COVID-19 and Travel: How Our Out-of-home Travel Activity, In-home Activity, and Long-Distance Travel Have Changed

Abstract

COVID-19 has made unprecedented impacts on our daily life. This paper investigates individuals’ immediate response to COVID-19, exploring out-of-home activities, in-home activities, and long-distance travel. Data for the Kelowna region of Canada comes from a web-based COVID−19 Survey for assessing Travel impact (COST). In addition to analyzing the survey, this research models adjustments in travel decisions by developing ordered logit models for in-home and out-of-home activities, and a binomial logit model for long-distance travel. Data analysis suggests a reduction of about 50% out-of-home activities/day/person during COVID-19 compared to the pre-pandemic period, with the only exception being picking up online orders which significantly increased in frequency. Individuals were engaged in longer duration of in-home activities; the average duration of teleworking, online shopping for groceries and other goods at-home was around 5.5 h/day/person, 32 min/day/person, and 26 min/day/person respectively. The out-of-home activity model results suggest that higher income, younger and middle aged individuals, and full-time workers are more likely to decrease their out-of-home activity; whereas, males, lower income groups, health care professionals, and picking up online orders are more likely to increase. The in-home activity model suggests that older and younger adults, higher and lower income, full-time workers, and highly educated individuals are most likely to increase their in-home activity frequency; in contrast, health care professionals are likely to decrease. Long-distance travel model results reveal that seniors, students, and airline travelers are more likely to reschedule; whereas, trips to visit friends and family are more likely to be cancelled.

Introduction

COVID-19 has had a significant impact on the way people travel all around the world. Many countries have issued social distancing measures, with some issuing stay-at-home orders. For example, Canada closed its border, including travel to/from USA, on March 21st. In British Columbia, Canada, strict physical distancing measures were put in-place on March 16th, which enforced a ban on gatherings of more than 50 people and closure of the majority of businesses including restaurants and bars, followed by the declaration of a public health emergency on March 17th (The COVID-19 Pandemic, 2016). These strict measures and travel restrictions have had a noticeable impact on the lifestyle of many people; specifically, on their daily and long-distance travel. The google mobility report reveals a 42.94% decrease in activities in public spaces such as retail locations and public transit stations from March to May of 2020 (COVID-19 Community Mobility Reports, 2020). Long-distance travel has also been significantly impacted by COVID-19, with only 59% of the average commercial air traffic flying in the last two weeks of March of 2020 (Charting the Decline in Air Traffic Caused by COVID-19 | Flightradar24 Blog, 2020). On the other hand, British Columbians have spent 17.96% more time at residential locations from March to May of 2020 (COVID-19 Community Mobility Reports, 2020). Around 50% of Canadians are tele-working during this time due to COVID-19 (Concerns About COVID-19 – April 21, 2020 – Leger, 2020). There has also been a surge in the demand of active transportation (AT); for instance, demand for shared bike services increased by 67% for March of 2020 in New York city (A Surge in Biking to Avoid Crowded Trains in N.Y.C., 2020). It is evident that individuals’ daily activities including travel and in-home activities have changed during COVID-19. However, due to the novelty of these pandemic restrictions, limited understanding exists on how individuals initially responded to the restrictions, in-terms of adjusting their daily out-of-home and in-home activities, as well as long-distance travel in small to medium sized communities.

This paper provides insights regarding how residents of medium sized cities have adjusted their travel behavior in response to the COVID-19 travel restrictions. The majority of the existing studies have focused on larger metropolitan areas such as Toronto, Chicago etc. (Loa et al., 2021, Shamshiripour et al., 2020). It is important to understand the behavioral response of individuals from larger, medium and smaller cities as well as rural areas, which is expected to assist in developing appropriate plans and policies for these distinct communities in a future pandemic scenario. Furthermore, this study provides results from the first wave of data, collected for the Kelowna region of British Columbia, Canada from March to May of 2020. Another set of data has been collected for the same region recently, which focuses more on the longer-term behavioral changes. Therefore, the findings of this study set the stage for future comparative retrospective studies taking into account individuals’ behavior at different timepoints of the pandemic. Data for this study comes from the first wave of a multi-wave web-based COVID-19 Survey for assessing Travel impact (COST) which collected information on daily travel and in-home activities, long-distance travel, and socio-demographic information from the respondents. This study adopts a logit modeling technique to investigate individuals’ adjustment of out-of-home activities, in-home activities, and long-distance travel. Specifically, ordered logit models were developed to examine the change in in-home and out-of-home activities, and a binary logit model was developed to investigate the change in long-distance travel. Furthermore, this study takes a holistic approach to investigate individuals’ initial response to the COVID-19 travel restrictions, in-terms of adjusting their daily out-of-home and in-home activities, as well as long-distance travel, which is limited in the existing literature. For example, the majority of the existing studies have focused on the adjustment of out-of-home activities during the pandemic (Abdullah et al., 2020, De Vos, 2020, Parady et al., 2020, Shamshiripour et al., 2020), ignoring how individuals have replaced it with in-home activity engagements.

Literature review

COVID-19 has forced individuals to spend more time at home with the closure of businesses, recreation facilities, workplaces, and schools, as well as local and international travel restrictions. As a result, individuals have started to find ways to replace or complement their out-of-home activities with in-home activities. For example, people have adapted to working and learning from a distance, eliminating their commutes as many physical workplaces and schools are closed (Concerns About COVID-19 – April 21, 2020 – Leger, 2020). Furthermore, online shopping has emerged as a viable alternative to avoiding travel for in-store purchases (As Online Orders Surge, Grocers Struggle to Deliver - The Globe and Mail, 2020). In addition to the daily activity, people's long-distance travel has been significantly impacted due to the world-wide closure of borders (Coronavirus Travel Restrictions and Bans Globally: Updating List - The New York Times, 2020). These changes are unprecedented as they occurred almost overnight and not been experienced in the recent past. The majority of the recent studies have investigated the way COVID-19 has impacted individuals’ travel. For example, Engle et al. (2020) found that even a very small infection prevalence in a community significantly reduced mobility in that region of the United States. They also discovered that socio-demographic characteristics had an impact on mobility reduction and government regulation compliance.

In the case of out-of-home travel activities, on average individuals made around 3.6 trips/day before the pandemic (Langerudi et al., 2016). The Google mobility report related to British Columbians reveals an increase in time spent at parks and residential locations along with a decrease in time spent at all other out-of-home destinations (COVID-19 Community Mobility Reports, 2020). This is in line with the BC government’s recommendations to avoid indoor and busy spaces (Phase 1 – BC’s Restart Plan - Province of British Columbia, 2020). The largest decrease, averaging around 58% reduction in engagement was observed in transit stations, closely followed by retail and recreation along with workplaces with an average reduction of 47% and 51% respectively. The reduction in transit station activities is in line with the findings of the global web index, which found that 54% of survey respondents reported avoiding public transit due to COVID-19 (Phase 1 – BC’s Restart Plan - Province of British Columbia, 2020). Further, Abdullah et al. (2020) reported the findings of a web-based survey conducted in different countries around the world and argued that during COVID-19 the primary trip purpose became grocery shopping and the primary modes became private car and active transport. A Chicago study found that 34% of online grocery shoppers placed their first order after social distance restrictions were imposed (Shamshiripour et al., 2020). They also discovered that 25% of the full-time workers and 52% of the part-time workers had their employment reduced or terminated, with the largest portions coming from those with lower annual income. This large share of job loss would likely result in an increase in time being spent in-home.

However, limited studies have examined how individuals are spending time at-home. Among the few, a recent study by Hossain et al. (2021) utilized a random parameter multinomial logit model to determine that out-of-home and in-home activity engagement are directly linked. This study found that individuals who conducted work related travel are less likely to engage in longer durations of in-home activity. Similarly, individuals performing more work in-home are likely to spend less time on household maintenance (Srinivasan & Bhat, 2005). Shabanpour et al. (2018) revealed that those who engaged in many out-of-home activities were less likely to engage in leisure and personal maintenance activities at home. Asgari et al. (2016) revealed that teleworkers spent more time on nonmandatory and out-of-home activities than commuters pre-pandemic. Fatmi (2020) found that higher income individuals were more likely to work from home during COVID-19.

In the case of long-distance travel, defining this term provides challenges as individuals having long commutes for work or school might not consider their daily travel as long-distance. This can be tackled by removing the strict distance limits from the long-distance travel definition. Miller (2019) recommended excluding daily or weekly travel from the long-distance travel category as those begin to classify as routine travel, even if they do cover large distances. Reichert & Holz-Rau (2015) investigated the mode split of long-distance travel in Germany and revealed that around 61.9% of long-distance travel was made by car, indicating that private car is a favourable method of long-distance transportation. The study also found that 50% of the respondents had not engaged in long-distance travel in the past 3 months, which was during the pre-pandemic period. In the last two weeks of march there was only 59% of the average air traffic compared to the same dates for previous years (Charting the Decline in Air Traffic Caused by COVID-19 | Flightradar24 Blog, 2020). Following the COVID-19 closure of international borders and the large number of cancelled flights, the impact of this pandemic on long-distance travel demands further investigation. Fatmi (2020), found that the majority of long-distance trips completed were within one province and the majority of cancelled or altered trips were international.

This paper will contribute to the literature by providing an overview of individuals’ out-of-home activities, in-home activities, and long-distance travel immediately following the implementation of COVID-19 travel restrictions. This paper investigates out-of-home activity engagement including the frequency, mode choice, companionship, along with subjective happiness and satisfaction during COVID-19. This paper also presents adjustment in out-of-home activities, which includes changes in frequency, mode, and companionship. Similarly, engagement and adjustment in in-home activities including their frequency and duration are analyzed. Furthermore, adjustment in long-distance travel including travel purpose, mode, and distance are investigated. Finally, logit modeling technique is adopted to explore how different types of out-of-home activities, in-home activities, and long-distance travel are adjusted. Adjustment in in-home and out-of-home activities are modeled using an ordered logit modeling technique and change in long-distance travel is modeled utilizing a binomial logit modeling technique.

Data description

The data used in this survey comes from a web-based COVID–19 Survey for assessing Travel impact (COST) conducted from March 24th until May 9th, 2020. The survey has three main components: daily activity, long-distance travel, and sociodemographic information. The daily activity component contained two modules: daily out-of-home activities, and daily in-home activities. The respondents were asked about their engagement in different activity types on the previous weekday. The out-of-home activity module collected information regarding the frequency, mode, companion, happiness, satisfaction, and type of all travel activities completed on the previous weekday. The activities can be categorized into: work which includes work/work related errands/school, household errands and other which includes personal business/household errands/shopping for major purchases/pick up or drop off passengers/health care, routine shopping, recreational/social which includes recreation/visit friends or family/civic/religious, and pick up online orders. Respondents were then asked how their out-of-home activities have changed during COVID-19 compared to the pre-pandemic period. They reported their change in mode, frequency, and companion of their out-of-home activities.

The next module collected information regarding the respondents’ in-home activities on the previous weekday. The in-home activities can be categorized into: sleep, personal maintenance which includes personal care/eating/drinking/grooming, household maintenance which includes general household activity/housecleaning/caring for household members, leisure which includes relaxing/socializing/watching tv/exercise/hobbies, discretionary which includes religious/spiritual/volunteering, work which includes work/school, online grocery/medical shopping, other online shopping, and other. The survey contains information regarding the duration and frequency of each activity. After that, respondents were asked to record the changes in their activity during COVID-19, which includes the change in frequency and duration of in-home activities. Following the weekday segment, similar questions regarding the out-of-home and in-home activities were asked for the previous weekend day.

The long-distance travel component of the survey asked respondents to report any long-distance travel that had been planned or conducted since January 2020. They were asked about the origin, destination, mode and purpose for their long-distance travel. The survey then asked respondents if COVID-19 had impacted their long-distance travel decisions or not. If so, they were asked to provide information related to the origin, destination, mode, and purpose of travel that had been affected. They were also asked about the actions taken (i.e. cancelled, rescheduled, or no change) for the impacted long-distance travel.

The sociodemographic component asked respondents about their gender, age, marital status, education, employment status, occupation, driving license, household size (i.e. HH size), number of children in the household (i.e. children), household income (i.e. income), and number of vehicles in the household (i.e. HH vehicles).

The data was compared to the census Canada data for Kelowna region (includes Kelowna, West Kelowna, Vernon, Lake Country, and Peachland) and validated using the iterative proportional fitting (Lomax & Norman, 2016). The weighted sample size for this study is 202. The validation results suggest that the sample reasonably represents the observed population of the Kelowna region. For example, males are under-represented in the COST survey by only 0.52%. For marital status, married population is over-represented by around only 1%. In the case of household-level attributes, two-person households and renters are over-presented by around 2% and 3% respectively. Overall, the validation exercise suggests that the COST sample shows satisfactory representation of the Kelowna population, as the majority of attributes fell within a few percent of the census distributions. For example, around 72.90% of the categories was found to be within 4.00% of the census distribution. Therefore, this validated sample was used for further analysis and modeling.

Modeling approach

Ordered logit model

The study adopts ordered logit (OLogit) modeling approach to analyze the change in individuals’ in-home and out-of-home activities. The OLogit model is based on the latent regression modeling approach which accounts for the ordinal nature of the response variable (Fatmi and Habib, 2019, Khaddar and Fatmi, 2021). The basic form of the model can be represented as:

$$y_i^{\ast }=\beta \text{'}{x}_i+{\epsilon }_i$$ (1)

Here, $y_i^{\ast }$ represent the change in in-home or out-of-home activities, $x_i$ is the vector of explanatory variables, $\beta \text{'}$ is the column vector of parameters corresponding to the explanatory variables, and ${\epsilon }_i$ is the random error term. The error term follows the assumption that it has a standard logistic distribution with constant variance. The dependent variable $y_i^{\ast }$ is continuous in nature and cannot be measured directly. The observable component $y_i$ can take values which is in discrete and ordinal scale (0, 1, …., J-1). For example, in case of changes in in-home activities the responses were recorded as decreased/stopped performing (0), no change (1), and increased (2). Therefore, the observable components take the following form:

$y_i=0$; when, $y_i^{\ast }\le {\mu }_0$

$y_i=1$; when, ${\mu }_0<y_i^{\ast }\le {\mu }_1$

$y_i=2$; when, ${\mu }_1<y_i^{\ast }$

Here, ${\mu }_j$ is the threshold parameter where $j$ = 0,1, 2, ……., J-1. The probability of an individual’s change in activity j is represented as follows:

$$P\left(y_i=j\right)=P\left({{\mu }_{j-1}\le y}_i^{\ast }\le {\mu }_j\right)=\frac{{\mathrm{e}}^{\left(x_i\beta \text{'}-{\mu }_{j-1}\right)}}{1+{\mathrm{e}}^{\left(x_i\beta \text{'}-{\mu }_{j-1}\right)}}-\frac{{\mathrm{e}}^{\left(x_i\beta \text{'}-{\mu }_j\right)}}{1+{\mathrm{e}}^{\left(x_i\beta \text{'}-{\mu }_j\right)}}$$ (2)

The above probability equation is in closed format. The parameter value is determined by maximizing the following log likelihood function:

$$\mathrm{L}\mathrm{L}=\prod _{j=0}^{J-1}\prod _{i=1}^N{\left(P\left({{\mu }_{j-1}\le y}_i^{\ast }\le {\mu }_j\right)\right)}^{{\delta }_{\mathit{ij}}}$$ (3)

where, N is the number of observations and ${\delta }_{\mathit{ij}}$ is a dummy variable. The value of ${\delta }_{\mathit{ij}}$ is 1, if individual i has experienced j change in out-of-home or in-home activity.

Binary logit model

The study utilizes binary logit modeling approach to analyze the change in individuals’ long-distance travel. According to the binary modeling approach, the study considers the value of the response variable as “1″ when a long-distance travel is cancelled and “0” when it is rescheduled, or no changes occurred. The general form of model can be represented as follows (Fatmi and Habib, 2017, Young and Liesman, 2007):

$$P=\frac{e^y}{1+e^y};y={\beta }_0+{\beta }_1{x}_1+{\beta }_2{x}_2+\cdots +{\beta }_k{x}_k$$ (4)

where, P is the probability of the long-distance travel being cancelled, ${\beta }_1,{\beta }_2,\cdots ,{\beta }_k$ are the parameter values corresponding to the explanatory variables $x_1,x_2,\cdots ,x_k$ respectively and ${\beta }_0$ is the intercept. The parameter values are estimated using the following log likelihood function:

$$\mathrm{L}\mathrm{L}=\prod _{j=0}^1\prod _{i=1}^N{\left(P\right)}^{{\delta }_{\mathit{ij}}}$$ (5)

where, N is the number of observation and ${\delta }_{\mathit{ij}}$ is a dummy variable. The value of ${\delta }_{\mathit{ij}}$ is 1, if an individual i has cancelled the travel.

Daily activity engagement and long-distance travel during COVID-19

This study investigates individuals’ immediate response to the COVID-19 travel restriction, by analyzing their daily activity engagement including out-of-home and in-home activities, and long-distance travel. In the case of daily activity engagement, this study only focuses on the weekday activities.

Out-of-home travel activities

On average, 1.6 out-of-home travel activities/day/person were done. Comparing this with Kelowna’s travel survey conducted every 5-years, this is around 50% lower than the pre-pandemic period. Recreational/social is the only activity type that was performed more than twice/day/person, 32% of these activities were reported as walk or bike. Around 84.13% of all activities were done using private car; whereas only 2.60% of all activities used public transit. This could be influenced by the higher-level of perceived health risk on public transit during a pandemic.

Among the out-of-home travel activities, the majority of the travel was done alone, with 83.33% of work, 75.32% of routine shopping, 69.70% of recreational/social, 60.00% of online order collection, and 48.39% of household errand activities being conducted alone. Fig. 1 a shows the distribution of out-of-home travel activities by companion. Note that this distribution excludes solo travel. The analysis suggests that the next most common companion for all but work activities was a spouse or child.

Fig. 1.

Distributions of out-of-home and in-home activities.

In the case of satisfaction and happiness associated with travel during COVID-19, data was recorded on a five-point scale of: 1 being very dissatisfied/very unhappy and 5 being very satisfied/very happy. The average score of the responses are reported in Fig. 1b as the happiness or satisfaction score. In general, respondents reported being happy and satisfied with their travel; specifically, recreational/social travel was reported to be the most satisfying and happy, which is in line with the prior literature (Zhu & Fan, 2018). Respondents also reported being the least happy and satisfied with travel performed to pick up goods that had been ordered online. This might be attributed by the wait times associated with picking these items up once the location is reached. Picking up online orders includes both online grocery orders and other online purchases.

In the case of adjustment in out-of-home activities, a very small portion of respondents indicated an increase in frequency (Fig. 1c). Approximately 17% of respondents experienced an increase in online order collection while engagement in other activities only increased within the range of 0.63–4.43%. This increase in online shopping shows a major shift away from traditional shopping, not only for clothing and household goods but for groceries as well. These changes during the pandemic could have a lasting impact on the way shopping is conducted in the future.

Most respondents reported no change of travel mode; specifically, 82.5% of respondents made no mode change over all of the activities. There was also a noticeable share of mode shift - either from active modes to private car or vice versa. For example, among the recreation/social activities, about 14.75% of the travel shifted to walk/bike from a private car. On the other hand, around 11% of routine shopping shifted from walk/bike to private car. In the case of change in travel companion, no change accounted for 84.24% of all activities, a shift from a companion to solo represented 12.96% of all activities, and only 2.80% of all out-of-home activities shifted from solo to with a companion.

In-home activities

With governments urging the public to remain at home during the pandemic, it is important to investigate the activities being done in-home. This will provide insights into the out-of-home activity adjustments and help provide a more complete picture of daily activities. The distribution of the duration of in-home activities along with the average activity duration are presented in Fig. 1d. This figure shows that the most time is being spent sleeping and doing leisure and work activities. For example, the average duration of in-home work activities is around 5.5 h/day/person. Among the individuals who are tele-working, more than 78% are working for more than 3 h/day. On average, individuals are spending more than 4 h/day on in-home leisure activities such as watching tv. The average time spent conducting online grocery and other shopping is around 32 min/day/person and 26 min/day/person respectively.

The change in duration and frequency of in-home activities is illustrated in Fig. 1e. Overall, the frequency of most of the activities has increased. Household maintenance and leisure activities have seen significant increases, likely due to the extra time spent at home. For example, more than 60% of the respondents showed an increase in household maintenance activities by more than 60 min/day. Around 85% of leisure activities have increased by more than 60 min/day. In the case of tele-working, more 40% has increased by more than 60 min/day.

Long-distance travel

With governments closing international borders and calling for repatriation, long-distance travel has been notably adjusted. The largest share of adjusted travel was international airline travel and the largest share of travel completed was regional travel using private car. Here, regional travel is defined as travel within the same province/state, domestic is within the same country, and international crosses an international border. The distribution of completed and adjusted travel is presented in Fig. 1. In total 42.08% of respondents reported completing some long-distance travel since January 2020 and 73.80% of respondents’ long-distance travel decisions were impacted by COVID-19. Among the adjusted trips, 89.23% were cancelled, 9.23% were rescheduled, and 1.54% returned home due to COVID-19. Airline travel account for around 58.52% of the adjusted travel and 48.98% of completed travel, while private car accounts for only 29.70% of adjusted travel and 40.82% of completed travel. This is likely due to the widespread cancellation of flights and the lower level of perceived safety in airlines during the pandemic.

Model results

Three models were developed in this study: two ordered logit models to investigate the adjustment in out-of-home activities and in-home activities, and a binary logit model for change in long-distance travel. The parameter estimation results suggest that the majority of the variables retained in the final model possess statistical significance at least at the 95% confidence interval.

Out-of-home activity model results

The change in activities is categorized into the following ordinal scale: increased, no change, and decreased/stopped performing. The parameter estimation results are reported in Table 1 . Model results suggest that male individuals, lower income group, and individuals with education below a bachelor’s degree, have a higher likelihood to experience an increase in out-of-home activities. This might be attributed by the lack of available space at home for longer work-from-home and limited access to teleworking equipment, as well as belonging to occupations that do not allow remote working. Health care workers reveal a higher likelihood for increased out-of-home activities, which could be attributed by the increased demand of the health care workers at the hospitals and other health care centres during the pandemic. This reveals that these demographics might be at a higher risk during this pandemic, as they are leaving their homes more frequently and likely having more social interactions in public. Among the activity types, individuals show a higher likelihood to pick up online orders; particularly, groceries. This might imply that people are taking advantage of the alternate methods for in-store shopping such as online ordering followed by curb side pick-up during this pandemic. This shift could result in a longer-term change towards online grocery shopping. Recreation, major shopping, and visiting friends/family tend to decrease, which is in line with the government’s recommendations urging the public to avoid social interactions outside of the household.

Table 1.

Ordered logit model results for out-of-home activities during COVID-19.

Variables	Definition	Coefficient	t-stat.
Income < $50,000	Household income below $50,000	0.43	4.03***
Income >= $100,000	Household income of at least $100,000	−0.21	−1.8*
Age 30–44	Individuals aged 30 and 44 years old	−0.34	−2.99***
Age 18–29	Individuals aged 18 to 29 years old	−0.82	−5.85***
Gender: male	Male individuals	0.45	4.89***
Employment status: full-time	Individuals with full-time employment	−0.23	−1.39
Occupation: health care	Health care professionals	0.36	1.95*
Occupation: management	Management professionals	−0.40	−3.91***
Education level: below bachelor’s degree	Individuals with education below a bachelor’s degree	0.57	5.93***
Activity type: recreational	Recreational activities	−0.57	−3.72***
Activity type: online groceries	Pick up groceries or medication ordered online	2.10	8.69***
Activity type: major shopping	Shopping for major items	−0.35	−2.03**
Activity type: visit friends or family	Visit friends or family	−1.29	−8.08***
Threshold parameters
Threshold 1		0	–
Threshold 2		1.14	24.93***
Goodness-of-fit measures
Log likelihood convergence	−71.96
Log likelihood constant	−88.83
Adjusted r-squared	0.19

Note: *, **, and *** represent 90%, 95%, and 99% confidence interval respectively.

In-home activity model results

The adjustment in in-home activities is classified into: increased, no change, decreased/stopped performing. The results of this model are reported in Table 2 . Model results suggest that health care professionals showed a higher propensity to decreased in-home activities. Health care workers are likely to experience an increased need to travel to their workplaces during the pandemic; therefore, they might spend less time at-home. The majority of variables show an increase in in-home activities, indicating that most demographics have increased their in-home activity frequency. This finding might indicate that individuals are replacing their out-of-home activities with higher engagement in in-home activity.

Table 2.

Ordered logit model results for in-home activities during COVID-19.

Variables	Definition	Coefficient	t-stat.
Income < $30,000	Household income below $30,000	0.59	2.91***
Income >= $150,000	Household income of at least $150,000	0.80	4.19***
Age 18–29	Individuals aged 18 to 29 years old	1.15	6.78***
age >= 65	Individuals aged 65 years and above	0.51	3.49***
Gender: male	Male individuals	0.52	4.24***
Employment status: full-time	Individuals with full-time employment	0.62	4.29***
Education level: Bachelor’s degree or above	Individuals with a minimum education of a bachelor’s degree	0.79	6.13***
Occupation: health care	Healthcare professionals	−0.91	−2.97***
Occupation: education	Education professionals	0.31	2.13**
Activity type: online shopping both groceries and others	Shopping online for groceries and other goods	0.26	1.89*
Activity type: discretionary	religious/spiritual/volunteering activities	−1.11	−6.09***
Threshold parameters
Threshold 1		0	–
Threshold 2		2.10	24.89**
Goodness-of-fit measures
Log likelihood convergence	−1062.97
Log likelihood constant	−1074.09
Adjusted r-squared	0.011

Note: *, **, and *** represent 90%, 95%, and 99% confidence interval respectively.

Long-distance travel model results

The change in long-distance travel is categorized into: cancelled and rescheduled. The model results are presented in Table 3 which suggest that middle-aged and older adults along with students are more likely to reschedule their travel than other groups. Students might have been forced to reschedule their travel for studying abroad and summer holidays. The unprecedented scenario created by the pandemic might trigger the general unwillingness of individuals to reschedule their travel, followed by border closures and flight cancellations which also pose a significant barrier to long-distance travel. Interestingly, travel for the purpose of visiting friends and family shows a likelihood to be cancelled rather than rescheduled.

Table 3.

Binary logit model results for long-distance travel during COVID-19.

Variables	Definition	Coefficient	t-stat.
Constant	–	2.63	5.54***
Income <= $50,000	Household income maximum $50,000	1.12	1.52
Age 30–44	Individuals aged 30 and 44 years old	−0.73	−1.02
age >= 65	Individuals aged 65 years and above	−2.69	−4.04***
Gender: male	Individuals aged at least 65 years old	0.85	1.63
Occupation: education	Education professionals	1.43	1.68*
Occupation: student	Individuals who are students	−2.52	−2.47**
Trip purpose: visit friend/ family	Visiting family or friends	2.21	1.85*
Travel mode: airline	Travel conducted or planned by airline	−0.70	−1.08
Goodness-of-fit measures
Log likelihood convergence	−1976.47
Log likelihood constant	−2096.55
Adjusted r-squared	0.057

Note: *, **, *** represent 90%, 95%, and 99% confidence levels respectively

Conclusion

This study presents the findings on individuals’ immediate response to the COVID-19 travel restrictions - in-terms of engagement in out-of-home activities, in-home activities, and long-distance travel, and how their travel decisions have changed compared to the pre-pandemic period. This study analyzes the engagement and adjustment in out-of-home travel activities, in-home activities, and long-distance travel immediately following the implementation of restrictions due to COVID-19.

Individuals were engaged in 50% less out-of-home travel activities/day/person than the pre-pandemic period. The average duration of in-home work activities was around 5.5 h/day/person. This finding highlights a shift from out-of-home activities to in home activities during the pandemic which can be largely attributed to the temporary closures of workplaces and businesses along with the travel restrictions. The only out-of-home activity that had a significant increase in frequency was picking up online orders. In the case of travel companion, the majority of trips were made alone or with another household member such as spouse. The decrease in out-of-home activities has resulted in an increase of in-home activities. There was an overall increase in the frequency of in-home activities with the largest increases in household maintenance, leisure, and work. These activities also had longer average durations, only surpassed by sleep. The majority of the respondents reported an increase in online shopping for groceries and other goods. In the case of long-distance travel, a large share of the travel was cancelled or rescheduled. The majority of the completed long-distance travel was regional using private vehicle. The largest share of the adjusted travel was international air travel, likely due to the cancellation of international flights.

The out-of-home activity model results suggest that the majority of the variables such as higher income groups, younger and middle-aged individuals, and full-time workers are more likely to decrease their out-of-home activity. Among the activity type, activity participation such as visiting friends and family, recreational, and major shopping activities have a higher probability to decrease. However, males, lower income groups, and health care professionals are more likely to experience an increase in out-of-home travel activities. Engagement in picking up online orders, specifically groceries, reveals a higher likelihood to increase. In the case of the in-home activity model results, the majority of the variables such as older and younger adults, higher and lower income groups, full-time workers, and highly educated individuals correspond to an increase in activity frequency. Interestingly, health care professionals showed a decrease in in-home activity frequency. Discretionary activities also showed a negative relationship. In the case of the long-distance travel model results, seniors and students have a propensity to reschedule their travel rather than cancel it. Airline travel was also more likely to be rescheduled than cancelled. Interestingly, trips to visit friends and family were more likely to be cancelled.

Although the study provides important insights in terms of individuals’ activity engagement immediately after the travel restrictions imposed by COVID-19, it has certain limitations. While the data has been validated for socio-demographic characteristics it is important to note that this study used a convenience sampling technique. Respondents voluntarily participated in this web-based survey delivered via social media advertisements, which might have resulted in data that did not encompass the entire population of the region. As a result, an iterative proportional fitting (IPF) method was used to weight the data and improve the representativeness of the sample for the Kelowna region. In addition, there might be some inherent reporting bias in the data. For instance, the reporting of data regarding the in-home activity duration is often associated with recalling bias and less accurate than that of actual duration. More advanced methods of data collection should be considered in future researches to avoid such biasness. Furthermore, future studies should consider utilizing more advanced econometric modeling methodology to capture multi-domain heterogeneity among individuals which will help the practitioners and policy planners in identifying long term behavioral changes.

The scope of this paper did not include perceived safety of out-of-home activities and long-distance travel which could be an important factor in mode choice and frequency. Future studies should investigate mode changes in long-distance travel as well as consider jointly modeling in-home, out-of-home, and long-distance travel to understand the interaction among them. A second wave of this survey has been distributed and will be analyzed to compare with this study to understand respondent’s activity before, immediately, and months after restrictions have been in place. Overall, the findings of the study provide important insights into how individuals adjusted to a completely new and challenging scenario imposed by the pandemic through modification of activity engagement. The study will assist future researches and policy makers by providing a better understanding of the changes in in-home, out-of-home activities, and long-distance travel in small to medium sized cities such as Kelowna and thereby help in effective policy making to better cope with such pandemic situation in the future.

Author contributions

The authors confirm contribution to the paper as follows: study conception and design: M. Fatmi, and C. Thirkell; data collection: M. Fatmi; analysis and interpretation of results: M. Fatmi, C. Thirkell, and S. Hossain; draft manuscript preparation: M. Fatmi, C. Thirkell, and S. Hossain. All authors reviewed the results and approved the final version of the manuscript. The authors do not have any conflicts of interest to declare.