Variations in Mode Choice of Residents Prior and during COVID-19: An Empirical Evidence from Johannesburg, South Africa

Abstract

There have been numerous studies on the impact of COVID-19 on mobility in most developed countries; however, few of the studies have focused on the impact of the pandemic in developing countries, especially in Africa. In view of this, our study examined the impact of the pandemic on residents’ transportation mode choice in South Africa. This study adopted the use of both primary and secondary data obtained from TomTom statistics and an online survey of respondents’ mobility patterns before and during the pandemic. The questionnaire was administered through emails, and respondents were asked to provide information about their socio-economic characteristics, travel characteristics (before and during COVID-19), and the effect of COVID-19 on their travel patterns. A multinomial logistic model was adopted for analysis, and the findings revealed that variations existed in trip frequency, trip purpose, and mode choice of people before and during the pandemic. It was also discovered that respondents shifted from the use of public transport to private cars during the pandemic as a result of the implications for their health. Based on this, we propose that an enabling environment and an efficient transport planning technique should be adopted by the government and relevant stakeholders in the transport sector. This will integrate all modes of transport to reduce the over-reliance on private automobiles and also to encourage the use of non-motorized transport (walk/cycle) for sustainable transport planning in the future.

1. Introduction

It has been noted from previous studies [1,2,3,4,5,6,7,8] that interactions between people and spaces contribute to the spread of infectious diseases, especially during pandemics. Based on this, travel is restricted during pandemics [9,10,11]. In order to control the spread of the virus, various control and preventive measures are imposed by the governments of different countries depending upon the local governance, socio-economic conditions, and cultural context.

South Africa has been the worst-hit country in Africa. The number of confirmed cases of COVID-19 on a single day in South Africa amounted to 8078. The total cases reached 3,684,319, which is the highest number of confirmed cases compared to other African countries. Gauteng, the province with Johannesburg as its capital, was the most affected regionally with over 1.2 million cases as of early March 2022 [12].

The COVID-19 pandemic has had a profound impact on all sectors of the economy globally, including the South African economy. This suffered due to government restrictions on movement, which hampered its economic growth and development. Transport is one of the sectors that has been severely affected by both demand- and supply-side impacts. The COVID-19 pandemic has resulted in unparalleled restrictions on travel and on social gatherings in many countries [13].

The different measures put in place globally to reduce the effect of the pandemic included national lockdowns, the closure of borders, the cancellation of events, and the shifting of work towards home offices. All this has had major implications for people’s daily work, as well as travel mobility; for example, studies analysing daily mobility patterns via phone data show a general reduction in global mobility compared to pre-COVID-19 times [14,15,16,17].

South Africa has suffered several setbacks since the outbreak in 2019. The country recorded its first case on 5 March 2020, and there has subsequently been a surge in the number of affected people. Recently, the country experienced a third wave of the pandemic with daily deaths recorded in Gauteng province. In light of this, the government adopted measures aimed at curbing or minimizing the effect of the virus on the population. One of these was travel restrictions on the movement of people and even goods and services within and outside the provinces. A study by Venter, Hayes, and Gyl [18] showed that prior to the pandemic, 60% of the population used public transport, but post the pandemic, public transport use currently remains substantially below pre-COVID-19 levels. For instance, at the end of August 2022, Google’s Mobility Report indicated public transport use to be 42% below the baseline which reveals a drop in traffic flow along major routes in the city.

The drop in traffic, even though it has benefited the people, has also shown a reduction in the economic activities of the city. This is seen in the closure of major commercial outlets which has given rise to increased unemployment and crime, to mention just a few, as some people lost their source of livelihood during the pandemic.

Shifts have also occurred in people’s work, shopping, and social activities, which in turn have led to a decline in travel volumes and traffic patterns. This has been brought about by changes such as work from home (WFH), staggered working hours, and retrenchments which have reduced the number of trips [19]. Studies [20,21,22] have addressed the impact of COVID-19 on transport and the environment, with a major focus on the transportation of workers, the aviation sector, and employment to mention just a few. Nevertheless, few studies have examined its impact on the mobility of people and, where such exist, they have not addressed its impact on the mode choice of respondents. Hence, this study examined the differences in mobility patterns of people in Johannesburg before and during the pandemic. It explored the travel characteristics of households in terms of trip frequency, trip purpose, and variations in mode choice of people. The main objectives explored the propensity for utilizing other modes of transport such as non-motorized travel (walking/cycling) as an alternative to private car use in future pandemics.

1.1. Related Work

1.1.1. General Overview of the Restrictions on Movement in Different Alert Categories and Their Impact on Mobility in South Africa

The restrictions were divided into Levels 1–5. The Lockdown Alert Level 5 was enforced from midnight on 26 March until 31 April 2020. Based on this, movement was severely curtailed, and individuals were not permitted to leave home except under strictly controlled circumstances. These included seeking medical care; buying food, medicine, and other supplies; or collecting a social grant.

Furthermore, all long-distance and inter-provincial public transport was stopped; public transport operations were prohibited except for transporting essential workers, and this was scheduled between 05:00 and 10:00 and 16:00 and 20:00 [23]. Vehicles were not permitted to carry more than 50% of their licensed seating capacity. These regulations under the Alert Level 5 impacted negatively on traffic flow as only few vehicles were seen on the road, and the few ones that were allowed had a maximum carrying capacity, with designated work hours.

The Alert Level 4 came into operation from 1 to 31 May 2020, and public transport services were permitted between 05:00 and 19:00, with a grace period until 20:00 to drop off passengers. Loading capacity remained at 70% for minibus taxis, and 50% for e-hailing and metered taxis. An increasing number of goods were deemed essential, and travel demand increased. On 1 June 2020, there was “the return of rush hour” [24] with Alert Level 3. Minibus taxis and buses were permitted to resume operations at all hours (still at 70% loading capacity), whereas e-hailing and metered taxis could not operate beyond 50% capacity. Train services suffered a decline or total loss in ridership, as their services were not in operation. On 12 July 2020, the decision was announced that minibus taxis could be fully loaded for short distances on the condition that risk-mitigation protocols related to masks, vehicle sanitizing, and open windows were followed. By Alert Level 2 (18 August to 20 September 2020), all road-based transport was permitted to operate at full capacity, and trains could carry a maximum of 70% capacity. The public transport systems such as metered taxis, minibuses and other paratransit received a form of recognition by the government during the pandemic. However, little or no recognition was given to active travel in the form of walking/cycling, thus implying their invisibility to decision-makers and stakeholders in the transport sector [25].

The lockdown affected the mobility of people across the country, as gatherings of people were prohibited and movement was only allowed in order to buy groceries, and for medical and essential purposes within the provinces. Nationwide, schools were closed and students were advised to go home. Offices and businesses were shut down as well, while security forces ensured that people complied with the ‘stay at home’ directive from the government to fight against the COVID-19 pandemic. Contrary to what had been, people’s mobility became restricted in order to reduce the transmission of the virus from one person to another.

1.1.2. COVID-19 and Transportation

Several studies have explored the impact of COVID-19 on transportation in both developed and the developing countries of the world.

For example, Vickerman [26] determined the use of public transport by residents in the UK during the pandemic. It was discovered that there was a reduction in public transport ridership despite the amount of financial support provided by the government. This was as a result of the prevailing model of a deregulated competitive public transport, as there are challenges in the current methods of delivery of public transport services. This cost implication was also corroborated by Dai et al. [27], as they explored the fare-free public transport policy implemented to lure passengers back to public transport in three Chinese cities of Hangzhou, Ningbo, and Xiamen. While passengers may be lured to public transport, Eisenmann [28] proposed that car transport became more important during the pandemic in Germany while public transport lost ground. This was also corroborated by Dai et al., [27] as they recognized a radical modal shift from the public to private transport mode in India.

In an attempt to establish the contextual differences or similarities in the COVID-19 impact on transport services in Africa, Porter et al. [29] illuminated women’s mobility in Africa as impacted by COVID-19. The study was conducted in Abuja, Tunis, and Cape Town, and findings revealed a greater need for the increased involvement of women in the transport sector both as commuters and transport operators, so as to reach regulatory positions that could limit the consequences of COVID-19 in Africa. Similarly, Mogaji [30] examined the consequences of COVID-19 for transportation in Lagos, the largest commercial city of Nigeria, and found that economic, religious, and social activities influenced transport in Lagos during the heat of the pandemic. Furthermore, Barbieri et al. [31] observed a significant reduction in the use of transport rules to meet passengers’ travel needs in their study on the impact of COVID-19 and the risk components associated with various transportation routes in 10 countries. The identified demographic factors influencing transport mode choice are age, income, gender, and education; others include transport specific factors such as cost of travel, purpose of trip, and time of travel. All these factors have been found to influence the transportation mode choices of people, thereby affecting their travel behaviour. The studies above have thus revealed that the pandemic had negative effects on major sectors of the economy and transportation. In view of this, our study examined the impact of COVID-19 on the mobility of people in South Africa; a developing country with the highest number of COVID-19 cases in Africa.

1.1.3. Active Travel in South Africa

Walking and cycling are seen as safe and low-carbon means of transport across the globe. In some developed countries, such as Europe and several US and Latin American cities, pop-up bicycle facilities have become increasingly common, but these were entirely absent in South Africa [25]. Instead, South Africa has directed its efforts towards attempting to regulate public transport capacities and proposing complex “peak-flattening” and costly facilities-cleansing measures. The country has largely failed in these attempts and has conceded in some cases to full-capacity paratransit. Furthermore, the authorities have paid scant attention to alternatives to public transport, such as ride-share, walking, and cycling. According to Jennings [32,33], the use of active travel such as walking and cycling was given more attention during the lockdown. However, it was rather seen as a form of exercise for the mobility-privileged in the society rather than a form or mode of travel. Even though there were pedestrian and cycling policies across the country, this was supposedly done to promote and support the mode as a means of a climate, health, and poverty alleviation strategy, as policymakers in the transport sector ignored the mode entirely.

Walking is a major mode of transport in South African cities (at least 30% of trips) and is common among people without private vehicles. Moreover, while bicycle travel is a minor mode, (at around 1% of trips), this is not because there is no latent demand but because bicycle promotion measures have fallen short. Strategies are insufficiently activated, and infrastructure development is frequently contested [34]. The majority of people who cycled for exercise and grocery shopping during the lockdown were seen as defaulters who only used it to violate the restrictions on movement. To promote non-motorized travel in the country, it was recommended by activists that bicycles must be provided for essential workers to enable accessibility to a safe, reliable, and convenient transport mode to reach townships. Furthermore, the government and private sector were to embrace and support non-motorized transport by providing cycling infrastructure, better urban design and planning, and incentivizing those who cycled to work [25].

2. Materials and Methods

The study adopted the use of both primary and secondary data. The secondary data was collected online from TomTom statistics which have been known to provide accurate real-time travel information. The traffic flow information accurately revealed the traffic situations in the city during the COVID-19 and pre-COVID-19 seasons. A monthly evaluation of traffic flow was provided, and the spatial analysis of traffic flow along different routes was also documented. A comparative study was thereafter conducted to access the traffic flow prior to the pandemic (December 2019–February 2020) and during the pandemic (March –September 2020). The primary data involved an online survey which was conducted through the administration of questionnaires. The survey was conducted from February–June 2021. Following COVID-19 standard operating procedures, it was difficult to collect a questionnaire survey with hard copies. Therefore, a questionnaire was prepared on Google forms and distributed randomly on various social media platforms (e.g., Facebook and WhatsApp) to collect responses consistent with previous studies [35,36,37,38].

Data were captured relating to the socio-demographics of the respondents, the characteristics of primary travel before and during the COVID-19 pandemic, and factors affecting mode choice for primary travel before and during the COVID-19 pandemic. The questionnaire designed for the research had three main sections. Section 1 recorded information relating to socio-economic characteristics such as gender, age, educational level, income, and marital status of respondents. Section 2 focused on the travel characteristics of the people such as trip frequency (average number of round trips), trip purpose (work, shopping, health, or recreation), travel time, transport mode, and travel cost prior to and during the pandemic. The Section 3 comprised information about the impact of the pandemic on the mobility of households, with questions relating to the frequency with which respondents went to work, shopping, and other activities before and during the pandemic.

A total of 456 questionnaires were retrieved and used for analysis. According to Lindemann [39], the acceptable response rate for an online survey is 33% and above, while 50% and above is considered acceptable for other means of surveys.

The sample size was determined using the Taro Yamane formula. This formula is applicable to a finite population, which implies that the population size is known, the Yamane formula for determining the sample size is given by:

$$n=\frac{N}{1+N{\left(e\right)}^2}$$

where

• n = Sample size,
• N = population size, and
• e = Margin of error (MoE), e = 0.05

A 2022 report by Statistics South Africa in the World Population Review [40] estimated the 2020 population of Johannesburg to 5,782,747. Based on this, we arrived at a sample size of 400. The calculation is provided below:

$$n=\frac{5,782,747}{1+5,782,747\times {\left(0.05\right)}^2}=\frac{5,782,747}{14,457.87}={399.9}^{ }$$

n = 400

Study Area

The City of Johannesburg is located in the Gauteng province and is the biggest metropolitan area in South Africa [41], covering an extended area of 1.645 square kilometres. According to SAIRR [42], South Africa’s large population lives in urban areas, an aftermath of urbanization over past years. The economic development in Johannesburg city has attracted people from all walks of life in search of better opportunities. The transport situation according to Sabest [43] has revealed that the two major cities with the most traffic challenges in South Africa are Johannesburg and Cape Town, and this was attributed to the economic activities in these towns.

3. Results

3.1. State of Traffic in Johannesburg

The City of Johannesburg (COJ) (see Figure 1) is the area in the republic with the most business activities and tourism concentrations; this makes traffic congestion very common on most of the roads. As seen in Figure 2, the congested roads are represented in the orange colour. Traffic is usually high during the peak periods (morning and evening) on the Main Reef and Sisulu roads located to the west of the map. These routes connect other nodes and most people going to work in the early hours of the morning (6:00–7:00 a.m.).

As seen in Figure 3, the month of February 2020 (pre-pandemic) had the highest traffic congestion level with 38%. Within the context of the daily fluctuations in traffic and travel speeds, the study considered the respective indicator of traffic congestion for each recorded instance. According to Hamad and Kikuchi [44], the value of congestion, on a scale of 0 to 1, is categorized as low, moderate, high, and very high. The distribution of traffic across all roads and their respective traffic congestion levels indicates that a better index of traffic congestion, level B, as compared to levels C & D, is maintained for capacity utilization under 40%. A steady index of traffic congestion, between 0.52 and 0.57, is categorized as moderate, and is maintained for varying levels, between 10% and 27%, of capacity utilization. The observed traffic congestion level of 38% on the road is considered by the authors to be due to the high amount of combination trucks on this roadway. Although only a limited volume of such vehicles occupy the roadway, their relatively lower speeds result in a state of apparent congestion even before the existing capacity is well utilized.

The traffic congestion dropped to 25% in March 2020 when the pandemic started in the country and restrictions were already put in place, thus disrupting movement in and out of the provinces. However, a sharp drop in the traffic flow was noticeable in the month of April after the government had put stricter measures in place which prevented unnecessary movement within provinces. Thus, people had to remain indoors, and only a few essential workers who could not work from home were allowed to move. The month of April 2020 (pandemic) had the lowest traffic with about 5% of the total traffic volume. This was due to the severity of the pandemic at that time and after the country had closed both local and international borders. Later in the year, precisely from September, the traffic flow began to rise as the number of active COVID-19 cases fell, and the restrictions of movement were relaxed. During this period, both local and international borders had opened and people were allowed to move and travel from one province to another. As a result, there was a gradual increase in traffic flow as major economic activities opened within the province.

3.2. Demographic Characteristics of Respondents:

The socio-economic characteristics of respondents as seen in

Table 1. Summary of Survey Table (Socio-economic characteristics).

Characteristics	Variables	Frequency	Percent
Gender	Male	142	31.2
	Female	314	68.8
Age	18–30 years	77	16.9
	30–39	269	58.9
	40–49	63	13.8
	50–59	28	6.1
	60–69	13	2.8
	70 years and above	6	1.3
Marital status	Single	155	34
	Married	285	62.5
	Divorced	4	0.9
	Widowed	12	2.6
Household size	Less than 6 persons	253	55.5
	6–9 persons	177	38.8
	10 persons and above	26	5.7
Educational status	No formal education	6	1.3
	Primary/secondary school education	22	4.8
	Bachelor	343	75.2
	Postgraduate	85	18.6
Average monthly income	<20,000 (rand)	12	2.6
	20,000–30,000	275	60.3
	30,001–40,000	85	18.6
	40,001–50,000	65	14.3
	50,001 and above	19	4.2

Source: Author’s Field work.

revealed that the majority of the participants were female (68.8%), married (62.5%), had bachelor’s degrees (75.2%), and earned between ZAR 20,000–30,000 (60.3%). Further information showed that more than half of those sampled (58.9%) were between the age of 30–39 years. Based on the above, it can be deduced that respondents are young adults, are educated, and hence had a good knowledge of the survey.

3.3. Trip Frequency of Respondents:

This section examines the differences between the number of trips made by respondents before and during COVID-19. This analysis is presented in

Table 2. Trip Frequency of Respondents before COVID-19, during first wave and second wave of pandemic.

	Mean	Std. Deviation	95% Confidence Interval for Mean		df1	df2	F	p-Value
			Lower Bound	Upper Bound
Daily trip before pandemic	3.12	1.92	2.98	4.12	2.00	240.00	7.96	0.01
Daily trip during first wave	1.63	1.31	1.18	1.91
Daily trip during second wave	2.32	3.10	1.31	2.24

Source: Author’s Field work.

, where the mean trip frequency of respondents is compared before the pandemic and during its first and second waves. This is necessary to determine any variations in household trip frequencies during the first wave when the pandemic started and the second wave when the restrictions of movement were relaxed. For the purpose of this study, trip frequency represents the average number of daily round trips made by respondents before and during the pandemic.

From our study, we discovered that the number of daily round trips made before the pandemic significantly exceeds that which was made during the pandemic (first and second waves). This implies that the pandemic had a negative impact on the mobility of people in Johannesburg (see Table 2). The result corroborates the findings from the TomTom statistics which revealed a decrease in the number of trips made. It can be seen from the ANOVA table that a noticeable and significant drop in movement occurred during the first wave of the pandemic. This can be attributed to the restrictions placed on both local and international movements in travel. During the second wave, there was an increase in movement with a mean daily trip frequency of 2.32, up from 1.63 during the first wave. This change is not difficult to understand, as major commercial outlets had started opening up due to the relaxation of movement restrictions by the government. The ANOVA result with F = 7.96, significant at ‘p < 0.05’, further showed the difference in the trip frequency of respondents before and during the pandemic.

3.4. Trip Purpose of Respondents

Information revealed that there was difference in trip purpose before and during the pandemic as seen in

Table 3. Frequency count and percentages for dominant trip purpose prior to and during COVID-19.

Variables	Levels	Frequency	% of Total
Dominant trip purpose before COVID-19	Work/School/Business	376	82.5
	Recreational	11	2.4
	Shopping	52	11.4
	Religious	17	3.7
Dominant trip purpose during COVID-19	Work/School/Business	96	21
	Recreational	21	4.6
	Shopping	312	68.4
	Religious	18	4
	Others	9	2

Source: Author’s Field work.

. While the majority (82.5%) of respondents reportedly made more work/school/business-related trips prior to the pandemic, a majority (68.4%) of the respondents made more shopping trips during the pandemic. This is as a result of the change of work patterns; the majority worked from home during the pandemic, hence the changes in trip purposes.

The t-test result as shown in

Table 4. Difference in dominant trip purpose before and during COVID-19.

	Grouping	Mean	Std. Dev.	Std. Error Mean	t	df	p-Value
Dominant trip purpose	Dominant trip before COVID-19	1.96	0.72	0.09	−8.21	170.00	<0.05
	Dominant trip during COVID-19	3.52	2.13	0.15

Source: Author’s Field work.

further revealed that significant differences existed in the dominant trip of respondents before and during the pandemic. The t-test result with t = −8.21 at ‘p = 0.05’ revealed that significant differences exist in the trip purpose of respondents.

3.5. Transport Mode of Respondents before and during the Pandemic

Information on the transport mode of respondents before and during the pandemic (see

Table 5. Frequency count of mode of transportation before and during pandemic.

	Transportation Mode	Frequency	Percentage (%)
Mode of Transportation before pandemic	Public transport	252	55.3
	Private car	177	38.8
	Bicycle/Walk	27	5.9
Mode of Transportation during pandemic	Public transport	102	22.4
	Private car	322	70.6
	Bicycle/Walk	32	7
	others

Source: Author’s Field work.

) showed that a larger percentage (55.3%) made use of public transport (taxis, Gautrain, and BRT), while 38.8% made use of private cars before COVID-19. However, there was a decline in the use of public transport during the pandemic. It was discovered that only 22.4% made use of this mode against the 70.6% of their counterparts who made use of private vehicles during the pandemic.

3.6. Travel Characteristics of Respondents by Mode before COVID-19

The result on travel characteristics of respondents (see

Table 6. t-test statistics between the mode of transportation before and during pandemic.

	Grouping	Mean	Std. Deviation	t	df	p-Value
Mode transportation	Transportation mode before pandemic	2.23	1.10	−4.08	180	0.01
	Transportation mode during pandemic	2.78	1.78

Source: Author’s Field work.

and

Table 7. Summary Statistics of Travel Characteristics of Respondents by Transport Mode Choice Before COVID-19.

Transport Mode	Average Travel Time (mins)	Average Travel Distance (km)	Dominant Trip Purpose
Public Transport	1.15	3.5	Work
Private Car	45	1.8	Work
Walking/Cycling	25	350 m	Recreation

Source: Author’s Field work.

) by transport mode before the pandemic revealed that they relied more on the use of public transport for long-distance trips and this mode recorded the trip with the longest travel time. It was discovered that the use of active travel was mostly used for recreational purposes. It thus revealed that people do not use active travel for most of their non-discretionary trips.

3.7. Travel Characteristics of Respondents by Transport Mode during COVID-19

The result of the travel characteristics of respondents (see

Table 8. Summary Statistics of Travel Characteristics of Respondents by Transport Mode Choice During COVID-19.

	Average Travel Time (mins)	Average Travel Distance (km)	Dominant Trip Purpose
Public Transport	15	1.2	Shopping
Private Car	35	2	Shopping
Walking/Cycling	10	250 m	Recreation

Source: Author’s Field work.

) during the pandemic revealed that respondents travelled shorter distances compared to the pre-pandemic period. Likewise, the average travel time of respondents was shorter when compared to pre-pandemic. This implied that differences exist in the travel characteristics of respondents before and during the pandemic. Based on this, we can assume that the pandemic had impacted the respondents’ mobility. This is evident in the choice of mode, trip purpose, travel time, and trip distance.

3.8. Multinomial Logistic Regression Model (MNL) for Mode Choice before and during the Pandemic

The last phase involved the testing and estimation of the factors influencing mode choice of respondents before and during the pandemic. The multinomial logistic model identifies the influence of the individual and household characteristics, and also trip characteristics of mode choice. Mode choices for the primary trip purpose before and during COVID-19 were modelled using multinomial logistic regression. In these models, the mode (public transport, private car, and walk/cycle) was set as the outcome variable, and the demographic and trip characteristics were entered as predictors/independent variables. The independent variables or predictors included: gender (male (1) and female (0), age, household size, income, travel time, and travel cost.

3.8.1. Model Specification

The primary data collected from the questionnaire administration were coded as different groups. The coded data were later used as the set of variables for model generation. The selected variables (see

Table 9. Description of variables used in the multinomial model.

Variables	Description	Measure
Age	Age of respondents	Scale
Household size	Number of People living and feeding from the same pot	Scale
Income	Monthly income of respondents (rand)	Scale
Gender	Gender (1 = male; 0 = female)	Scale
Travel Cost	Average cost of making a round trip	Scale
Travel Time	Average Travel time spent on trip	Scale
Transport Mode	Walk/Cycle, Private Car, Public Transport	Nominal

Source: Author’s Field work.

) for the logit model are based on previous theoretical and empirical research on the mode choice model. Thus, the final specification of the variables based on statistical testing is arrived at here. The dependent variables in this study are transport mode, which includes non-motorized transport (walk and cycle), private car, and public transport.

3.8.2. Goodness-of-Fit Tests

The multinomial logistic regression reports Pearson and deviance goodness-of-fit statistics. From the information presented in

Table 10. Goodness-of-fit.

Goodness-of-Fit before COVID-19				During COVID-19
	Chi-Square	df	Sig.	Chi-Square	df	Sig.
Pearson	263.321	307	0.992	202.198	307	1.000
Deviance	221.232	307	1.000	199.592	307	1.000

Source: Author’s Field work.

, it is believed that the attributes were viable and provide the best fit to the data. The Pearson and deviance chi-square test before COVID-19 indicates that the model is a good fit, [X²(321) = 263.321, p = 0.992], [X²(321) = 221.232, p = 1.000], respectively while during COVID-19, it was [X²(307) = 202.198, p = 1.000], [X²(307) = 199.592, p = 1.000].

In addition to this, the result of the likelihood ratio test, which explains the overall contribution of each independent variable, revealed that a significant relationship exists between the dependent variable and the set of independent variables. A likelihood ratio test indicates whether the model fits the data better than a null model. The chi-square statistic is the difference between the −2 loglikelihoods of the null and final models. The statistical significance (0.001) and (0.000) before and during COVID-19, respectively, indicate that the full model represents a significant improvement in fit over the null model.

With respect to the parameters sign, the significant negative coefficients which was presented in

Table 11. Model Development (Parameter Estimates) before COVID.

Mode		B	Std. Error	Wald	Sig.
Public Transport	Intercept	2.986	1.396	4.256	0.011
	Age	0.198	0.0102	0.876	0.001 *
	Travel Time	1.982	0.0763	2.108	0.423
	Travel Cost	−0.192	0.0710	0.322	0.010 *
	Household Size	2.132	0.0381	2.106	0.000 *
	Gender (Male, 1, Female 0)	0.842	0.502	1.712	0.562
	Income	0.312	0.011	0.612	0.020 *
Non-Motorized Transport (Walk/Cycle)	Intercept	1.752	0.131	1.823	0.001
	Age	0.232	0.106	0.561	0.010 *
	Travel Time	0.072	0.461	0.256	0.623
	Travel Cost	5.013	2.496	6.214	0.601
	Household Size	0.311	0.108	0.624	0.070
	Gender (Male, 1, Female, 0)	1.402	0.315	1.743	0.851
	Income	0.885	0.252	1.342	0.011 *

Source: Author’s Field work. * Significant at the 0.05 level. N = 456. Reference Category: Private car. Notes: Pseudo R²: Cox and Snell: 0.725, Nagelkerke: 0.810, McFadden: 0.692.

are interpreted to result in a decrease in the likelihood of that response category with respect to the reference category. On the other hand, the parameters with positive coefficients were considered to result in an increase in the likelihood of that response category.

3.8.3. Public Transport Relative to Private Car (Before COVID-19)

From the model equation and parameters estimates, it was discovered that only gender and travel time did not significantly influence the use of private cars. This implied that gender and time of travel does not play a major role or a deciding factor in using a private car. It was, however, discovered that the negative sign (−0.192) for travel cost revealed that travel cost would be expected to increase by 0.192 units while holding all other variables in the model constant. The Wald test statistic for the predictor cost value is 0.322, an associated p-value of 0.01, which is less than 0.05; thus, the regression coefficient has been found to be statistically significant.

3.8.4. Non-Motorized Transport Relative to Private Car (Before COVID-19)

The model equation revealed that age and income of respondents were significant influencers for the use of private cars. This implied that an increase in age and income of individuals might lead to an increase in the use of private vehicles against non-motorized transport.

3.8.5. Public Transport Relative to Private Car (During COVID-19)

From the model equation and parameters estimates (see

Table 12. Model Development (Parameter Estimates) During COVID-19.

Mode		B	Std. Error	Wald	Sig.
Public Transport	Intercept	4.218	0.968	5.161	0.001
	Age	1.112	0.801	1.402	0.001 *
	Travel Time	0.610	0.265	0.682	0.302
	Travel Cost	2.721	0.815	2.805	0.428
	Household Size	0.406	0.140	0.687	0.001 *
	Gender (Male, 1, Female, 0)	0.854	0.312	0.978	0.065
	Income	0.451	0.112	0.872	0.011 *
Non-Motorized Transport (Walk/Cycle)	Intercept	0.823	0.921	1.082	0.000
	Age	0.592	0.231	0.896	0.402
	Travel Time	0.106	0.248	0.321	0.070
	Travel Cost	1.123	0.872	1.243	0.321
	Household Size	0.221	0.101	0.321	0.010 *
	Gender (Male, 1, Female, 0)	1.601	0.032	2.453	0.010 *
	Income	1.521	0.912	1.702	0.001 *

Source: Author’s Field work. * Significant at the 0.05 level. N = 456. Reference Category: Private car. Notes: Pseudo R²: Cox and Snell: 0.701, Nagelkerke: 0.812, McFadden: 0.684.

), it was found that age, household size, and income significantly influenced the use of a private car compared with public transport during the pandemic. The implication of this is that considerations were placed on age, household size, and income of respondents in mode choice during the pandemic. It was, however, discovered that the positive sign (1.112) for age, (0.406) for household size, and (0.451) for income revealed that increasing the age, size of the households, and income would be expected to increase the use of a private car by 1.112, 0.406, and 0.451 units respectively while holding all other variables in the model constant.

3.8.6. Non-Motorized Transport Relative to Private Car (During COVID-19)

As revealed from the model equation, household size, gender, and income were found to be significant factors influencing the mode choice of respondents during the pandemic. The model equation revealed that an increase in household size, income, and gender variation might lead to an increase in the use of private vehicles against non-motorized transport. Overall, it was discovered that people placed value on age, cost of travel, household size, income level, and gender variation in choosing their mode before and during the pandemic. Hence, considerations have to be placed on these factors in mode choice selection in order to achieve sustainable transport planning in the future. The result of the pseudo R² explains the percentage contribution by which the independent variables influenced the dependent variable.

4. Discussions

This work has several implications with regard to new knowledge generation for academic purposes and policy formulation. It sets out to identify the impact of the pandemic on the mobility of people and how it can inform policy growth in developing countries. The pandemic has changed travel behaviour, and people are adjusting their commuting [13]. Based on this, organizations and individuals should be incentivized for developing initiatives which support change in travel behaviour towards sustainable transportation. Such initiatives supporting the use of other modes such as non-motorized transport may reduce over reliance on private automobiles and public transport. These actions can change travel behaviours resulting in less need for commuting and less congestion at terminals and bus stops.

The trip frequency of respondents before and during COVID-19 showed that there was a decline in trip-making. Studies by Mahmudur [45] and Palma et al. [46] also attested to a reduction in trip-making of people during the pandemic. A majority felt that the restrictions on movement had a large impact on their mobility during the pandemic due to the closure of major economic activities and concern about their health. Respondents were confined to their homes, as there was a fear of contacting the virus, particularly for those who did not have their own private vehicles. Hence a majority were confined to their homes. Viri and Tikkaja [47] support the assumption that trip frequencies and ridership declined, especially in public transport during the pandemic. In addition, the study of Venter, Gyl, and Cheure [19] on the trip frequency of people before and during the pandemic revealed that perceptions about trip reduction differed markedly across modes, with car users far more likely to report a reduction in trips than public transport users. For instance, only 40% of Gautrain users and 51% of minibus taxi users felt that their number of trips was still below pre-COVID-19 levels. This was compared to 87% of car and e-hailing users. The reason for this could be that car users are mostly higher and medium-income individuals who are more able to adapt to the pandemic by working from home.

In addition to this, the result on mode choice of respondents saw a decline in the use of public transport during the pandemic compared with the pre-pandemic period. From our study, it was discovered that 70.6% of respondents agreed that they used private cars during the pandemic against 38.8% of those who utilized this mode before the pandemic. The studies of Abdullah et al. [36], Jenelius and Cabecauer [48], Mogaji et al. [49], Zafri et al. [50], Munawar et al. [51], Palma et al. [46], Eisenmann et al. [28], and Dai et al. [27] also found a significant reduction in the use of public transport during the COVID-19 pandemic, hence corroborating this study. The reason for this modal shift could be linked to the health implications of the virus, as people had a conscious need to protect themselves. Contrary to our findings, the studies of Luiu et al. [52] on the impact of COVID-19 on the mobility of Kenyan people and Loa et al. [53] revealed that a majority made use of non-motorized transport modes during the pandemic. This was attributed to the increased transport fare which made it impossible for people to use public transport or private vehicles, hence resorting to the use of non-motorized transport.

The results from our findings can also be explained on the basis of the utility model. This proposes that the probability of an individual choosing an option is a function of socio-demographic and economic characteristics and the associated attractiveness of that option in relation to others in the choice set [54]. Individuals will choose the alternative with the highest utility. Hence, from the consideration of age, income, household size, and travel cost in the respondents’ mode choice, it can be deduced that individuals’ socio-economic characteristics, together with the cost of making a trip, play a major role in mode choice selection.

Findings from our study further revealed that there were changes in the trip purposes of people before and during the pandemic. It was evident that, while a majority (82.5%) made more of non-discretionary trips (work and school) before the pandemic, discretionary trips (shopping trips) accounted for a larger percentage (68.4%) of the trip purpose of respondents during the pandemic. Thus, shopping became the primary purpose of traveling during COVID-19. The significant shift from work and study to shopping trips revealed the effect of the pandemic on work- and school-related trips. The reason is not unrealistic, because self-isolation or lockdowns imposed by the authorities also reduced trips for work or education. However, shopping could be the primary reason why people needed to make trips in order to feed themselves, regardless of the level of restriction. Thus, shopping trips during a pandemic would generally be made for buying groceries and other household items, and would likely be shorter in distance and time compared to those for work and study. Abdullah et al. [36], Mahmudur [45], Jameel et al., [55], and Public Health England (PHE), [56] for example, also revealed the shift from work to shopping trips during the pandemic, thus corroborating our findings. Our findings also revealed that varying factors influenced individual transport mode choice. These factors include age, household size, gender, travel time, cost of travel and income. Scholars [57,58,59] also found a statistically significant relationship between mode choice and age, income, household size and gender of respondents. Based on the above, it is pivotal that developing cities such as Johannesburg consider integrating all modes of transport, especially non-motorized transport which can promote sustainable mobility in future.

5. Implications on Integrated Transport Planning in South Africa in the Future

A rethink is needed of the pathway towards upgrading and integrating all the modes, and especially the use of active transport should be made attractive to the public. Venter, Gyl and Cheure, [19] found that car users foresaw a much quicker recovery of demand in the future, while public transport changes will largely continue on present trends. The exception is rail patronage, which is likely to decline further in line with the collapse of rail services. It is also notable that walking and cycling seem not to be making any gains in usage. The commuters perceptions about active travel (walk/cycle) was no different from the findings of Venter, Gyl and Cheure, [19]; we found that the modal share of walking/cycling before and during the pandemic was only 5.9% and 7% respectively. This is very low compared to the modal share of private vehicles and public transport before and during the pandemic. The reaction to the use of active transport (walking/cycling) implies that people may not be willing to adopt this mode as an alternative to physical movement; thus measures must be put in place to make it encouraging and attractive for use. This will not only improve the living conditions of the people, but it will also help in environmental sustainability as this mode constitutes little or no harmful substances into the environment. There is, therefore, a need to integrate all the modes of transport in order to reduce reliance on private cars which has been found to have a detrimental effect on healthy living and the environment.

6. Conclusions

Our study discovered differences between the mobility patterns of respondents before and during the pandemic. These variations have been observed in trip frequency, trip purpose and mode choice, hence factors influencing these differences, especially from the aspect of the mode choice should be considered in order to achieve better transport planning in the future. In recent times, people and urban communities are experiencing a shift towards the use of non-motorized transport (NMT) which releases little or no gaseous substances detrimental to the health of human beings and the environment. Hence the need exists to integrate other modes of transport so as to ensure sustainable transport planning and development. In view of this, the provision of enabling and other mediating environments, and attention to self-concepts and capabilities regarding new transport behaviours are necessary to build a tractable, shared vision of any “new normal”. There is also the need to promote the use of non-motorized travel. This can be made possible when organisations encourage their employees to opt for the use of ride-sharing or cycling to work. In addition to the above, government should provide transport infrastructures which promote non-motorized travel such as bike racks, pedestrian walkways, and cycling paths that are safe and secure for users. The use of active travel will also help to eliminate some health issues such as depression, as people will still be able to conduct their activities and interact with their environment. They will not have to rely on the use of private cars or public transport thus limiting confinement to their homes in future pandemics.