Performing Structural Equation Modeling in Public Transport Through COVID-19 Pandemic Time

During the COVID-19 pandemic, transportation has been the sector most severely impacted. Global spread of the COVID-19 virus leads WHO to declare a pandemic on March 11, 2020 [1]. This pandemic has changed the government’s policy in order to reduce the faster spread of the COVID-19. These policies included the implement curfews and safer-at-home orders which affected on all economic activity and increasing unemployment rates. So that, the public transport services and ridership are having significantly effort to keep physical distancing continued until the danger from COVID-19 is outcaste. All this reason reduction the mobility on the streets which decrease the traffic volume during the pandemic lockdown [2]. The World Health Organization (WHO) recommended that it should keep 1m between people and don’t touching the surface mainly in close areas to prevent virus transmission. Public transport identifies as a crowded space thus it is challenging to keep people within a social distance which makes public vehicles an excellent indoor environment for virus transmission. Using public transportation can increase the spread of viruses by a factor of up to six [3]. As a result, there was a worldwide drop in public transportation utilization during the COVID-19 pandemic [4]. In other wise, some people who are unable to drive, walk and cycle, public transport may be the best choice.

After a lot of cities reopen with hope to live without worry from virus, there is a pressing demand to take a quickly strategic to prevent completely lockdown. After the lockdown restrictions ease, the mobility shows a significant shift to private cars in cities after the experience of COVID-19 [5]. More private cars on the road would make cities to get a traffic jams which increased air and noise pollution, greenhouse gas emissions, obstruct walking and cycling, which make life difficult for people with lower-income who have not car [6]. Therefore, it should make some strategies as soon as possible for return the public transport system to transport sector under some restriction in order to confined transmission of virus.

The vehicle manufactures and the operators of public transport have a great effort to make space condition of public transport more safely by make door sensors, use hand sanitizer dispensers and redesign the internal seats for having physical barrier to reduce direct contact between the people [7]. Precautions, such as wearing face masks and maintaining the necessary physical distance, have been shown to reduce the risk of virus transmission [8]. It's worth noting that public transportation's physic social qualities-imposed limits on its use long before the COVID-19 [9]. Also, Aditjandra et al. [10] reported by development structural equations model that travel attitude characteristics and accessibility features of the urban area may change the use of public transport.

The researchers try to understand and predict travel modes and people behavior during and post the pandemic. Though there is research studying people's intentions to use public cars in general, under restrictions, and when infected with COVID-19, the focus on alternative transportation modes during the pandemic has been lacking [11]. The study by Dzisi and Dei [12] took a prominent route in Kumasi, where researchers observed and analyzed the habits of over 850 of the city's most popular paratransit buses. While Ministry of Transport criteria were used to assess passengers' adherence to the policy on physical separation, the researchers counted the number of passengers on each bus to establish whether or not the policy on face masks was being followed. The results show that most buses adhere to the social distance regulations established by the Ministry of Transportation; nevertheless, the policy on face masks is only partially adhered to in most vehicles. Only around 12.6% of the vehicles had less than three passengers without face masks, and only about 21.3% of the buses had fewer than three passengers with face masks. COVID transportation policy and practice responses were aided by the introduction of the new concept of Responsible Transport [13]. What makes this concept novel is that it takes into account not just environmental factors related to sustainability, but also the health and well-being of individuals and communities. More than that, it highlights the role of the person as an independent, responsible agent in achieving socially desirable transportation outcomes. COVID-19 transportation policy and practice responses were aided by the introduction of the new concept of Responsible Transport. What makes this concept novel is that it takes into account not just environmental factors related to sustainability, but also the health and well-being of individuals and communities. More than that, it highlights the role of the person as an independent, responsible agent in achieving socially desirable transportation outcomes. Abdulrazzaq et al. [14] devised a methodology for people to use in order to make the transition from driving alone to taking public transportation. According to the results, an individual's mode of transportation selection is heavily influenced by their trip distance, daily trip frequency, trip duration, gender, age, and occupation. During COVID-19, Javid et al. [15] looked at how people felt about taking public transit on their travels. Using a questionnaire, researchers in Lahore, Pakistan, investigated passengers' knowledge, sense of duty, and moral duties toward adhering to safety standards and norms when using public transportation. The results of the structural equation model showed a favorable relationship between passengers' awareness of consequences and their willingness to take responsibility for their actions. Passengers' willingness to use public transit is significantly affected by their perception of the difficulty of complying with safety rules, such as wearing a facemask, using sanitizers, and maintaining a social distance. Travelers' preferences for public transportation may be significantly influenced by their level of familiarity, sense of duty, and trust in its use during a pandemic. During the pandemic, Abdullah et al. [16] constructed a binary logistic model to represent travel behavior in terms of individual modes (i.e., private cars, taxis, etc.) and public modes. It has been found that women are more likely than men to ride in public transportation vehicles of a different mode.

A conceptual model according to earlier studies has been proposed in this study. The concept of public transport user model are behavior, attitude, perception, and responsibility. Theoretically, the process in this study can provide a reference by employing model users’ intentions to use public transport while adhering to the safety instructions during the COVID-19 pandemic in Iraq.

3.1 Reliability and validity measurements for SEM model

In order to determine the reliability of the dimensions and to examine the variables' degree of internal consistency, the Cronbach's alpha test was carried out. The fact that the Cronbach's alpha value for the Responsibility variables is less than 0.500 makes it unacceptable as a measure of reliability [18, 19]. The value was found to be 0.584. The item code Res1 was removed so that we could go within the range of the Cronbach's alpha test that is considered acceptable. After one item was removed, the values of all the variables and items that make up Cronbach's alpha lie between 0.623 and 0.692, as shown in Table 3. Following the completion of the reliability test, a multitude of additional tests can be carried out to validate the suitability of the factor analysis (FA) of the data sets. Kaiser-Meyer-Olkin (KMO) is a test that determines whether or not there was sufficient sampling, and factor loading identification and Bartlett's test of sphericity are tests that are used in the process of analyzing the dependability of each factor on its own. The Kaiser-Meyer-Olkin (KMO) measure and Bartlett's test of sphericity were applied to the findings of the pilot study, and the results of both of these tests are shown in Table 4. This allowed for each of the 17 items on the questionnaire to be validated. The KMO measure of sample adequacy was found to be 0.688, which ensures the occurrence of significant partial correlations across the entire investigation. The requirement that the measure of sampling adequacy be more than 0.50 is a prerequisite for the criterion that was proposed by Hair et al. [20] to be evaluated to validate the existence of correlation.

Table 3. Reliability test results for pilot survey before and after items deleted

Table 4. KMO and bartlett's test for pilot study

3.2 CFA measurement model

In order to validate the results of the measurement scale test, a confirmatory factor analysis (CFA) was carried out. During the course of the investigation, a total of 234 valid samples were found. In the context of the CFA, dimensions are referred to as latent constructs, whilst attributes are referred to as reflective indicators. In order to evaluate the measurement model, we looked at how well it suited the data overall, as depicted in Figure 4. In this analysis, nine goodness-of-fit indices derived from SEM-based research and the SEM references [20-22] were taken into consideration. In order to validate the results of the measurement scale test, a confirmatory factor analysis (CFA) was carried out. During the course of the investigation, a total of 234 valid samples were found. In the context of the CFA, dimensions are referred to as latent constructs, whilst attributes are referred to as reflective indicators. In order to evaluate the measurement model, we looked at how well it suited the data overall, as depicted in Figure 3. In this analysis, nine goodness-of-fit indices derived from SEM-based research and the SEM references [20-22] were taken into consideration. These indices are listed as follows:

1. P-Value;
2. Chi-square;
3. Chi-square/df (ratio of Chi-square to degree of freedom);
4. RMR (Root Mean Square Residual);
5. GFI (Goodness-Of-Fit index);
6. AGFI (Adjusted Goodness-Of-Fit index);
7. NFI (Normed Fit Index);
8. TLI (Tucker-Lewis Index);
9. CFI (Comparative Fit Index);
10. RMSEA (Root Mean Square Error of Approximation).

4.jpg

Figure 4. Structural equation model research process [17]

According to Figure 3, values that are more than 0.600 reflect the leading factors, providing evidence of convergent validity. Figures with an R² greater than 0.600 were found to be common for all indicators, which shows that definitive dependability has been reached. On the basis of the results of AMOS 24.0, this model is an adequate match because its indices have satisfied the requirements of a model fit, as indicated in Table 5.

Table 5. Fit indices for study SEM

According to Table 5, CFA was used to estimate the convergent and discriminant validities by using the AMOS 24.0 software. The overall goodness-of-fit indexes of the Confirmatory Factor Analysis (CFA) measurement model suggest a satisfactory fit. All measurements in Table 5 confirm that the structural model adequately fits the data, according to references [20, 23, 26, 27]. However, Baseline Comparisons refers to the models automatically fitted by Amos for every analysis which is not adequate due to not reaching thresholds.