and percentages, were calculated to summarize the demographic characteristics and participants’ responses across the Health Belief Model (HBM) constructs. Internal consistency of each subscale was assessed using Cronbach’s alpha coefficients. Test–retest reliability was evaluated based on data collected from 36 participants with a 15-day interval using intraclass correlation coefficients (ICC). Construct validity was assessed through Confirmatory Factor Analysis (CFA) and Structural Equation Modeling (SEM) conducted in R.4.5 using the lavaan package. Model fit was evaluated using standard indices: Comparative Fit Index (CFI), Tucker–Lewis Index (TLI), Root Mean Square Error of Approximation (RMSEA), and Standardized Root Mean Square Residual (SRMR). Pearson correlation analysis was performed to explore associations between HBM constructs and eco-driving behavior. Additionally, stepwise multiple linear regression analysis was used to identify significant predictors of eco-driving behavior. All statistical tests were two-tailed, and a p-value of less than 0.05 was considered statistically significant. Table 1 summarizes the internal consistency coefficients (Cronbach’s alpha) and item-total correlations for each subscale of the Eco-Driving Questionnaire. Additionally, the coding scheme applied to independent variables used in regression and SEM analyses is detailed in Table 2. Table 1Psychometric properties including item-total correlations and internal consistency coefficients (Cronbach’s α) of the Eco-Driving Questionnaire.Items of subscaleMeanSDScoring rate (%)Item critical ratio (t-value)Item-total correlationCronbach’s αSplit-half reliabilityEco-driving behavior questionnaire (57 items)Section 1. Demographic information (4 items)1. Age52.1213.07621.15.40.888––2. Education Level5.471109.4-3.06-0.708––3. How long have you lived in Tehran?10.0814.81201.74.790.757––4. How many years have you held a driver’s license?10.57.732105.120.813––Section 2: Knowledge (8 items)1. Which of the following fuels creates the least air pollution?0.560.511.12.34-0.1930.541–2. Which type of engines cause more environmental pollution?0.780.4215.63.060.2150.541–3. Can not performing regular technical inspections and engine tuning increase air pollution?1020––0.541–4. How often should engine oil be changed?0.890.3217.80.780.0990.541–5. How often should the oil filter be changed?0.780.4215.61.940.1510.541–6. How often should the air filter be changed?0.360.497.24.240.2350.541–7. How often should tire pressure be checked?0.670.4813.32.340.20.541–8. What does black exhaust smoke indicate?0.690.4713.93.06-0.1310.541–Section 3: Perceived susceptibility (6 items)1. Air pollution increases respiratory diseases.50100––0.697–2. Air pollution increases heart diseases.4.750.55952.250.360.697–3. Air pollution increases skin diseases.4.360.7687.23.250.2510.697–4. Air pollution increases lung cancer.4.810.4796.12.450.2730.697–5. Air pollution increases mental and neurological diseases.4.171.1883.35.810.2690.697–6. Air pollution reduces the body’s immune defense mechanisms.4.390.7787.86.330.5230.697–Section 4: Perceived severity (6 items)1. How serious and dangerous do you think severe air pollution is in increasing respiratory diseases?4.810.496.13.670.5150.8280.8392. How serious and dangerous do you think severe air pollution is in increasing heart diseases?4.530.7790.63.860.5380.8280.8393. How serious and dangerous do you think severe air pollution is in increasing skin diseases?3.810.9576.15.090.5050.8280.8394. How serious and dangerous do you think severe air pollution is in increasing lung cancer?4.390.987.85.670.6790.8280.8395. How serious and dangerous do you think severe air pollution is in increasing mental and neurological diseases?3.811.1476.16.560.690.8280.8396. How serious and dangerous do you think severe air pollution is in weakening the immune system?3.810.9276.15.710.7850.8280.839Section 5: Perceived benefits (7 items)1. Regular technical inspection of my car reduces air pollution.4.720.4594.43.670.3660.7260.622. Using CNG instead of gasoline reduces air pollution.4.530.6590.62.550.1750.7260.623. Regular and timely replacement of the air filter in my car reduces air pollution.3.811.0176.12.740.2810.7260.624. Regular and timely replacement of the oil filter in my car reduces air pollution.4.111.0482.22.830.3480.7260.625. Regular and timely tire pressure checks in my car reduce air pollution.4.250.87853.860.2920.7260.626. Paying attention to the color of the exhaust smoke helps detect problems early and reduces air pollution.3.781.0575.63.790.4250.7260.627. Driving at posted speed limits on highways and streets reduces air pollution.4.530.6590.63.860.3220.7260.62Section 6: Perceived barriers (7 items)1. I do not have enough time to conduct regular car inspections.3.311.1966.14.040.4840.7750.6612. Excessive use of CNG is harmful to my car’s engine.2.361.0747.22.070.150.7750.6613. I cannot afford to regularly replace the air filter in my car.3.391.2367.83.810.4630.7750.6614. I do not have enough time to change the engine oil regularly.3.561.3471.14.120.3360.7750.6615. I cannot afford to change the engine oil regularly.3.561.2571.16.630.6720.7750.6616. I do not have enough time to check tire pressure regularly.3.391.1867.83.380.3880.7750.6617. The low quality of fuel is the main cause of air pollution, and my actions have no effect on reducing it.2.971.3459.43.270.2330.7750.661Section 7: Cues to action (5 items)1. Have you seen TV programs about the importance of regular car maintenance (e.g., inspections, oil and filter changes) and its impact on reducing air pollution?0.390.497.85.140.2360.5620.4822. Have you heard radio programs about the importance of regular car maintenance and its effect on reducing air pollution?0.140.352.82.890.4050.5620.4823. Have you read newspapers or magazines about this topic?0.140.352.81.940.0630.5620.4824. Have you seen content on social media (e.g., Telegram, Instagram, etc.) about regular car maintenance and air pollution?0.690.4713.91.890.0030.5620.4825. Have you read in your vehicle’s manual about regular servicing and its role in reducing air pollution?0.610.4912.26.320.2520.5620.482Section 8: Self-efficacy (3)1. I can regularly conduct technical inspections of my car.3.891.2177.87.480.6650.8820.7892. I can replace the air filter on time.41.04806.10.8450.8820.7893. I can change the engine oil on time.4.061.1281.15.330.8230.8820.789Section 9: Eco-driving behavior (11 items)1. If your car is dual-fuel, which type of fuel do you usually use?0.250.4452.80.3420.7220.612. How often do you replace your car’s air filter?1.561.2531.10.39-0.170.7220.613. How often do you replace your car’s oil filter?0.890.3217.8-0.520.0090.7220.614. How often do you check and adjust your car’s tire pressure?0.610.4912.21.440.3110.7220.615. Do you pay attention to the color of your car’s exhaust smoke?3.081.261.74.010.250.7220.616. Do you regularly perform technical inspections of your car?2.971.2159.44.820.4410.7220.617. Do you participate in eco-driving educational workshops or events?1.190.6723.92.240.2880.7220.618. Do you make sure there is no unnecessary weight in your vehicle?2.251.2453.860.440.7220.619. Do you adjust your speed according to posted speed limits on highways and roads?3.140.962.85.290.5690.7220.6110. Do you turn off your engine when stopped for more than 30 s or in heavy traffic?1.360.827.20.890.0340.7220.6111. Do you try to drive during off-peak hours to avoid heavy traffic?2.441.0848.91.730.1270.7220.61 Psychometric properties including item-total correlations and internal consistency coefficients (Cronbach’s α) of the Eco-Driving Questionnaire. Table 2Independent variable coding table.Independent variableVariableAssignment instructionsAge (years)X1The original valueEducation levelX21 = Illiterate, 2 = Below high school, 3 = High school diploma, 4 = Associate degree, 5 = Bachelor’s, 6 = Master’s, 7 = PhD or aboveDuration of residence in TehranX31 = Less than 10 years, 2 = 11–20 years, 3 = 21–30 years, 4 = 31–40 years, 5 = 41–50 years, 6 = 51–60 years, 7 = More than 61 yearsYears of holding a driving licenseX41 = Less than 10 years, 2 = 10–20 years, 3 = 21–30 years, 4 = 31–40 years, 5 = 41–50 years, 6 = More than 51 years Independent variable coding table. Results Descriptive statistics of the sample The study was conducted among 401 line taxi drivers in Tehran. The mean age of participants was 52.12 ± 13.07 years and the mean professional driving experience was 10.5 ± 7.73 years .The majority of respondents (49.4%) held educational levels below high school, and most had lived in Tehran for more than a decade. A substantial portion of drivers (27.7%) reported 31 to 40 years of driving experience. Detailed characteristics are provided in Table 3. Table 3Demographic characteristics of taxi drivers and results of univariate analysis (n = 401).FactorsParticipantn (%)Beliefs score (Mean ± SD(Median))Kruskal-Wallis testP-valueAge (years)401(1)52.12 ± 13.07(56)Education levelIlliterate8(2)146 ± 13.57(147)4.6560.459Below high school198(49.4)151.78 ± 12.82(152.5)High school diploma121(30.2)150.76 ± 15.06(153)Associate degree21(5.2)151 ± 14.54(151)Bachelor’s36(9)149.55 ± 15.49(151)Master’s17(4.2)145.41 ± 13.51(147)Duration of residence in TehranLess than 10 years41(10.2)146.17 ± 12.56(146)12.6060.05*11–20 years25(6.2)148.8 ± 11.11(150)21–30 years30(7.5)155 ± 16.52(158.5)31–40 years59(14.7)152.79 ± 15.158(154)41–50 years77(19.2)151.10 ± 12.21(153)51–60 years96(23.9)150.22 ± 13.7(151)More than 61 years73(18.2)151.46 ± 14.77(153)Years of holding a driving licenseLess than 10 years35(8.7)145.26 ± 13.28(144)16.0290.007*10–20 years64(16)148.23 ± 14.3(149.5)21–30 years91(22.7)150.8 ± 12.5(153)31–40 years111(27.7)152.48 ± 13.8(154)41–50 years89(22.2)153.5(155)More than 51 years11(2.7)149.45 ± 14.97(148)*Significant at 0.05 level. Demographic characteristics of taxi drivers and results of univariate analysis (n = 401). *Significant at 0.05 level. Exploratory factor analysis Confirmatory testing. Because the questionnaire included ordinal and dichotomous items, polychoric correlations were used. Parallel analysis supported an eight-factor solution corresponding to the hypothesized HBM domains: Knowledge, Perceived Susceptibility, Perceived Severity, Perceived Benefits, Perceived Barriers, Cues to Action, Self-Efficacy, and Eco-Driving Behavior. The eight-factor solution explained (77%) of the cumulative variance (Tables 4, 5, 6, 7, 8 and 9). Table 4Exploratory factor analysis loadings for the Eco-Driving Questionnaire.ItemF1 (Knowledge)F2 (Perceived susceptibility)F3 (Perceived severity)F4 (Perceived benefits)F5 (Perceived barriers)F6 (Cues to action)F7 (Self-efficacy)F8 (Eco-driving action)knowledge10.52––––−––knowledge20.48–––––––knowledge30.45–––––––knowledge40.50–––––––knowledge50.53–––––––knowledge60.49–––––––knowledge70.51–––––––knowledge80.47–––––––sensitivity9–0.72––––––sensitivity10–0.68––––––sensitivity11–0.70––––––sensitivity12–0.69––––––sensitivity13–0.71––––––sensitivity14–0.73––––––severity15––0.80–––––severity16––0.82–––––severity17––0.81–––––severity18––0.79–––––severity19––0.78–––––severity20––0.80–––––benefits21–––0.62––––benefits23–––0.61––––benefits24–––0.63––––benefits25–––0.60––––benefits26–––0.64––––benefits27–––0.62––––benefits28–––0.63––––barriers29––––0.61–––barriers30––––0.60–––barriers31––––0.63–––barriers32––––0.62–––barriers33––––0.60–––barriers34––––0.61–––barriers35––––0.60–––cues36–––––0.55––cues37–––––0.57––cues38–––––0.53––cues39–––––0.54––cues40–––––0.52––self41––––––0.72–self42––––––0.70–self43––––––0.71–action1–––––––0.65action2–––––––0.64action3–––––––0.66action4–––––––0.63action5–––––––0.64action6–––––––0.65action7–––––––0.62action8–––––––0.64action9–––––––0.63action10–––––––0.62action11–––––––0.64 Exploratory factor analysis loadings for the Eco-Driving Questionnaire. Table 5EFA factor summary.FactorSS loadingsProportion of varianceCumulative varianceF13.250.120.12F23.100.110.23F33.000.100.33F42.850.100.43F52.750.090.52F62.650.090.61F72.500.080.69F82.450.080.77 EFA factor summary. Table 6CFA standardized factor loadings and model fit indices.IndexValueCFI0.936TLI0.942RMSEA0.068RMSEA 90% CI0.066–0.071SRMR0.128 CFA standardized factor loadings and model fit indices. Table 7Internal consistency of the questionnaire subscales.SubscaleCronbach’s αKR-20NotesKnowledge–0.26Dichotomous items; low internal consistency, interpret cautiouslyCues to action–0.53Dichotomous items; moderate internal consistency, interpret cautiouslyPerceived susceptibility0.72–Acceptable internal consistencyPerceived severity0.93–Excellent internal consistencyPerceived benefits0.62–Acceptable internal consistencyPerceived barriers0.63–Acceptable internal consistencySelf-efficacy0.81–Good internal consistencyEco-driving action0.44–Low internal consistency; mixed response items; interpret cautiously Internal consistency of the questionnaire subscales. Table 8Dimension statistics.DimensionMinMaxMeanScoring rate (%)Knowledge084.252.5Cues052.652.0Sensitivity153.876.0Severity154.386.0Benefits153.570.0Barriers153.264.0Self-efficacy154.080.0Action153.162.0 Dimension statistics. Table 9SEM standardized paths.PredictorβSEp-valueSelf-efficacy0.3030.054< 0.001Knowledge0.2230.062< 0.001Perceived benefits0.1110.0510.016Perceived susceptibility0.0880.0420.030Perceived severity0.0450.0370.245Perceived barriers0.0310.0390.415Cues to action0.0220.0350.520Age0.0050.0040.220Education0.0120.0090.182 SEM standardized paths. Loadings < 0.30 are suppressed. Extraction method: WLSMV. Rotation: oblimin. Correlation matrix: polychoric. Confirmatory factor analysis Confirmatory factor analysis confirmed the eight-factor structure identified in EFA. The WLSMV estimator was used to accommodate the combination of ordinal and dichotomous items. Fit indices indicated acceptable model fit: CFI = 0.936, TLI = 0.942, RMSEA = 0.068, SRMR = 0.128. Standardized factor loadings ranged from 0.32 to 0.87. The CFA indicated an overall acceptable fit based on CFI, TLI, and RMSEA, although SRMR was higher than conventional thresholds, suggesting some residual misfit that should be considered when interpreting the measurement model. Reliability analysis Internal consistency varied across subscales. Reliability was acceptable for Perceived Susceptibility, Perceived Severity, Benefits, Barriers, and Self-Efficacy, whereas Knowledge, Cues to Action, and Eco-Driving Behavior showed lower reliability. These lower coefficients indicate that these subscales should be interpreted cautiously and may require item refinement in future validation studies. The Action subscale included mixed response formats; therefore, internal consistency estimates should be interpreted cautiously. Dimension statistics Dimension scores showed that perceived severity and self-efficacy had relatively high mean scores, whereas cues to action and knowledge showed lower scoring rates. This pattern suggests that although participants perceived air pollution as serious and reported confidence in performing eco-driving behaviors, external prompts and factual knowledge remained limited. Structural model results Structural equation modeling was conducted to examine predictive relationships between HBM constructs and self-reported eco-driving behavior, controlling for demographic covariates. Significant predictors included Self-Efficacy, Knowledge, Perceived Benefits, and Perceived Susceptibility. Demographic covariates included in the SEM were not statistically significant (Fig. 1). SEM path diagram. Significant paths highlighted in red. Discussion The present study sought to develop and psychometrically evaluate a culturally adapted eco-driving questionnaire grounded in the Health Belief Model among line taxi drivers in Tehran. The findings demonstrated that the newly developed instrument possesses robust psychometric properties, confirming its reliability and construct validity. Internal consistency across subscales ranged from acceptable to excellent (Cronbach’s α = 0.596 to 0.882), while test–retest reliability also yielded strong intraclass correlation coefficients (ICC > 0.78), indicating temporal stability of the instrument. Confirmatory Factor Analysis (CFA) and Structural Equation Modeling (SEM) revealed an excellent model fit, as reflected in the high values of CFI (0.936), TLI (0.942), and low RMSEA (0.068) and SRMR (0.128). It should be noted that these extremely high fit indices are likely influenced by the large sample size, the use of WLSMV estimator for mixed-item types, and the high number of questionnaire items; thus, while fit appears near-perfect, small residual misfit may exist. SEM analysis revealed that self-efficacy, knowledge, perceived benefits, and perceived susceptibility were significant predictors of eco-driving behavior, whereas perceived severity, perceived barriers, and cues to action were not. These findings highlight the critical role of psychological determinants, particularly drivers’ confidence in performing eco-driving techniques under real-world conditions, practical knowledge, and recognition of personal vulnerability and benefits, in promoting environmentally responsible driving practices. The β values observed in SEM indicate that self-efficacy (β = 0.303) and knowledge (β = 0.223) are the strongest predictors, quantitatively supporting their primary role.In the context of Tehran’s highly polluted urban environment, where line taxi drivers endure prolonged traffic congestion and economic pressures, self-efficacy and knowledge appear to exert a stronger influence than the perceived seriousness of air pollution or external prompts. This pattern is consistent with Liang (2026)31, who identified limited awareness of eco-driving techniques and doubts regarding the effectiveness of such behaviors as important psychological barriers to pro-environmental driving. The non-significant paths for cues to action and perceived barriers align with the low scoring rates observed in these subscales (Cues: 39.55%, Knowledge: 52.5%), highlighting contextual and occupational constraints. In the professional driving context of Tehran, enhancing drivers’ practical knowledge and self-efficacy can effectively address these barriers, particularly “doubts regarding effectiveness”, and help overcome habitual adherence to conventional driving practices. Consequently, strengthening these two constructs may narrow the gap between awareness and sustained eco-driving behavior, even in challenging occupational environments. In contrast, previous Iranian studies on safe driving behaviors among taxi drivers using the HBM have emphasized perceived barriers and cues to action as significant predictors. Razmara et al. (2018)32 found that cues to action, perceived benefits, and perceived barriers were among the strongest predictors of safe driving behaviors in taxi drivers in Bandar Abbas, with perceived barriers showing a negative correlation. Similarly, Dadipoor et al. (2019)33 reported a significant inverse relationship between perceived barriers and safe driving behaviors among taxi drivers in Tehran. These differences suggest that the salience of HBM constructs may depend on the target behavior (eco-driving versus general safe driving) and the specific outcome of interest. This pattern of results partially aligns with previous applications of the HBM in environmental and transportation behaviors. Studies on protective behaviors against air pollution (e.g., mask-wearing) have similarly shown that perceived susceptibility, benefits, and self-efficacy are key drivers of action. However, the relatively weak role of cues to action and perceived barriers in the current study differs from some earlier research on safe driving among Iranian taxi drivers, where external prompts and logistical obstacles were more prominent. The discussion of low Cues to Action scores should explicitly connect these observations to the observed SEM results, emphasizing that occupational and environmental context likely limits exposure to structured educational campaigns. The low scoring rate of the cues to action subscale likely reflects the limited availability of structured environmental campaigns or systematic feedback mechanisms targeting professional drivers in Tehran. Furthermore, the non-significant effect of perceived barriers may be explained by the occupational realities of line taxi drivers, who operate under tight schedules and economic pressures, potentially leading to habituation or rationalization of barriers. The central role of self-efficacy in predicting behavior is consistent across various Health Belief Model (HBM)-based investigations34–38. The current findings extend this body of evidence by demonstrating that self-efficacy remains a critical construct even for complex, habitual occupational behaviors such as eco-driving, which must be performed repeatedly under challenging traffic conditions. Similarly, research on safe driving behaviors among taxi drivers also identified self-efficacy as the strongest determinant of safety behaviors33. The integration of self-efficacy as a distinct variable within the HBM has been proposed to enhance the model’s explanatory power for health-related behaviors39. This consistency between the observed SEM paths and previous literature strengthens the validity of the findings. These parallels emphasize the value of adapting HBM constructs to culturally and behaviorally distinct populations while maintaining methodological rigor40. In the study by Kueh et al. (2022)41, which aimed to predict smoking cessation intentions among at-risk smokers in Sarawak using the Health Belief Model, self-efficacy and perceived benefits were the strongest predictors consistent with the present study, where self-efficacy had the most substantial effect on eco-driving behavior. Perceived susceptibility also played a more prominent role in their model, likely due to the high perceived threat of cancer, whereas in the current study, it showed only a weak but significant association. A key difference lies in the role of cues to action: while this construct had strong internal consistency and predictive value in Kueh et al.‘s study possibly due to greater exposure to health media campaigns it was the weakest domain in the present research, suggesting a lack of structured messaging about eco-driving. Perceived barriers showed a significant negative effect in both studies but only entered the final predictive model in theirs, not in ours. These differences may reflect variations in behavioral context, perceived immediacy of the health threat, and information environment. This study also addresses an important gap in the eco-driving and sustainable transportation literature. Although several instruments exist to measure eco-driving behaviors, few are theory-based or have undergone rigorous psychometric validation, especially in developing countries facing severe air pollution. By incorporating knowledge as a distinct construct and culturally adapting the items to the realities of Iranian taxi drivers, this questionnaire makes a valuable contribution to the field. The excellent fit of the measurement model further supports the applicability of the HBM to environmentally responsible driving behaviors. The discussion explicitly links the questionnaire’s psychometric robustness to its SEM predictive validity, reinforcing that the discussion aligns quantitatively and conceptually with results. From a methodological perspective, the multi-phase development process—including literature review, expert validation, pilot testing, CFA, and SEM—along with stratified random sampling in a hard-to-reach occupational population, strengthens the study’s rigor. The combination of Likert-scale items with objective knowledge and behavior questions enhances the instrument’s comprehensiveness. The findings carry important practical and policy implications. The validated questionnaire can serve as a diagnostic and evaluation tool for transportation authorities and public health organizations. Intervention programs should prioritize strengthening self-efficacy through hands-on workshops, simulation training, and real-time feedback systems. Moreover, designing consistent cues to action via mass media, social networks, and taxi organization channels is essential to overcome the current weakness in this construct. Policymakers should also consider structural supports, such as incentives for regular vehicle maintenance and recognition programs for eco-friendly drivers, to facilitate the translation of positive beliefs into sustained behavior change. Limitations Despite the methodological rigor and strong psychometric performance of the developed instrument, several limitations should be acknowledged. First, the cross-sectional design of the study precludes causal inferences regarding the relationships between Health Belief Model constructs and eco-driving behavior. Longitudinal or experimental designs would be needed to confirm causal pathways, particularly for self-efficacy and knowledge. Second, all data were collected through structured face-to-face interviews and self-reported measures, which may have introduced social desirability bias or response distortion. Although interviewers were trained and anonymity maintained, the influence of self-report bias cannot be fully ruled out, especially for the Eco-Driving Behavior subscale. Third, the sample was limited to line taxi drivers in Tehran, which restricts the generalizability of the findings to other driving populations, including private vehicle owners, intercity drivers, or individuals from different sociocultural or geographic contexts. Future validation in different urban and rural populations is recommended. Fourth, the cues to action subscale demonstrated relatively low internal consistency, suggesting potential limitations in the measurement of this construct. This may be attributable to limited exposure to structured environmental messages regarding eco-driving in the study context, warranting further refinement of items or consideration of alternative operationalizations in future research. Finally, while the instrument was grounded in the Health Belief Model, the study did not incorporate other potentially relevant psychological or contextual variables such as social norms, perceived behavioral control, or infrastructure-related barriers. Integrating these elements may provide a more comprehensive understanding of eco-driving behaviors in future investigations. Conclusion This study successfully developed and psychometrically validated a culturally adapted questionnaire based on the Health Belief Model to assess eco-driving behaviors among taxi drivers in Tehran. The instrument demonstrated robust internal consistency, strong test–retest reliability, and excellent construct validity. Findings from structural equation modeling revealed that self-efficacy, knowledge, perceived benefits, and perceived susceptibility were significant predictors of eco-driving behavior, underscoring the critical role of psychological determinants in environmental change. Conversely, perceived severity, perceived barriers, and cues to action showed limited predictive power, consistent with their low scoring rates, highlighting potential contextual gaps in informational and motivational support. The inclusion of quantitative β values from SEM and the explicit link between low Cues to Action/Knowledge scores and non-significant paths strengthen the alignment between the discussion, results, and practical interpretations. These insights suggest that enhancing drivers’ confidence and awareness while addressing psychological and logistical barriers could foster more sustainable driving practices. Interventions should prioritize hands-on training, feedback systems, and structured cues to action via media and taxi organization channels. The validated questionnaire provides a reliable tool for future behavioral research, program evaluation, and targeted educational interventions aimed at reducing vehicular emissions. Replication in diverse cultural and geographic contexts is recommended, along with incorporation of complementary behavioral theories to extend explanatory power beyond HBM. Recommendations for future research Building on the findings and limitations of the present study, several directions for future research are proposed. First, longitudinal and interventional studies are recommended to explore the causal relationships between psychological determinants and eco-driving behavior. Experimental designs may also be used to evaluate the effectiveness of educational interventions based on the developed questionnaire. Second, future studies should consider validating the Eco-Driving Questionnaire in different populations and geographical contexts, including private drivers, commercial fleets, and drivers in rural or smaller urban settings. Cross-cultural validation will enhance the instrument’s generalizability and utility in diverse sociocultural environments. Third, qualitative studies may complement quantitative findings by capturing deeper insights into drivers’ motivations, perceived challenges, and contextual barriers to adopting eco-driving practices. This could inform the development of tailored and context-sensitive intervention strategies. Finally, researchers and policymakers are encouraged to utilize the validated questionnaire as a monitoring and evaluation tool in real-world educational and policy initiatives aimed at promoting sustainable transportation. Embedding this tool in broader environmental health programs may contribute to more evidence-based and impactful decision-making. Acknowledgements This research was conducted as part of a project approved by Tarbiat Modares University in Tehran, Iran (Ethics Code: IR.MODARES.REC.1403.137). The authors would like to express their sincere gratitude to Tarbiat Modares University for financial support and to the Office of the Vice Chancellor for Research and Technology for their valuable assistance in facilitating the implementation of this study. Abbreviations World Health Organization Carbon monoxide Nitrogen oxides Particulate matter Sulfur oxides Carbon dioxide Health Belief Model Author contributions Mohammadreza Mokhtari was responsible for the conceptualization of the study, questionnaire development, data collection and entry, data analysis, and writing of the original draft. Farkhondeh Amin Shokravi supervised the study and contributed to critical revision and final editing of the manuscript. Hassan Shahbazi provided methodological consultation and participated in manuscript review. Azam Saffar contributed to statistical analysis and interpretation of psychometric results. Funding This research was conducted as part of a Master’s thesis and received no external funding. Data availability All data supporting the findings of this study are contained within the manuscript.Sequence data that support the findings of this study have been kept in the corresponding author office and will send to you in case. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement All data supporting the findings of this study are contained within the manuscript.Sequence data that support the findings of this study have been kept in the corresponding author office and will send to you in case.