Study design and participants
A cross-sectional study was conducted to investigate factors affecting pedestrians’ road crossing beliefs and behaviors in potentially risky situations based on the Theory of Planned Behavior.
Since, the young adults (18-25 years old) are the potential victims in pedestrian road traffic accidents in Iran [22] and because of higher rate of mortality due to road traffic accidents among young adults [10], a total of 562 young adults (285 male, 277 female), aged 18-25 years, living in Tehran, Iran were recruited.
A stratified multistage sampling method was applied to recruit participants. After stratifying based on district and size of residence the blocks (units) were randomly selected and every 10th house in a block was selected. Finally, every 18-25 year- old residence from all the 22 districts in Tehran was considered as part of the population of the study. Every household within 22 different districts in Tehran had the same probability of being sampled.
Next, participants responded to the structured written questionnaire of pedestrian road crossing behavior (PEROB) via self-report manner. More detail about the instrument is reported elsewhere [23].
Informed consent was obtained from all study participants before the project began and they had been taught how to complete questionnaires. All of the process of study including data collecting, data analyzing and providing the final report lasted from 27 January 2018 to 20 May 2018.
Measures
TPB Theoretical Variables
The items which assessed components of the TPB were derived from the developed scale for pedestrian road crossing behavior. There were 18 items for measuring following constructs: (1) subjective norms, (2) attitude, (3) perceived behavioral control, and (4) behavioral intention which all the psychometrics characteristics of this instrument had been reported previously [23].
Subjective Norms: Four items measured subjective norms toward the risky road crossing behaviors based on two provided scenarios which assessed potentially risky road crossing behaviors. Example of scenarios for measuring subjective norms was "You’ve left your wallet in a bank. When you decide to come back and get it, find yourself on the opposite side of the street that you need to go back and the environmental condition is as follows: 1-The street is a busy two-way arterial and 2-Right side, about 50 meters away from where you are standing, there is a pedestrian bridge, but you are tired and anxious, and you have to get there before closure of the bank. Will you crossing the street without using the pedestrian bridge?” All items were rated on a 5-point Likert scale from 1 to 5 (strongly agree to strongly disagree). The reliability coefficients of (α = 0.79) indicated internal consistency of the measure.
Perceived behavioral control: Four other items measured the perceived behavioral control beliefs about pedestrians' road crossing behaviors (e.g., If I'm in a hurry, I'll wait in the pedestrian area for the traffic light to turn green). The reliability coefficients of (α = 0.86) verified internal consistency of the measure.
Intention: To measure behavioral intention two items (e.g., I intend to keep my road crossing behavior safe in the next 6 months) were used. All items were rated on a 5-point Likert scale from 1 to 5 (strongly agree to strongly disagree). The reliability coefficients of (α = 0.79) indicated internal consistency of the measure.
Attitude towards traffic rules and regulations: Attitude was measured using 4 items that negatively worded like ‘traffic rules have been adopted just for drivers’ and ‘walk signals are mainly designated for elderly people and children’ (scored 1= strongly agree to 5= strongly disagree). For the current study, the estimates of internal consistency for attitude, as measured by Cronbach's Alpha, were (α=0.88).
Perceived risk: Perceived risk was measured by presenting a list of risky situations in road crossing (9 items) and asking the respondents to indicate to what extent will involve in an accident when crossing in these situations. These items were categorized on a 5point Likert scale (very small extent scores 5 and very large extent scores 1).
Examples of items were ‘texting messages on the phone while crossing the street’ and ‘crossing the street without using the pedestrian bridge’. An estimated reliability coefficient (α = 0. 83) indicated that the measure of perceived risk was internally consistent.
Perceived severity: Perceived severity of involvement in an accident was measured by 3 items including ‘an accident that may happen to me could be costly’, ‘an accident may cause me a permanent disability’, and ‘an accident that may happen to me be able to also troublesome to the driver’. Each item was rated on a 5-point scale (strongly agree scores 5 to strongly disagree scores 1) (α=0.678). The reliability coefficient for the scale was 0.68. Higher scores on the scale indicated higher perceived severity.
Self-reported unsafe road-crossing behaviors: Pedestrian behaviors in potentially risky situations were assessed by 7 items such as: ‘I use the pedestrian “crosswalk” when crossing the street’, and ‘I used to wait until the pedestrian “crosswalk” turns green to cross’. Each item was rated on a five-point Likert scale (never to always). (Positively worded items: never scores 1 and always scores 5; negatively worded items: never scores 5 and always scores 1).
An estimated reliability coefficient (α = 0. 80) indicated that the measure of risky road-crossing behaviors was internally consistent. The total score calculated and transformed into a 0 to 100 result where a higher score represented better condition.
Socio-demographics: the socio-demographic variables included age, gender, marital status (single, married) and questions on past involvement in a vehicle-collision (accident history).
Statistical analyses
Data were analyzed using SPSS statistics for windows version 25.0. The normality of the numeric variables was checked by the Kolmogorov-Smirnov test. Data were presented using mean (SD) for the numeric normal variables and frequency (percentage) for categorical variables. The between-group comparisons of demographic variables were done by the Chi-square test. To determine whether perceived risk and severity, attitude towards traffic regulations and risky road crossing behaviors differed between participants who had and had not been involved in an accident, univariate independent t-test was used and to adjust for sex and education a univariate analysis of covariance (ANCOVA) was utilized.
Additionally, to compare all constructs of TPB between participants with and without history of vehicle-collision, a Multivariate Hoteling T2 Tests based on Wilks' Lambda was conducted and also to adjust for sex and literacy level, multivariate ANCOVA based on Wilks' Lambda was done.P-values less than 0.05 were considered as significant.