Travel behaviour changes and risk perception during COVID-19: A case study of Malaysia

The COVID-19 pandemic has disrupted travel behaviours due to the need for movement restrictions. The restrictions adversely affected various aspects of health and the economy. This study aimed to investigate factors affecting trip frequency during the recovery period of the COVID-19 pandemic in Malaysia. An online national cross-sectional survey was conducted to collect data in conjunction with different movement restriction policies. The questionnaire includes socio-demographics, experience with COVID-19, risk perception of COVID-19, and trip frequency on several activities during the pandemic. Mann Whitney U was conducted to determine whether there were statistically significant differences between the socio-demographic factors for the respondents in the first and second surveys. Results show no significant difference in socio-demographic factors except for the level of education. The results indicate that the respondents from both surveys were comparable. Next, Spearman correlation analyses were conducted to find significant correlations between trip frequencies toward socio-demographics, experience with COVID-19 and risk perception. There was a correlation between the frequency of travel and risk perception for both surveys. Regression analyses were performed based on the findings to investigate trip frequency determinants during the pandemic. Perceived risk, gender, and occupation influenced the trip frequencies for both surveys. By understanding the influence of risk perception on the frequency of travel, the government can identify the appropriate policy during a pandemic or health emergency to avoid impeding normal travel behaviour. Thus, people's mental and psychological well-being are not negatively affected.


Introduction
In December 2019, the World Health Organization reported 44 cases of severe pneumonia caused by a novel coronavirus  in China (WHO, 2020a). The virus was then spread rapidly to other regions of the world; by January 2020, 2,798 cases had been reported worldwide (WHO, 2020b). Two months later, the situation even aggravated when the cases rose sharply in 114 countries, i.e., 118,000 cases with 4,291 deaths. The fast spreading of the virus caused a worldwide declaration of a pandemic (WHO, 2020c). To curb the transmission of the virus, governments worldwide have implemented the lockdown protocol (Haider et al., 2020;Jang et al., 2021;Sarda et al., 2022;Verma et al., 2020). The lockdown protocol has led to social isolation due to suspending all public transport services, shop closures, and curfews. This situation has caused emotional distress among the general population worldwide and, at the same time, increased the risk perception in daily travel (Motta Zanin et al., 2020).
Due to the severe impact of COVID-19, vaccination programmes were launched worldwide to restrain the spreading of viruses (Eccleston-Turner & Upton, 2021). In December 2020, the United Kingdom became one of the first countries to begin mass vaccination with the COVID-19 vaccine. The vaccine acceptance rate amongst countries was mixed. Some countries with the highest vaccine acceptance rates were Ecuador, Malaysia, Indonesia, and China (Sallam, 2021). On the other hand, the countries with the lowest acceptance rate were Poland, the US, and France. Despite the said scenarios, the effort of vaccination programmes has been continuous, and positive outcomes have been reported after a few months it was launched, i.e., the decreases in the number of deaths (Public Health Ontario, 2022; UK Health Security Agency, 2022) and infection rates (AstraZeneca, 2021) in addition to the faster economic recovery rate (Deb et al., 2022). High acceptance and positive attitudes toward the vaccination program are due to the high-risk perception of the effect of COVID-19 (Guazzini et al., 2022;Bughin et al., 2020). Many people perceived vaccination as an effort to overcome the profound impact of the restriction movement on work, income, and daily life Lin et al., 2020;Abedin et al., 2021). After the vaccination programme has acquired herd immunity, many people start recommencing routine activities.
Vaccination is indeed one of the interventional strategies to combat COVID-19. Despite the acceptance of vaccination programs worldwide, travel activity did not instantly return to normal. Consequently, throughout the pandemic era, there has been a decline in travel, particularly those involving work trips (Pawar et al., 2020), shopping (Schmidt et al., 2021), and recreation (Bh & TM, 2021). Such trip reduction is related to the lockdown and movement restrictions (Politis et al., 2021) and the perception of infection risk (Ozbilen et al., 2021). However, some individuals need to travel to earn a living and meet their daily necessities. It is interesting to identify who is still involved in travelling during the pandemic and how they perceive risk.

Malaysia context
Malaysia is no exception to the impact brought by COVID-19. In January 2020, Malaysia reported its first 3 cases of COVID-19 Shah et al., 2020). Those confirmed cases kick-started Malaysia's first wave of COVID-19 infection. Based on being concerned about the adverse effects of the spread of this virus, the Malaysian government has announced a national restriction movement starting March 18, 2020 (Elengoe, 2020). New Standard Operating Procedures (SOPs) were introduced, such as mandatory usage of face masks, physical distancing, and a temperature check at the entrance of every premise (Fadzlina et al., 2020;Mei Ching et al., 2021).
Since then, Malaysia has undergone a variety of restriction movement policies, such as the Movement Control Order (MCO) and the National Recovery Plan (NRP). The restrictions made by the Malaysian government have caused activity-related travel behaviour to be limited in which; people in essential sectors are allowed to go out to work, while those in non-essential sectors are encouraged to work from home (Rajendran et al., 2021;Tay et al., 2021), and influenced the people's risk perception of getting an infection while travelling (Md Yusuf & Azhar, 2021). The public's mental health and well-being are impacted when their ability to travel is impeded (Knolle et al., 2021). Additionally, travel restrictions also have detrimental effects on the economy (Rossi et al., 2020;Meyerowitz-Katz et al., 2021;Brodeur et al., 2021).
Thus, in line with WHO policy to restrain the transmission of COVID-19, the Malaysian government has launched the National Immunisation Programme (NIP) to facilitate the distribution of vaccines in the country. The first phase of the massive vaccine rollout in Malaysia began on February 26, 2021, followed by the second phase in April 2021. It kickstarted Malaysia's early stage of achieving herd immunity. This movement relaxation, coupled with National Recovery Plan (NRP), thus marked the country's beginning of recovery from the pandemic. During the third phase of NIP, the government announced that Malaysians with two vaccine doses could perform interstate travel and travel overseas (MalaysiaNow, 2022).
Some researchers studied the impact of COVID-19 on travel behaviour in Malaysia, including mode choice during the pandemic (Dias et al., 2021) and the effect on the tourism supply chain (Hamid et al., 2021). In addition, some of the researchers also investigated travel behaviour and risk perception during the Movement Control Order (MCO). For example, Aziz and Long (2022) focused on perceived travel risk among tourists, Wong and Alias (2021) focused on psycho behaviour using Health Belief Model (HBM), and Dias et al. (2021) focused on risk perception, trip frequency and mode choice. Because of the travel restriction in effect during the MCO, it is conceivable that these studies report fewer trips were made due to the high risk of COVID-19. In addition, the acceptance towards the vaccination program was still in the early stage.
Meanwhile, this paper focused on trips after the movement restrictions were lifted. This study was conducted in two different periods to indirectly examine the effect of the National Recovery Plan (NRP) and the National Immunisation Programme (NIP). Fig. 1 displays the timeline of COVID-19 cases, restriction movement periods, and data collection. The first data gathering was performed after the National Recovery Plan (NRP) announcement that enables free movement within state borders. In addition, even though daily cases are still rising during this period, the National Immunisation Programme (NIP) was in phase two, where the elderly groups have been vaccinated.
Once nationwide approval for the free movement was announced, the second round of data collection was conducted. At the same time, the number of cases was declining, and the National Immunisation Programme (NIP) began the vaccine phase with a focus on all adults. However, policy changes related to movement during the pandemic indirectly influence the trip frequency and community perception. Fewer trip frequency indicates that the situation has not normalized and that there is still a sense of pandemic infection, which might prolong the struggle with mental health and financial hardships. Therefore, the ultimate objective of this study is to determine the factors influencing the trip frequencies and activities during the recovery phase of COVID-19. The findings of this study are essential for the transportation authorities and stakeholders to plan a comprehensive strategy to face a pandemic or any emergency health situation. In addition, travelling activities should be resilient in any condition to avoid adverse mental and psychological health effects.

Travel behaviour changes during COVID-19
The spread of the novel coronavirus has affected people's travel behaviour around the globe, especially regarding trip purposes and trip frequencies, as it has forced people to remain at home. Previous studies reported that trip frequency on public transport usage reduced tremendously during the COVID-19 pandemic Jamal and Paez, 2020). In addition, offices, shops, institutions, workplaces, and schools were not operating physically. Politis et al. (2021) observed that the most significant travel decrease was in shopping, leisure and working. Meanwhile, a qualitative study by Yang et al. (2021) in Huzhou, China, reported that people only travelled for essential supplies such as food, face masks and medical supplies during the pandemic.
Bh & TM (2021) investigated a reduction in trip frequency based on psycho-attitudinal technique and found that people were more likely to reduce shopping trips for daily necessities compared to work trips. However, Abdullah et al. (2020) reported an increase of 40% in travelling primarily for shopping during COVID-19. In other countries, shopping for daily necessities tremendously increased to 164% (Anwari et al., 2021). This situation portrays the panic buying at the early of the pandemic attack. An increase was also observed in online shopping during the pandemic (Shamshiripour et al., 2020). It was caused by a decrease in physical shopping, where grocery shopping decreased by 34% (Parady et al., 2020). Another study stated that from late March to early April, the number of people who did grocery shopping outside their homes fell to around 8%. (de Haas et al., 2020).
Regarding outdoor leisure or recreational activities, a 78% decrease was reported in Bangladesh (Anwari et al., 2021). Parady et al. (2020) reported a decreasing trend for eating out and leisure, with an 87% decrease in the said activity. However, in the Philippines, travelling for recreation and visiting relatives increased during that period (Mayo et al., 2021). de Haas et al. (2020) found that in terms of outdoor activity, the youngest age group showed a significant reduction during the pandemic. Meanwhile, Irawan et al. (2020) found that younger people were more inclined to engage in leisure activities after the outbreak.
Some studies have found that outdoor leisure and recreational activities shifted to indoor activities. New hobbies such as calligraphy, planting flowers and baking had risen from 27.6% before to 36.5% during the pandemic for older adults in Shanghai, China (Ding et al., 2022). In Canada, (Fatmi et al., 2021) noted a decrease in outdoor travel activities by almost half, whereas indoor activities such as watching tv showed an increase, likely due to additional time spent at home. A study conducted in mid-May 2020 in Japan showed a 49% decrease in time spent outside, with 7% of that decrease being substituted by leisure activity/entertainment using the internet. (Yabe et al., 2021).
The closure of the workplaces also has affected the workers in several aspects. A study in Bangladesh done by Anwari et al. (2021) found that online working has increased tremendously by 953%. A similar result was observed in the Philippines (Mayo et al., 2021), as travel to work dropped by 11.2%. Understandably, during the pandemic, the number of people working from home has risen dramatically from 6% to 39% (de Haas et al., 2020). Before the pandemic, 71% of respondents in Chicago had never experienced working from home, but then the number decreased to 37% during the pandemic (Shamshiripour et al., 2020). Several studies also found that the behaviour toward working from home post-COVID-19 has changed. As Olde Kalter et al. (2021) pointed out in their study, the main factor was job characteristics. They found who did not telework prior to or during the lockdown have no intention of doing so in the future and vice versa. Office workers were more inclined to work from home after COVID-19. Interestingly, they found that teaching staff did not have the intention to work from home once the pandemic was over despite most of their work being shifted online during the pandemic. The marketing, sales, information technology and administrative workers tended to work from home during the pandemic (Barbour et al., 2021). Likewise, Politis et al. (2021) found that males were involved in higher work trips during the pandemic; meanwhile, women's duties increased at home.
With movements being restricted during the pandemic, social interactions are also affected. In a study done globally by research organisations from Europe, North Africa, Western Asia, and the Americas, there was a 58% decrease in family visits and approximately a 45% decrease in visiting neighbours/friends (Ammar et al., 2020). It has reduced the respondents' satisfaction with life by half (60% before confinement and 30% during confinement). In the USA, most studied participants either cancelled or postponed their social visits (Choi et al., 2022). However, during this pandemic, the use of online media has become more critical, as shown by Juvonen et al. (2021). Their study noted that social media and video calls were the most frequent methods of contacting friends during COVID-19. As the restrictions eased, more people were more comfortable meeting with friends as the planned number of trips significantly increased (Beck & Hensher, 2020).

Risk perception towards COVID-19
Risk perception is the people's instinctual assessment of a hazard to which they may be exposed . A previous study on the perception of risk of the COVID-19 pandemic found that people in Spain mainly reported a high-risk perception, especially in terms of getting infected with COVID-19 while travelling (Vich et al., 2022). In addition, individuals' preferences to avoid specific transport modes during the pandemic can be influenced by risk perceptions relating to personal health, crime, and traffic safety (Hotle et al., 2020;Ma et al., 2019). Ozbilen et al. (2021) stated that sharing public transport, such as buses and taxis, was perceived to have high transmission of COVID-19 compared to private modes, such as cars. It is parallel with Zafri et al. (2022) reported that most Bangladeshi perceived the highest risk of getting an infection while travelling by bus and ridesharing. Meanwhile, travel by private transport such as motorcycles and cars, cycling and walking were given a low perceived risk of getting infected. Vich et al. (2022) reported that the presence of foreign tourists in local buses increased the perceived risk of COVID-19 transmission by local bus users in Mediterranean cities. Thus, with public transport getting a negative representation during this pandemic, there is a fear that there will be a massive shift from public to private modes of transport in the future (Das et al., 2021;Habib & Anik, 2021;Kłos-Adamkiewicz & Gutowski, 2022).
Risk perception also plays a role in the success of protective measures implemented by a government. Siegrist et al. (2021) noted that the people who performed more hygienic behaviour and minimised their contact with other people accepted the policies implemented by the government compared to those who were against it. Through regression analysis, they also found that risk perception was the most crucial factor in their willingness to accept the policy measures to stop the spread of the disease.
Meanwhile, in South Korea and China, there was a difference in specific aspects of their risk perception (Chen et al., 2021). In South Korea, the risk perception of COVID-19 was affected by familiarity with the situation locally and levels of satisfaction with the government's policies. Meanwhile, people in China show their level of satisfaction and trust in the information on the government's websites.

Survey design
The questionnaire was designed to measure trip frequency changes and risk perception during the early stage of the nationwide immunisation programme (the first period) and recovery stage (the second period). The questionnaire was distributed Malaysian-wide in two different periods, with the first period being from 7th June 2021 to 27th August 2021 and the second period being from 18th October 2021 to 9th Mac 2022. Due to the social distancing requirements, the questionnaire was distributed through social media platforms such as Telegram, WhatsApp, Facebook, and personal contacts. Prior to data collection, the questionnaire was pre-tested through an online pilot study with 30 selected respondents.
The reliability of the questionnaire was evaluated using Cronbach Alpha. Furthermore, the result shows an acceptable value of 0.70 which signifies that the questionnaire has high reliability (Hinton et al., 2004). Face and content validities were also performed, whereby the respondents were asked about the complexity of the questions and their comprehensiveness towards the terminologies used in the questionnaire.
The questionnaire consisted of several sections. Section 1 aimed to obtain the respondents' demographic information (i.e., gender, age, current occupation, household income, and highest education level) with a mixture of binary and multiple-choice answers. In addition, the respondent's experience in relation to COVID-19 was also asked, which included the status of comorbidity, contracting COVID-19, and experience as a close contact.
Meanwhile, in Section 2, a 10-point Likert scale was used to assess respondents' risk perception of COVID-19. The questions consist of statements that refer to their level of worry toward the current pandemic (Answer scale; 1 = Very Not Worried, 10 = Very Worried), their perceived risk of getting infected (Answer scale; 1 = Very Low, 10 = Very High), whether the pandemic can be controlled and their belief that self-control can avoid COVID-19 infection (Answer scale; 1 = Strongly Disagree, 10 = Strongly Agree).
Section 3 of the questionnaire was to obtain trip frequency information. Respondents were asked about their travel frequency for the following purposes: work; purchasing daily necessities; meeting family members; meeting friends; and leisure activities. For the first survey, the respondents were asked to answer the question, "My frequency of travelling for …… DURING the pandemic is.." Meanwhile, the question for the second survey was altered to "My frequency of travelling for …… AFTER interstate travel was allowed is…". A 10-point Likert Scale was used for this section (1 = Very Rare, 10 = Very Frequent).

Sampling
The snowball sampling technique was used to obtain as many respondents as possible. The public was provided with the questionnaire links and encouraged to distribute the link to their contacts. The target group of the study included people over 18 years old and residing in the country at the time of answering. After excluding several respondents under 18 years old, 2,129 respondents were obtained in the first survey and 2,217 respondents in the second survey. According to Sekaran & Bougie (2016), for a population with a minimum of 1 000 000 people, the minimum sample needed is 384 people. Therefore, the number of respondents obtained in this study was sufficient.

Statistical analyses
The descriptive analysis for the socio-demographic factors of the respondents and a Kolmogorov-Smirnov normality test was first conducted. The normality result indicated a significant violation of the normality assumption (p < 0.05). Due to the non-normality of data, the Mann-Whitney test was performed to determine whether there was a statistically significant difference between the data of the first and second surveys by focusing on age, gender, occupation, income, and highest education level.
Next, the correlation between travel behaviour and the sociodemographic including age, gender (1 = male, 2 = female), occupation (1 = essential, 2 = non-essential), income level, education level (1 = lowest level, 5 = highest level), an experience of COVID-19 and risk perception were examined for both groups of respondents based on the value of Spearman's correlation coefficient. The variables significantly correlated in Spearman's correlation test were then chosen as the independent variables in Ordinal Logistic Regression (OLR) analyses.
Ordinal Logistic Regression (OLR) is a regression model wherein a dependent variable has an ordinal response (Harrell, 2015). OLR were analysed for both groups of respondents by considering the five dependent variables of trip frequencies, including i) travelling to work, ii) purchasing daily necessities; iii) meeting family members; iv) meeting friends; and v) leisure activities.

Socio-demographic characteristics for first and second surveys
In this study, the first survey was conducted during the early stage of recovery, and the second was conducted after the interstate travel was allowed. Table 1 shows the socio-demographics of the respondents in both surveys. The average age of respondents in the first survey was 38 years old. Most respondents were female (67%) and from non-essential sectors (69%). Non-essential employment refers to jobs not considered critical during the epidemic. 49% of them earned less than RM 4,850. Most respondents (83%) had a university education, followed by secondary education (17%) and others.
Meanwhile, the average age of the respondents in the second survey was 39 years old, most of whom were female (65%). Similar in the first survey, 69% of respondents worked in non-essential industries and had an annual income of less than RM 4,850 (50%). University-educated respondents dominated the majority of responses (80%), followed by secondary education (20%) and primary education (1%). The Mann-Whitney U test was performed to ensure the respondents' criteria for both surveys could be compared for further analysis. The P-value in Table 1 shows that only education level significantly differs among the respondents, while the rest of the demography exhibited minor variations. The findings indicate that the characteristics of the respondents for both periods were equivalent, allowing for a comparative study between groups.

Health status
In terms of the health status during the pandemic, the findings from the first survey show that most of the respondents (79%) did not have comorbidity illness (i.e., heart disease, high blood pressure, diabetes, or obesity). Most of them (97%) had never contracted COVID-19. In addition, most respondents (67%) did not have a family member or close contact that contracted COVID-19. In the second survey, most respondents also reported not having comorbidity illness (80%). However, more respondents contracted COVID-19 (12%) in the second survey compared to only 3% during the first period. Moreover, most respondents reported that their family members or close contacts had contracted COVID-19 (56%) compared to only 33% reported in the first survey.
The statistical results from the Mann Whitney-U test presented in Table 2 show a significant difference in the numbers of respondents, family members, and close contacts infected by COVID-19 for both groups. Meanwhile, no significant difference was reported in the status of comorbidity. The findings show a significant increase in COVID-19 infection from the first to the second.

Risk perception toward COVID-19
The respondents' mean risk perception scores for COVID-19 for the two surveys are depicted in the bar chart in Fig. 2. In the first survey, most respondents reported a high level of worry (M = 8.88, SD = 1.62) compared to the second survey (M = 8.13, SD = 2.18). However, most of the respondents in the first survey claimed that they could control themselves from being infected (M = 8.83, SD = 1.65) compared to the respondents in the second survey (M = 6.85 SD = 2.57). The risk perception of getting infected by the virus was not significantly different in the first survey (M = 6.55, SD = 2.50) and the second survey (M = 6.54, SD = 2.52). Meanwhile, the respondents in the first survey perceived that the infection could be controlled (M = 7.06, SD = 2.54) compared to the second survey (M = 6.85, SD = 2.57).
The increasing number of infections also contributed to the perception that the COVID-19 infection is hard to control with ease of movement. Even though the restriction remained in place for an extended period, infection cases continued to rise. Thus, it may be caused a decrease in the overall level of worry, and the respondents' perceptions of their abilities to control the pandemic were also reduced. However, the respondents from both periods perceived equivalent risk levels of getting COVID-19 infection. Fig. 3 depicts the respondents' travel behaviour during the first and second surveys in terms of trip frequencies for five travel activities, including work travel, travel for goods and daily necessities, visiting family, visiting friends, and leisure activities. Likert Scale was used for this section (1 = Very rare; 10 = Very frequent).

Travel behaviour during the pandemic
The findings show that all travel behaviours were less than the average mean of 5 points during the first survey. It indicates that most of the travel activities during the restriction movement were focused on travelling to work (Mean = 4.5, SD = 3.13) and getting goods and daily necessities (Mean = 4.07, SD = 2.24). The findings from the second survey show increasing frequency in travel activities for all purposes, especially travel to work (Mean = 5.53, SD = 3.09), getting good and daily necessities (Mean = 5.35, SD = 2.35), and visiting families (Mean = 4.72, SD = 2.52). The results indicate that people wanted to resume everyday activities once the restriction movement was eased. After almost two years of staying at home, people craved human interaction; hence when the movement ban was lifted, people went out for leisure activities, meeting families and friends. Though most companies in the country have shifted to online working during the pandemic, having employees under one roof was still the way to ensure productivity amongst the workers. Therefore, most companies demand to have their employees return to the office.

Correlation between trip frequencies, socio-demographic, experience with COVID-19 and risk perception
In the first survey, work trips had a positive correlation with gender and perception of getting infected but had a strong negative correlation with occupation. The findings also show a negative correlation between travel to work with self-experienced and family members getting COVID-19 (Refer to Table 3). The results indicate that male respondents were frequently involved in more work trips, especially for the essential sectors. In addition, the respondents who were never infected by COVID-19 and had no close contacts infected by COVID-19 reported involving more in travel to work. However, these respondents also reported a high perception of virus infection.
Regarding travelling for daily errands, a negative correlation was found between age, occupation, level of worry, and perceived selfcontrol. Meanwhile, gender and perceived risk of getting infected show a weak but significant positive correlation. The results indicate that young males involved in essential sectors tended to travel for daily errands. In addition, they also have a low level of worry and self-control toward the infection even though they have a high perception of getting infected.
For travelling to visit family members, results show a negative correlation with age, household income, level of worry and belief that selfcontrol can avoid infection. Besides that, travel behaviour was also positively correlated with gender, comorbidity, and perceived risk of infection. It can be concluded that young males who worked in the essential sector with lower income levels tended to visit their family members more often during this period. In addition, this group of respondents tended not to have any comorbidity and were less worried about the infection. This situation led them to believe that self-control could not avoid COVID-19 infection. However, they had a high perception of getting infected with the virus.
Significant negative correlations were found between travelling to meet friends and age, household income, level of worry, and belief that self-control can avoid COVID-19 infection. In addition, a positive correlation was discovered between gender and perceived risk of getting infected with said travel purpose. It illustrates that young males with lower incomes travelled more to meet their friends. These respondents had a lower worry about COVID-19 and less belief that self-control can avoid COVID-19. They also perceived high risk of getting infected with the virus.
Meanwhile, travelling for leisure activities was positively correlated with gender and perceived risk of getting infected, but negatively correlated with age, level of worry and self-control. It shows that young males who favour travel for leisure tended to worry less about COVID-19 infection. However, they also perceived a high possibility of getting infected and less belief that self-control can avoid COVID-19.
In terms of risk perception, Table 4 depicts that worry had negative correlations with gender, household income, education level and  (44) experiencing any close contact get an infection. The results indicate that males with less income and less educated tend to worry more, particularly when they have never had family members or friends infected with COVID-19. Meanwhile, those with essential sectors and higher education levels perceived a greater risk of infection, particularly in the absence of comorbidities and exposure to infected family and peers. The results also show that older males with higher incomes believed COVID-19 could be controlled. Likewise, older males, less educated and had never had close contact with an infected person, believed that selfcontrol could prevent COVID-19.
In the second survey, Table 5 shows that work travel was positively correlated with gender, education level, perceived risk of getting infected and belief that COVID-19 is a pandemic that can be controlled. Trip frequency was also negatively correlated with occupation, comorbidity, and level of worry. The results implied that males working in essential sectors travelled more to work than females. Meanwhile, those with a higher level of education and suffering from comorbidity tended to have a lower level of worry about the pandemic. At the same time, they also perceived higher risk of getting infected and believed that the pandemic could be controlled.
Next, travel to purchase daily goods showed a positive correlation with gender and a negative correlation with age, occupation, level of worry, and self-control. It indicates that young males working in essential sectors still travel for daily errands, similar to the first survey's condition. They were also less worried about the pandemic and believed self-control could avoid COVID-19 infection.
Regarding trips to meet family, the result shows that it positively correlated with gender, education and belief that COVID-19 can be controlled. A negative correlation was found between age, level of worry and self-control. The findings demonstrated that younger males with a higher level of education travelled more to meet their families during the second survey. They showed a lower worry about the pandemic and believed it could be controlled. However, they also believed that selfcontrol could not necessarily avoid COVID-19 infection.
Meanwhile, only gender and belief that COVID-19 can be controlled for travelling to meet friends showed positive correlations with the purpose. Age, occupation, the experience of COVID-19, level of worry and self-control were negatively correlated with travel to meet friends. The findings indicate that young males with a higher level of education were likely to travel to meet their friends. These respondents were less worried about the pandemic, as they had a higher perception that it could be controlled. However, they are less likely to believe that selfcontrol can avoid infection.
Lastly, trips for leisure activities were found to positively correlate with gender, education, comorbidity and belief that COVID-19 can be controlled. On the other hand, a negative correlation was found for age, occupation, family members, level of worry and self-control. Younger males with high-level education working in the essential sectors tended to travel for leisure activities. In addition, respondents who did not have comorbidity and had no close contact with those infected by COVID-19 were likely to travel for leisure activities. In addition, those who perceived that the pandemic could be controlled and had a lower belief that self-control can avoid COVID-19 infection seemed to be involved in travelling for leisure activities as well. Table 6 shows an in-depth correlation between socio-demographics, experience, and risk perception during the pandemic for the second survey. The survey was performed in conjunction with permission for interstate travel. The level of worry about the pandemic significantly correlates with the experience of contracting an infection. The finding indicates that after interstate travel was allowed, those still uninfected had a higher level of anxiety. Meanwhile, the high-risk perception of getting infected still occurs among essential workers, especially those with high-level education and no comorbidity, who never get infected themselves and their close contacts. In addition, an older male with a higher level of education who worked in essential sectors and had direct contact with infected COVID-19 patients believed that the contagious      could be contained. However, those who work in essential sectors are more likely to perceive that self-control can avoid infections.

Factors affecting travelling to work
The results from OLR analysis for trip frequency to work during the first and second surveys are presented in Table 7. The independent variables for the analysis were based on the significant correlation. The finding shows that gender, occupation, close contacts, and risk of getting infected were significant predictors during the first survey. In gender, the significant predictor is female (B = -0.226, s.e. = 0.083, p = 0.007), with the male being the reference sub-category. The odds ratio indicates that a trip frequency for female travel to work decreases by a factor of Exp(B) = 0.798 compared to the odds of those who are male. The significant predictor under occupation is the essential sector (B = 1.160, s. e. = 0.090, p < 0.001), with non-essential as the reference sub-category. The odds ratio indicates that the odds of a person travelling to work in essential sectors increase by Exp(B) = 3.191 compared to those working in non-essential sectors. Next, the significant predictor for close contacts is those who do not have close contact with COVID-19 patients (B = -0.236, s.e. = 0.083, p = 0.004), with those having close contacts as the reference sub-category. The odds ratio indicates that the odds of a person travelling to work when they have no close contacts decreases by a factor of Exp(B) = 0.790 compared to those who have close contact with COVID-19 patients. Lastly, regarding the risk of getting infected, results show that it was a significant predictor of travel for work (B = 0.112, s. e. = 0.016, p < 0.001). The odds ratio shows that the odds of a person travelling to work increase by Exp(B) = 1.118 when they perceive a higher risk of getting infected.
Unlike the first survey, education level was added as the significant predictor for this purpose. For gender, the significant predictor is female (B = -0.266, s.e. = 0.079, p = 0.001), with the male being the reference sub-category. The odds ratio indicates that a person was travelling to work when female decreases by a factor of Exp(B) = 0.766 compared to the odds of those who are male. The significant predictor under occupation is the essential sector (B = 0.502, s.e. = 0.083, p < 0.001), with non-essential as the reference sub-category. The odds ratio indicates that the odds of a person travelling to work in essential sectors increase by Exp(B) = 1.653 compared to those working in non-essential sectors. The sub-category which is a significant predictor of education level is secondary level (B = -0.219, s.e. = 0.095, p = 0.021), with tertiary level as the reference sub-category. The odds ratio indicates that the odds of a person travelling for work when they have completed secondary education decreases by a factor of Exp(B) = 0.803 compared to those who have tertiary education. Next, the significant predictor under the comorbidity category is those who did not have a comorbidity (B = -0.195, s.e. = 0.093, p = 0.036), with those who have comorbidity as the reference sub-category. The odds ratio indicates that the odds of a person travelling to work during the second survey when they have no comorbidity decreases by a factor of Exp(B) = 0.823 compared to the odds of those who have comorbidity. The level of worry is also a significant predictor for this purpose (B = -0.070, s.e. = 0.018, p < 0.001). The odds ratio shows that the odds of a person travelling to work decreases by a factor of Exp(B) = 0.932 when there is an increased worry about COVID-19. Lastly, the risk of getting infected by COVID-19 is a significant predictor for travelling to work (B = 0.058, s.e. = 0.016, p < 0.001). The odds ratio shows that the odds of a person travelling to work increase by a factor of Exp(B) = 1.060 when there is an increased perceived risk of getting infected by the virus.

Factors affecting travelling for daily necessities
The parameter estimates for trip frequency for purchasing daily necessities for the first and second survey is shown in Table 8. For the first survey, gender, age, occupation, level of worry, risk of getting infected, and self-control are significant predictors in travelling to work. In gender, the significant predictor is female (B = -0.400, s.e. = 0.083, p < 0.001), with the male being the reference sub-category. The odds ratio indicates that a person was travelling to purchase daily necessities when female decreases by a factor of Exp(B) = 0.671 compared to the odds of those who are male. Age was also a significant predictor (B = -0.010, s.e. = 0.004, p = 0.003), with the odds ratio showing that the odds of someone travelling to purchase goods decreases by a factor of Exp(B) = 0.990 when they are older. The significant predictor under occupation is the essential sector (B = 0.370, s.e. = 0.085, p < 0.001), with nonessential as the reference sub-category. The odds ratio indicates that the odds of a person travelling to purchase daily necessities when working in essential sectors increases by a factor of Exp(B) = 1.448 compared to those who are working in non-essential sectors. In addition, level of worry was the significant predictor (B = -0.105, s.e. = 0.026, p < 0.001). The odds ratio shows that the odds of a person travelling to buy necessities decrease by a factor of Exp(B) = 0.900 when worrying about the pandemic increases. The next significant predictor was the risk of getting infected by COVID-19 (B = 0.090, s.e. = 0.016, p < 0.001). The odds ratio indicates that the odds of a person travelling to purchase goods increase by a factor of Exp(B) = 1.094 when there is an increase in the perceived risk of getting infected by the virus. Lastly, in terms of the belief that self-control can avoid COVID-19 infection, results show that it *p-value is significant at p < 0.05. ** Reference category within socio-demographic characteristics and COVID-19 characteristics.
was a significant predictor to travel for purchasing daily necessities (B = -0.073, s.e. = 0.024, p = 0.003). The odds ratio shows that the odds of a person travelling to purchase daily necessities decrease by a factor of Exp(B) = 0.930 when there is an increase in the belief that self-control can avoid COVID-19 infection. For this purpose, only occupation was not a significant predictor in the second survey. For gender, the significant predictor is female (B = -0.340, s.e. = 0.080, p < 0.001), with the male being the reference subcategory. The odds ratio indicates that females travel to work when it decreases by a factor of Exp(B) = 0.712 compared to the odds of those who are male. The next significant predictor was age (B = -0.007, s.e. = 0.003, p = 0.050), with the odds ratio showing that the odds of someone travelling to purchase goods decreases by a factor of Exp(B) = 0.994 when they are older. In addition, the level of worry was the significant predictor (B = -0.064, s.e. = 0.018, p < 0.001). The odds ratio shows that the odds of a person travelling to purchase daily necessities decrease by a factor of Exp(B) = 0.938 when worrying about the pandemic increases. Finally, in terms of the belief that self-control can avoid COVID-19 infection, results show that it was a significant predictor to travel for purchasing daily necessities (B = -0.073, s.e. = 0.021, p = 0.001). The odds ratio shows that the odds of a person travelling to purchase daily necessities decrease by a factor of Exp(B) = 0.929 when there is an increase in the belief that self-control can avoid COVID-19 infection.

Factors affecting travelling to visit family
The parameter estimates for trip frequency to meet family members for the first and second survey is shown in Table 9. Gender, age, level of worry and risk of getting infected are significant predictors. In gender, the significant predictor is female (B = -0.354, s.e. = 0.085, p < 0.001), with the male being the reference sub-category. The odds ratio indicates that females travelling to meet family members decreases by a factor of Exp(B) = 0.702 compared to the odds of those who are male. Age was also a significant predictor (B = -0.012, s.e. = 0.004, p = 0.004), with the odds ratio showing that the odds of someone travelling to purchase goods decreases by a factor of Exp(B) = 0.988 when they are older. Level of worry was also one of the significant predictors (B = -0.109, s.e. = 0.026, p < 0.001). The odds ratio shows that the odds of a person travelling to purchase goods decrease by a factor of Exp(B) = 0.896 when worrying about the pandemic increases. The next significant predictor was the risk of getting infected by COVID-19 (B = 0.040, s.e. = 0.017, p = 0.019). The odds ratio indicates that the odds of a person travelling to meet family members increase by a factor of Exp(B) = 1.041 when there is an increase in the perceived risk of getting infected by the virus.
The parameter estimates for trip frequency to meet family members for the second survey show gender, level of worry, risk of getting infected, and self-control as significant predictors. Firstly, the significant predictor is female (B = -0.366, s.e. = 0.079, p < 0.001), with the male being the reference sub-category. The odds ratio indicates that females travelling to meet their families decreases by a factor of Exp(B) = 0.693 compared to the odds of those who are male. Furthermore, the level of worry was the significant predictor (B = -0.067, s.e. = 0.018, p < 0.001). The odds ratio shows that the odds of a person travelling to meet family members decrease by a factor of Exp(B) = 0.935 when worrying about the pandemic increases. The next significant predictor was the risk of getting infected by COVID-19 (B = 0.042, s.e. = 0.015, p = 0.006). The odds ratio indicates that the odds of a person travelling to meet family * p-value is significant at p < 0.05. ** Reference category within socio-demographic characteristics and COVID-19 characteristics.

Table 9
Regression coefficients and odds ratio for travelling to meet family members. members increase by a factor of Exp(B) = 1.043 when there is an increase in the perceived risk of getting infected by the virus. Lastly, in terms of the belief that self-control can avoid COVID-19 infection, results show that it was a significant predictor to travel to meet family members (B = -0.046, s.e. = 0.021, p = 0.031). The odds ratio shows that a person travelling to meet their family decreases by a factor of Exp (B) = 0.955 when there is an increase in the belief that self-control can avoid COVID-19 infection.

Factors affecting travelling to visit friends
The parameter estimates for trip frequency to meet friends for the first and second survey is shown in Table 10. The significant predictors for this purpose during the first survey are gender, age, level of worry, risk of getting infected, and self-control. The sub-category which is significant is female (B = -0.613, s.e. = 0.089, p < 0.001), with the male being the reference sub-category. The odds ratio indicates that females travelling to meet friends decreases by a factor of Exp(B) = 0.542 compared to males. Age was also a significant predictor (B = -0.018, s.e. = 0.004, p < 0.001); travelling to meet friends decreases by a factor of Exp(B) = 0.983 when they are older. Next, the level of worry was the significant predictor (B = -0.138, s.e. = 0.028, p < 0.001). The odds ratio shows that the odds of a person travelling to meet their friends decrease by a factor of Exp(B) = 0.871 when worrying about the pandemic increases. The next significant predictor was the risk of getting infected by COVID-19 (B = 0.082, s.e. = 0.018, p < 0.001). The odds ratio indicates that a person travelling to meet their friends increases by a factor of Exp (B) = 1.1.086 when there is an increase in the perceived risk of getting infected by the virus. Finally, in terms of the belief that self-control can avoid COVID-19 infection, results show that it was a significant predictor to travel to meet friends (B = -0.084, s.e. = 0.024, p = 0.001). The odds ratio shows that the odds of a person travelling to meet friends decrease by a factor of Exp(B) = 0.919 when there is an increase in the belief that self-control can avoid COVID-19 infection. Table 11.
For the second survey, gender reports as the significant predictor, with females (B = -0.459, s.e. = 0.081, p < 0.001) and males being the reference sub-category. The odds ratio indicates that females were travelling to meet friends decreases by a factor of Exp(B) = 0.632 compared to males. Age was also a significant predictor (B = -0.009, s.e. = 0.003, p = 0.005), with the odds ratio indicating that someone travelling to meet friends decreases by a factor of Exp(B) = 0.991 when there is an increase in age. In addition, the level of worry was the significant predictor (B = -0.079, s.e. = 0.018, p < 0.001). The odds ratio shows that the odds of a person travelling to meet friends decrease by a factor of Exp(B) = 0.924 when worrying about the pandemic increases. The next significant predictor was the risk of getting infected by COVID-19 (B = 0.069, s.e. = 0.015, p < 0.001). The odds ratio indicates that the odds of a person travelling to meet their friends increase by a factor of Exp(B) = 1.071 when there is an increase in the perceived risk of getting infected by the virus. Lastly, in terms of the belief that self-control can avoid COVID-19 infection, results show that it was a significant predictor to travel to meet friends (B = -0.058, s.e. = 0.022, p = 0.007). The odds ratio shows that a person travelling to meet their friends decreases by a factor of Exp(B) = 0.944 when there is an increased belief that selfcontrol can avoid COVID-19 infection.

Factors affecting travelling for leisure activities.
The parameter estimates for trip frequency for leisure activities for the first survey are shown in Table 9. For this purpose, all measured categories were significant predictors. Firstly, the significant predictor is female (B = -0.443, s.e. = 0.088, p < 0.001), with the male being the reference sub-category. The odds ratio indicates that females travelling for leisure activities decrease by a factor of Exp(B) = 0.642 compared to the odds of those who are males. Next, age was also a significant predictor (B = -0.015, s.e. = 0.004, p < 0.001), with the odds ratio showing that the odds of someone travelling to meet their friends decreases by a factor of Exp(B) = 0.985 when they are older. Level of worry was also the significant predictor (B = -0.135, s.e. = 0.027, p < 0.001). The odds ratio shows that the odds of a person travelling for leisure activities decrease by Exp(B) = 0.874 when worrying about the pandemic increases. In addition, the risk of getting infected by COVID-19 (B = 0.050, s.e. = 0.018, p = 0.004) is a significant predictor. The odds ratio indicates that the odds of a person travelling for leisure activities increase by a factor of Exp(B) = 1.052 when there is an increase in the perceived risk of getting infected by the virus. Lastly, in terms of the belief that selfcontrol can avoid COVID-19 infection, results show that it was a significant predictor to travel for leisure activities (B = -0.056, s.e. = 0.025, p = 0.026). The odds ratio shows that the odds of a person travelling to meet friends decrease by a factor of Exp(B) = 0.945 when there is an increase in the belief that self-control can avoid COVID-19 infection.
Similar to travelling for work, education level was also one of the new significant predictors compared to other travel purposes. In gender, the significant predictor is female (B = -0.317, s.e. = 0.080, p < 0.001), with the male being the reference sub-category. The odds ratio indicates that travelling for leisure activities for females decreases by a factor of Exp(B) = 0.728 compared to the odds of those who are male. The subcategory which is a significant predictor of education level is secondary level (B = -0.216, s.e. = 0.097, p = 0.026), with tertiary level as the reference sub-category. The odds ratio indicates that the odds of a person travelling for leisure activities when they have completed secondary education decreases by a factor of Exp(B) = 0.806 compared to the odds of those who have tertiary education. Having no close contact was another significant predictor (B = -0.175, s.e. = 0.077, p = 0.022), with having close contact as the reference sub-category. The odds ratio indicates that the odds of a person travelling to carry out leisure activities when they have no close contact with COVID-19 patients decreases by a factor of Exp(B) = 0.839 compared to the odds of those who do have close contact. Furthermore, the level of worry was the significant predictor (B = -0.101, s.e. = 0.018, p < 0.001). The odds ratio shows that the odds of a person travelling for leisure activities decrease by Exp(B) = 0.904 when worrying about the pandemic increases. Finally, in terms of the belief that self-control can avoid COVID-19 infection, results show that it was a significant predictor to travel for leisure activities (B = -0.064, s.e. = 0.021, p = 0.003). The odds ratio shows that the odds of a person travelling to carry out leisure activities decrease by a factor of Exp(B) = 0.938 when there is an increase in the belief that self-control can avoid COVID-19 infection.

Discussion
Surveys were conducted with a focus on two different groups of participants for two separate periods of the COVID-19 pandemic in Malaysia. The Mann-Whitney U statistical analyses show that there were no significant differences in socio-demographics for both groups except in the level of education. Thus, the comparative study is acceptable for this study. Both surveys successfully acquired a high number of respondents. Previously, several comparative studies are concentrating on the COVID-19 pandemic, but the majority do not compare two distinct stages of COVID-19 in the same country (Alanezi et al., 2021;Hoffman et al., 2021;Islm et al., 2021;Novak et al., 2021;van den Berg et al., 2023;Vohra et al., 2022;Wang & Mao, 2021). There is a relationship between human mobility and the transmission of contagious diseases since respondents who reported close contact with COVID-19 patients increased during the second survey period. It indicates that as the movement grew, so did the likelihood of transmitting viruses. This outcome is consistent with earlier findings. (Li et al., 2020;Zhu et al., 2020;Nouvellet et al., 2021).
Regarding risk perception towards COVID-19, respondents to the first survey in 2021 have a generally high-risk perception. During these periods, Indonesia, Pakistan, and Turkey likewise indicated a high-risk perception of the transmission of the virus (Gul, 2021;Tejamaya et al., 2021). However, the global immunisation programme decreased the risk perception of contracting the disease. As a result, the worldwide communities were more cognizant of the means to control the transmission of viruses. In addition, the impact of the infectious was not as severe as it had been in the early stages of the pandemic. Several studies also report this condition (Kirchberger et al., 2021;Ma et al., 2020;Yang et al., 2020).
Similarly, Malaysia's National Immunisation Programme (NIP) goal is to achieve herd immunity. Consequently, it provides Malaysians with a sense of safety once they have been immunised. This study's findings also prove a change in travel behaviour after the National Immunisation Programme (NIP). The restriction movement was lifted after 70% of the population was vaccinated. After implementing the vaccination campaign, the number of COVID-19-related deaths in Malaysia dropped, but the number of infections continued to rise.
This study's frequency of travelling for all trip purposes showed a significant increase once travel was allowed. With the ban lifted, people were now able to travel much more accessible than during the early period of the pandemic. The finding of this study is similar to previous studies (Kolarova et al., 2021;Molloy et al., 2020). However, the results also show the behaviour changes for the second survey. People tend to keep in less frequently of travelling to work and socialising. It might be because people were now more comfortable shifting their activities to online platforms, especially for work and meetings. Meanwhile, delivery services have become significantly necessary to get daily necessities. The result parallels several previous studies (Ding et al., 2022;Novianto et al., 2022;Semple et al., 2021).
The correlation for travel behaviour in the first and second surveys reveals that younger males were involved in more trips related to all purposes than females. It could be because, in some Southeast Asian countries, males were seen as the breadwinner and the head of the family. Therefore, for travel purposes involving work or purchasing daily necessities, this group is more prominent. People from the younger age group were also seen as more resilient in health since they are more energetic and healthier than older groups; hence, they tended to travel more during this pandemic, similarly explained in previous studies (Anwari et al., 2021;Irawan et al., 2020;Parady et al., 2020). For social activities such as meeting family members and friends, or leisure activities, both survey periods show that those with a lower worry tend to travel more for said purposes. Less worried people probably think they are not affected by a risk or, in this case, a virus, making them perceive that they are safer than everyone else. In return, it caused them to be more confident in going out, performing outdoor activities and socialising.
The regression analysis found that for travelling to work, someone who works in the essential sector tends to travel more for such purpose compared to those in the non-essential sector during both surveys, similar to the result found by (Politis et al., 2021). Those working in essential sectors, such as health, security, or water treatment, were allowed to travel with local authorities' permission. Therefore, most had little chance to switch to online working, which could explain the result. For purchasing goods or daily necessities, an increase in the belief that self-control can avoid COVID-19 infection shows the decreased odds of a person going out for the said purpose, similarly found in previous studies (Earley & Newman, 2021;Liu et al., 2021). They believe that self-control behaviour, such as not going out and being in contact with or meeting other people to work against COVID-19, causes them to not engage in such activity. In addition, the local authorities have advised people to avoid being in a public place for a long period, something that one would do if they went to the supermarket or grocery store.

Conclusion
This study explored the effects of COVID-19 on risk perception and travel behaviour changes in two groups of respondents at two different periods during the pandemic. A questionnaire survey was distributed online to collect data about demographics, health status, precaution behaviour, risk perception and travel behaviour changes. The statistical analyses showed no significant differences in the demographics of both groups except for education level. This result indicates that the data from two different groups of respondents are sufficient to be compared in this study.
The findings show that there were indeed changes in risk perception and travel behaviours between both periods. The risk perception of the pandemic decreased in parallel with the ease of movement, which that maybe because of the National Immunisation Program (NIP). Although there were new variants during the second phase of data collection, the respondents' risk perception of the infection of COVID-19 was significantly reduced. The less perceived towards infection also leads to a decrease in practising precautionary behaviour. The risk perception also, coupled with the increase in the frequency of travelling, led to a significant increase in COVID-19 cases. However, fewer mortality cases were reported during the second phase of the data collection.
Based on the results of this study, policy implications regarding the Standard Operating Procedure (SOP) during an emergency event related to a pandemic can be improved. The risk perception and travel behaviour changes could also give authorities more information regarding peoples' mobility and better equip them to predict people's demands on public facilities (public transport or open spaces) during a crisis such as the pandemic. In conclusion, the high-risk perception will reduce if people understand the causes of contagion and acknowledge how to avoid the infection. However, this study also had some limitations. First, the sample did not focus on the same respondents because of the lack of experience predicting the situation. Therefore, further studies on the topic that relates to the pandemic should consider the time series analyses by targeting the same respondents for the different survey periods to understand the short-term and long-term effects of the pandemic.
This study is based on a questionnaire survey conducted nationwide in Malaysia. The findings of this study might apply to other countries, especially developing countries. It is interesting to compare the effect of COVID-19 on risk perception and travel behaviour changes in other societies. In addition, the understanding of the factors that affect risk perception and travel behaviour during the pandemic attack should be explored in future studies.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.