RISK PERCEPTION, SELF-CONTROL, SELF-REGULATION, ABNORMALITIES ON RISKY DRIVING TENDENCY WITH DEFENSIVE DRIVING AS MEDIATOR

Background: Reckless bus drivers are still roaming on the highway. They usually make reasons like they are chased by deposits or so as not to be sleepy so they do the action. Even though when riding a bus, the driver certainly carries passengers and must be responsible for their safety. Aim: To provide an overview or model of intervention in dealing with risky driving behavior that is influenced by perceptions of safety risk, self control, self regulation, normlessness as factors behind risky driving behavior in Intercity Interprovincial Bus drivers, which is mediated by defensive driving behavior. Method: This study used a mix method, namely the calculation of questionnaire results using SmartPLS to see the model and the discussion of research uses more qualitative. The data used are secondary data, and primary data collected through filling out a closed questionnaire with a Likert scale of 1-5. The total respondents sampled in this study ranged from 50 bus drivers from PO Santoso and PO Handoyo at Bekasi City terminal, so the total number of respondents is around 100 people. Findings: There is a correlation or influence between risk perception, self control, n ormlessness and self regulation, significantly on risky driving behavior with defensive driving as a mediator. Risk perception had a significant influence on risky driving behavior. On the other hand, normlessness is very influential significantly because there is defensive driving as a mediator so that it is very effective if there is no defensive driver as a mediator.


INTRODUCTION
Transportation is growing more and more over time.Most people prefer to use land transportation compared to other types of transportation such as water transportation or air transportation (Onokala & Olajide, 2020;Park et al., 2019).There are more types of land transportation modes and easy to reach such as urban transportation, buses, motorcycles, trains to rickshaws are also used as traditional means of transportation (Newman et al., 2016;Vasconcellos, 2014) to meet the needs of the community for transportation which is expected to provide smoothness in carrying out activities out of town or province.Judging from this fact, it is not wrong if public transportation is considered to be needed by the community, especially for people who come from the island of Java.
For intercity transportation between provinces such as Bandung, Jakarta, Surabaya, to Sumatra, many bus fleets have been served, both based in Magelang and other cities.Especially for the direction of the western islands of Java and Sumatra, bus services are dominated by night buses.Some of the fleets that serve residents for this purpose include PO Santoso, Handoyo, Ramayana, and Putra Remaja (Teenage Son).One of the intercity and interprovincial (AKAP, Antar Kota Antar Provinsi) buses that serve the Java-Bekasi travel route is the intercity and interprovincial (AKAP) buses of PO Santoso and PO Handoyo which are one of the buses that are widely chosen by the people of Java and Bekasi in determining their choice of transportation mode for cross-provincial travel.On average, all of these fleets declare themselves as fast night buses.
With the development of the transportation system, the procurement of supporting and complementary facilities as well as road quality becomes very important (Bauchinger et al., 2021;Shladover, 2018).For this reason, the existence of facilities and infrastructure and other supporting facilities must have a size and standard both in quality and quantity that is safe, smooth, comfortable, and guaranteed.A transportation is said to be good, the travel time is fast enough, does not experience congestion, the frequency of service is sufficient, safe (free from possible accidents), and comfortable service conditions (Chalirafi et al., 2021;Dwiatmoko et al., 2022;Naumov et al., 2020;SInulingga, 1999).To achieve ideal conditions like this, it is determined by various factors that become components of this transportation, namely, the condition of infrastructure (Roads) and network systems and the condition of facilities (Vehicles) which is no less important is the mental attitude of users of these transportation facilities.Since this mental attitude is not taken into account, there will be many traffic violations committed by the bus drivers.
Traffic violations committed by bus drivers are endless in Indonesia.In videos that have gone viral on social media, it can be seen among them the drivers of inter-provincial (AKAP) buses.They do not seem to care about themselves, let alone passengers and even other people around them."Reckless" behavior or irregularities trip by bus drivers not only disturbs other road users, but also might cause fatal accidents (Bösehans & Massola, 2018;Celiński, 2023).The victims not only suffered injuries, but could die (Ahmed et al., 2023;Radam et al., 2022).This "reckless" behavior can occur due to certain factors from public unrest, such as traffic jams and other road traffic violations (Alavi et al., 2017;Martí-Belda et al., 2019).Those behaviors can make one forget their own safety, passengers, and even others around them.
Data from the Central Statistics Agency (BPS, Badan Pusat Statistik) (2015) states that in 2015 in Indonesia, the number of traffic accidents reached 98.9 thousand cases.This figure increased by 3.19% compared to the previous year which reached 95.5 thousand cases.The number of traffic accidents in the last 10 years has fluctuated, the highest increase occurred in 2011, reaching 108 thousand cases.In fact, in 2010 there were only 66.5 thousand cases.While the most cases occurred in 2012 with 117.9 thousand cases.The figure continues to fluctuate from year to year.This shows that the number of traffic accidents in Indonesia is in a state of concern.
The World Health Organization or WHO states that every year 1.3 million out of 50 million people die in traffic accidents.According to the Director General of Land Transportation, Pudji Hartanto, during the preparation of the Road Safety Campaign in Jakarta, the Police data, in Indonesia, an average of 3 people die every hour due to road accidents.The data also states that the large number of accidents is caused by several things, namely 61% of accidents are caused by human factors, namely those related to the ability and character of the driver, 9% are caused by vehicle factors (related to meeting roadworthy engineering requirements) and 30% are caused by infrastructure and environmental factors.While the human factor, which is related to the ability and character of the driver, turns out to be an influential factor in road safety.
Various efforts have also been made to deal with this problem until the Ministry of Transportation.Via its social media accounts, The Ministry of Transportation has recommended for drivers to avoid "reckless" behavior on the road which is now rife viral on social media lately.The instruction is also written in Law Number 22 of 2009 article 115, which reads: "It is forbidden to drive a vehicle beyond the highest speed limit allowed, and or race with other vehicles".
It does not stop there, still in the same rules there are also sanctions that will be imposed when someone dares or "recklessly" commits a violation.The high rate of injuries and deaths due to traffic accidents cannot be separated from individual behavior when traffic.In general, traffic accidents occur because individuals carry out risky behaviors when driving (Ragin, 2015).Risky driving behavior is manifested in various forms such as driving at high speed, drunk driving, driving without a license, not using helmets, man-belts, vehicle modifications, violation of signs, and others (Ragin, 2015;Toroyan, 2015).
Risky driving behavior is behavior that can endanger individual riders and other individuals around them.Risk behavior becomes something that is concerning if it has been commonly done since adolescence.According to Ragin (2015), risky behaviors carried out when adolescents can be carried into adulthood.Risky behavior can become a repetitive habit.The repetition of such behavior occurs because the individual feels accustomed to gaining an advantage when doing it Based on the background of the above problem, the researcher is interested in analyzing the influence of risk perception, self-control, self-regulation, abnormality and responsible driving on risky driving behavior mediated by defensive driver.The current research is one of the studies related to transport psychology with the theme road savety, as a consideration in improving the implementation of policies related to the traffic transportation system that prioritizes road safety and psychological aspects of drivers, passengers and the surrounding community.

METHOD
The research method will use a mix method, namely the calculation of questionnaire results using smart pls to see the model and the discussion of research uses more qualitative.First secondary data then primary data collected through filling out closed questionnaires with a likert scale of 1-5.The total respondents who were sampled in this study ranged from 50 bus drivers both from PO. Santoso and PO Handoyo at the Bekasi City terminal, so the total respondents were around 100 people.This research will be conducted using Structural Equation Modeling using the Partial Least Square approach and processed with the help of SmartPLS software.

Figure 1. Chart of Research Flow
The research was an exploratory, conducted in several places located in Bekasi City, West Java from February to September 2021.The population of this study was intercity and interprovincial (AKAP) bus drivers PO Santoso and PO Handoyo who qualified as research subjects.The requirements for the subjects of this study are drivers who work at PO Santoso and PO Handoyo on the Java-Bekasi route who have the status of permanent drivers.The locations were in the Pool PO Handoyo and PO Santoso located on Jalan H.M Joyo Martono Margahayu kec.East Bekasi, Bekasi City, West Java and Bekasi Terminal.
The sampling technique in this study was non-probability techniques, namely incidental sampling techniques, sampling techniques based on chance, that is, anyone who incidentally meets the researcher can be used as a sample, if considered suitable as a data source.While the research questionnaire was given directly to respondents because the scope in this study was still affordable by researchers.Each indicator item is measured using the Differential Semantic Scale, where there are two different poles, the negative pole is given a score of 1 and the positive pole is given a score of 5.The score of each indicator is the total value of the indicator items.The population in this study was unknown.The formula for calculating samples in unknown populations uses the Lemeshow4 formula as follows:
Analysis of research data was carried out using the SEM analysis model.Before testing the hypothesis, researchers go through three kinds of assumption tests in the form of normality tests, linearity tests, and multicollinearity tests.All forms of statistical analysis from the trial stage to research were carried out using SPSS for Windows version 20 and Smart PLS programs.

RESULTS AND DISCUSSION Path Drawing to Show Causality Relationship
The theoretical model that has been built on the hypothesis will be depicted in a flowchart.The path diagram will make it easier for researchers to see the causality relationships they want to test.Here is the flowchart in this study: The blue round shape in the figure above represents the variables observed using the questionnaire.For example, Risk Perception is formed from two indicators (X.1 with a value of 25,306, X.2 with a value of 45,470) and each indicator has an error rate of e1, e2 respectively.Likewise, other constructs are formed from each indicator.In SmartPLS applications, the causality relationship is simply described in a flowchart.Then, the program language converts the image into an equation, the equation into an estimate.

Test Validity and Reliability
The validity test is used to determine whether or not the question is feasible (valid).The decision criterion is to compare the value of Corrected Item -Total Correlation compared to the r value of the table with a level of (α) 0.05 which is 0.176.If the value of Corrected Item -Total Correlation is greater than r of the table, then the indicator is feasible (valid) and vice versa.
After being tested for validity, the next step is reliability testing, which is related to the problem of accuracy of a data, while for reliability testing through the value of alpha coefficient compared to a value of 0.60.A construct or variable is said to be reliable if it has an alpha value above 0.60 and vice versa.Based on the results of calculations with the SPSS program, validity and reliability testing can be presented in Table 3 below.The table above shows that all indicators (observed) are valid, this is indicated by the value of Corrected Item -Total Correlation > r table (0.176).This proof shows that all indicators (observed) are worthy of being used as indicators of constructs (latent variables).The alpha coefficient (cronbach alpha) has a value above 0.60 so that it can be explained thatthe above variables are reliable or have high reliability, so they have high accuracy to be used as variables (constructs) in a study.

Data Normality Test
Data normality is one of the conditions for SEM operations to process the modeling made.In assessing the normality of the data in this study was carried out by looking at the z value owned.This z value will be seen with a critical value determined through a specified significance level of 0.01 and a critical value of ± 2.58.In data processing with IBM Amos 20, the z value is seen from the critical ratio.In the Assessment of Normality output above, it appears that there is no critical ratio value above + 2.58 or below -2.58.For Skewness, the highest value is -2.462 and for Kurtosis, the highest value is -1.715 (KP2).Based on these results, it can be stated that the assumption of normality is fulfilled.

Hypothesis Test Factor Confirmatory Analysis
To test the uni-dimensionality of exogenous and endogenous constructs, confirmatoryfactor analysis techniques are used.If the resulting probability is significant, then the hypothesis stating that there is no difference between the sample covariance matrix and the estimated population covariance matrix cannot be rejected or the null hypothesis accepted.Confirmatory factor analysis is done by making a relationship of two arrows between each construct that symbolizes the correlation between the two constructs.The following are the results of the confirmatory factor analysis in this study.

Figure 3. CFA Analysis Results
The estimation results in the figure above give a Chi Square value of 47.936 with a value of p = 0.848.It appears that p exceeds the significance level value (p>0.05) which states that there is no significant difference between the sample covariance matrix and the population covariance matrix.The GFI value of 0.936 and AGFI = 0.901 are also above the recommended value (0.90).The Cmin/df value gives a value of 0.812 (less than 2.00), then the RMSEA value of 0.000 (less than 0.08).The TLI value is 1.018 (exceeding the value of 0.95) and the CFI is 1.000 (exceeding the value of 0.90), so that the results of the CFA analysis have met the expected requirements.This means that the model is marginally acceptable.

Causality Analysis (Structural Equations)
Estimation analysis is performed by analyzing the full-model to see the suitability of the model and the causality relationship built in the model being tested.The full model is done by replacing two correlation arrows with one arrow that symbolizes the hypothesis given in this study.The following are the results of the full model estimation in this study: The estimated standard error value displayed in the Output shows that the largest standard error for the indicator is 0.34 (e5).The value is relatively small (< 0.4) so there is no problem identifying the standard error.The problem occurs if there are one or more standard errors whose values are greater than 0.4.

Evaluation of Goodness of Fit Model
Model conformity testing is evaluated through analysis of various goodness of fit criteria.For this reason, the first action taken is to evaluate whether the data used can meet the assumptions needed in SEM analysis.When this assumption has been met, then the model can be tested.Some important measurements in evaluating goodness-of-fit criteria along with cut of value are as follows.The Chi-Square value is 47.936 with a probability value (p-value) of 0.848 (p > 0.05).The probability value meets the recommended value standards so that the model is feasible to be used as a tool to confirm the observed data.The GFI value of 0.936 is above 0.90, then the Cmin/df value of 0.812 is below 2, the RMSEA value of 0.000 is below 0.08.Furthermore, the TLI value = 1.018 and CFI = 1.000 is above 0.95, the AGFI value of 0.901 is above the recommended value of 0.90.Based on the results of the analysis, the model is declared marginally feasible.

Hypothesis Test
The test was carried out by looking at the critical ratio of regression weight.Here is the hypothesis test in this study:  1) Hypothesis 1: The c.r. value between Normlessness  Defensive Driving is 3.596, greater than 1.96.Thus, this means that Normlessness has a significant influence on Defensive Driving, thus hypothesis 1 in this study which states that "There is a significant effect of Normlessness on Defensive Driving" is accepted.2) Hypothesis 2: The c.r. value among Devensive Driving  Risk Perceptions is 4.467, greater than 1.96.This means that Risk Perception has a significant influence on Defensive Driving.Thus hypothesis 2 in this study which states that "There is a significant effect of Risk Perception on Defensive Driving" is accepted.3) Hypothesis 3: The c.r. value among Defensive Driving  Risky Driving is -2.080 greater than 1.96.This proves the significant influence of Risk Perception on Risk Driving.Thus hypothesis 3 in this study which states that "There is a significant effect of Defensive Driving on Risky Driving" is accepted.4) Hypothesis 4: The c.r. value between Normlessness  Risky Driving is 3.062 or greater than 1.96.This proves the significant influence of Normlessness on Risky Driving.
Thus hypothesis 4 in this study which states that "There is a significant effect of Normlessness on Risky Driving" is accepted.5) Hypothesis 5: The c.r. value among Risk Perceptions  Risk Driving is 2.539 or greater than 1.96.This proves the existence of a significant effect of Normlessness on Risky Driving Thus, hypothesis 5 in this study which states that "There is a significant effect of Normlessness on Risk Driving" is accepted.

Direct and Indirect Influences Direct Effect
Direct influence is shown with one arrow on the diagram and according to the hypothesis given.Here is a direct influence on this research model.

Table 7. Direct Influence
N PR DD MB DD 0,249 0,379 0,000 -0,376 MB 0,000 0,000 -0,376 0,000 Source: Primary data processed The table above shows that the direct effect of Normlessness on Risk Perception is 0.249; The effect of Risk Perception on Risk Perception is 0.379 and the effect of Defensive Driving on Risk Driving is -0.376 because the diagram is connected by one arrow ().The effect between other variables is 0 because there is no direct relationship with one arrow ().Based on the results of the analysis, it appears that the influence of Risk Perception on Defensive Driving is greater than the influence of Normlessness on Defensive Driving (0.379 > 0.249).

Indirect Effect
The advantage of SEM is that it is able to identify indirect influences between exogenous constructs on endogenous constructs that are not aimed at and arrows directly.Here are the magnitude of indirect influences on the research model: Table 8.Indirect Influence N PR DD Risky Driving DD 0,000 0,000 0,000 0,000 Risky Driving -0,094 -0,142 0,000 0,000 Source: Primary data processed Based on the table above, it appears that Normlessness has an indirect influence on Risk Driving of -0.094 and there is also an indirect influence between Risk Perception of Risk Driving of -0.142.This empirical evidence also proves that the effect of Risk Perception on Risk Driving through Risk Perception is higher than the effect of Normlessness on Risk Driving through Defensive Driving (0.379 > 0.249).

Total Effect
Total influence is the sum of the direct and indirect influences contained in the research model.Here is the magnitude of the total influence on the research model: Based on the table above, it appears that Risk Perception has a total influence on Defensive Driving of 0.370 and on Risk Driving of 0.283, so that the total value reaches 0.653.Normlessness has a total influence on Defensive Driving of 0.494 and on Risk Driving of 0.206, so that the total value reaches 0.7.The effect of total Defensive Driving on Risk Driving is -0.379.

Discussion
The test results against hypothesis 1 state that Risk Perception has a significant effect on Defensive Driving.Risk Perception as variable X has a significant effect on variable Y.What distinguishes the results of this study from the results of other researchers is that the researchers tested the direct influence of variable X on variable Y, while this study used an intervening variable or intermediate variable, namely Defensive Driving, which bridges the indirect influence of variable X (Risk Perception) to variable Y (Risky Driving).Thus it can be formulated as the findings of this study is that "The higher the Risk Perception owned by intercity and interprovincial (AKAP) drivers at risk, the more confident and comfortable consumers will be in using intercity and interprovincial (AKAP) buses because the drivers have a good level of risk perception." The test results against hypothesis 2 state that Normlessness has a significant effect on Defensive Driving.From this study, trait Normlessness as variable X has a significant effect on variable Y.What distinguishes the results of this study from the results of other studies is that the researchers tested the direct influence of variable X on variable Y, while this study used an intervening variable or intermediate variable, namely Defensive Driving, which bridges the indirect influence of variable X (Normlessness) to variable Y (Risky Driving) so that The result of Defensive Driving mediation results in Normlesness that can be minimized Thus it can be formulated as a finding of this study is that "the higher the benefits offered by the AKAP Bus (in terms of the Driver and the comfort of the Transportation Facilities) to customers, the greater the customer's trust in PO Santoso and PO Handoyo AKAP Buses".
The test results against hypothesis 3 state that Defensive Driving has a significant effect on risky driving.This is in accordance with the results of research conducted by that Defensive Driving has a significant effect on variable Y.What distinguishes it from other researchers is that they test the direct effect of Defensive Driving as an independent variable X on variable Y, while this study makes Defensive Driving as an intervening variable.Thus, it can be formulated that "The more customers trust PO Santoso and PO Handoyo intercity and interprovincial (AKAP) buses, the less risky driving tendencies will be." The test results against hypothesis 4 state that Risk Perception has a positive and significant effect on risky driving tendencies.It can be concluded that Risk Perception as variable X has a significant effect on variable Y.The difference is that this study uses an intervening variable or intermediate variable, namely Defensive Driving, which bridges the indirect influence of variable X (Risk Perception) to variable Y (Risk Driving).Thus, it can be formulated that "The more savety driving offered by PO Santoso and PO Handoyo intercity and interprovincial (AKAP) buses, the more consumers will trust the buses.And the higher the customer's trust in PO Santoso and PO Handoyo, the greater the customer's monat to continue using their services".

CONCLUSION
The results of this study show that there is a significant positive relationship between risk perception, Self Control, Self Regulation, Normlessness on the tendency of risky driving behavior in intercity and interprovincial (AKAP) bus drivers PO Santoso and PO Handoyo Java-Bekasi route.The higher the risk perception, Self Control, Self Regulation and trait normlessness mediated by defensive driving, the lower the tendency of risky driving behavior.Vice versa, the lower the perception of risk, Self Control, Self Regulation the higher the tendency of behavior to drive riskily.Thus the hypothesis proposed in this study is accepted, namely there is an influence between risk perception, Self Control, Self Regulation and risky driving behavior in intercity and interprovincial (AKAP) bus drivers PO Santoso and PO Handoyo.
For the Transportation Service Provider, it is expected to provide defensive driving training to each driver in order to get competency skills so that the driver also knows his competence as a driver who must carry passengers as comfortably as possible so that customers can still give their trust and loyalty to the bus.As for the police, trusted figures can provide examples of good driving behavior to the public and road users.
For future research, it should be carried out in a population with a wider scope to expand the generalization of research results.The components of the measuring instrument can be adjusted so that the measuring instrument can be filled by respondents according to the existing culture.It is better to update the measuring instrument so that it is expected to be able to measure variables more stably.Further researchers are also advised to pay more attention to gender categorization by choosing one gender category so that it is easier to categorize risky driving tendencies.

Table 3 .
Reliability and Validity Test Results of the Questionnaire The results of the normality test can be seen in the table below:

Table 5 .
Model Accuracy Test

Table 9 .
Total Influence