Modeling Telecommuting and Teleshopping Preferences in the Post-Pandemic Era

Abstract

COVID-19 mitigation measures triggered a sharp increase in the adoption of teleshopping and telecommuting activities. However, there is a need to understand the extent to which past frequencies and experiences will affect post-pandemic teleactivity behavior. Moreover, teleshopping and telecommuting are interconnected, and a relationship may exist between them in the post-pandemic world. This study investigates post-pandemic preferences toward online grocery shopping, online food ordering, and working from home by using a multivariate ordered probit (MVOP) model. The data come from a web-based survey conducted for the Central Okanagan region of Canada. Model results confirm the presence of unobserved factors influencing telecommuting and teleshopping choices. Looking at endogeneity, working from home after the pandemic revealed a positive effect on online grocery shopping. However, results were not the same for post-pandemic online food ordering. Model results also confirm the significant impact of past teleactivity frequencies and experiences on post-pandemic preferences. Overall, the findings provide important insights into post-pandemic activity and travel patterns which can be used for robust policymaking.

The COVID-19 pandemic has caused drastic changes in day-to-day mobility patterns and lifestyles ( 1 ). These changes occurred as a result only of government-imposed travel restrictions/social distancing measures but also of self-awareness of the need to maintain physical distance and avoid exposure to crowds and public spaces. These countermeasures led individuals to perform many of their daily activities virtually instead of in-person. For instance, telecommuting (working from home), teleshopping (online shopping), telelearning (online learning), and telehealth (online medical consultation) are some of the online activities that replaced the need for travel to engage in in-person interactions ( 2 ). Specifically, during the early stages of the pandemic, a substantial increase in teleshopping (online shopping) and telecommuting (working from home) was noticed. According to a global consumer behavior survey, a significant number of consumers have shifted to online shopping, causing a 35% dip in in-store shopping ( 3 ). A sudden rise in online grocery shopping (65%) and online food ordering (31%) was noticed in 2020 ( 4 ). Similarly, a boost in working from home was triggered in several parts of the world ( 5 ). This shift to teleactivities led to changes in traditional mobility patterns, reducing traffic congestion and traffic-related emissions ( 6 ). Given the continuous engagement in teleactivities during the pandemic (for over a year), individuals became accustomed to this “new normal” ( 7 ). However, it is important to ascertain whether they will continue to engage in teleactivities in the post-pandemic period or prefer returning to the pre-pandemic culture. For example, individuals with no experience of telecommuting might prefer an in-person office work environment ( 8 , 9 ) and vice versa. Similarly, individuals with past exposure to teleactivities (e.g., online grocery shopping) might prefer to continue to perform them remotely in the future ( 7 ). In this line of investigation, some studies confirm the interaction between teleshopping and telecommuting as they are interconnected and can influence each other ( 10 – 12 ). However, the COVID-19 pandemic has triggered a long-term behavioral change which has the potential to affect interactions among different activity types ( 13 ). For example, individuals working from home can choose to engage in other activities remotely or they may perform them in person as this provides them with the opportunity to travel outside ( 14 ). There is, therefore, a need to explore how one life domain decision, such as telecommuting, can significantly affect other domains such as teleshopping in the post-pandemic culture. Moreover, to understand the post-pandemic activity-travel behaviors, it is important to consider the extent to which activity-travel behavior before and during the pandemic will influence them.

This study aims to investigate how telecommuting and teleshopping preferences are likely to evolve in the post-pandemic era. Particularly, this study explores the preferences for online grocery shopping, online food orders, and working from home. The data comes from a web-based survey conducted in the Central Okanagan region of British Columbia, Canada from November 2020 to January 2021, capturing behavioral preferences of different time frames (pre-pandemic, early stages of pandemic, later stages of pandemic, and post-pandemic). This study extends the understanding of how teleshopping and telecommuting are interconnected and can influence each other. To understand the unobserved factors influencing telecommuting and teleshopping behavior, this study adopts a joint multivariate ordered probit (MVOP) modeling framework. In addition, this study considers possible endogenous effects among the teleactivities, exploring the effect of one life domain, such as working from home, on another life domain, such as online grocery shopping and online food ordering. One of the key features of this study is to examine the effects of previous travel behavior, such as the frequency of online activity engagement, and related experiences during different time points, that is, pre-pandemic (i.e., before March 2020), early stages of the pandemic (i.e., during March 2020), and later stages of the pandemic (i.e., during the survey which was which was conducted between November 2020 and January 2021). Moreover, this study tests the effect of sociodemographic, accessibility of different travel modes, and built environment attributes. The rest of the article is structured as follows: the literature review section provides insights on the existing literature on people’s activity-travel behavior. The data section then provides comprehensive information about the survey design and a general description of the collected data. Thereafter, the discussion is expanded by providing an overview of the methodology adopted and the study findings. Finally, the article concludes with a summary of the findings and potential future research agendas.

Literature Review

In recent decades, transportation planners and researchers have started investigating factors influencing teleactivities such as telecommuting and teleshopping because of their potential to influence activity-travel behavior ( 15 , 16 ). Studies have explored various methodological avenues to model the preferences toward teleactivities in the past, for example, a nested logit model to investigate the online shopping preferences of different socioeconomic groups during the pandemic ( 3 ). Another study adopted a structural equation modeling approach to investigate the effect of teleworking on activity-travel behavior during the pandemic ( 17 ). In addition, a latent class cluster analysis approach was used to capture the heterogeneity that exists among different groups of teleworkers ( 18 ). However, past literature confirms that teleshopping and telecommuting are intertwined and can mutually influence each other ( 12 ). Therefore, few attempts were made to model teleshopping and telecommuting jointly before the pandemic, shedding light on activity-travel behavior. For instance, ( 10 ) used a copula-based joint bivariate ordered-response system to model teleshopping and telecommuting frequencies. Another study adopted a joint binary logit regression framework to analyze the effect of ICT usage and experiences, geographical accessibility, and other sociodemographic characteristics on home-based online working and shopping ( 11 ). Although various methodologies have been used to study the impact of several determinants on teleactivities ( 19 – 22 ), unprecedented events such as pandemics warrant additional attention to this topic. For instance, there is a need to understand how working from home can affect other teleactivities such as online shopping. Additionally, to understand post-pandemic activity-travel behavior there is a need to consider the effects of experiences and frequencies before and during the pandemic.

The COVID-19 pandemic has altered people’s daily activity-travel patterns and choices ( 23 ). To prevent the spread of the virus, governments across the world adopted various countermeasures, which have triggered a shift from in-person activity engagement to the adoption of teleactivities ( 1 ), such as online shopping and working from home ( 24 ). In the case of telecommuting engagement rates, the number of people working from home substantially increased during the pandemic as compared with the pre-pandemic levels across the world ( 8 , 24 ). Many individuals found working from home to be productive because of there being “no commute time” ( 25 ). By contrast, distractions at home were one of the reasons for a reduction in the propensity toward working from home ( 26 ). Factors such as low travel costs, time saving, and higher flexibility from employers were some motivational determinants for adopting working from home ( 27 ). Another teleactivity that significantly replaced in-person shopping during the pandemic was teleshopping ( 28 ), more specifically online food ordering ( 29 ), and online grocery shopping ( 30 ). Recently, Li et al. revealed a significant increase in online grocery shopping during the initial stages of the pandemic as compared with pre-pandemic settings ( 31 ). Furthermore, a study found convenience and delivery times to be the most significant factors affecting teleshopping choices ( 32 ).

In addition to the COVID-19 pandemic, these teleactivities are influenced by determinants such as access to ICT (e.g., the internet), attitudes/experiences toward teleactivities, access to mobility tool ownership and built environment, and sociodemographic characteristics. For example, familiarity with the internet and its frequent use is positively linked with the adoption of teleactivities ( 33 , 34 ). Most importantly, internet access at home via a variety of ICT devices allows one to engage in working from home and also to participate in online shopping ( 11 ). In relation to people’s attitudes and experiences toward working from home, studies suggest that reducing family–work conflicts and distracting environments can lead to positive experiences with work productivity ( 35 ). Good access to home internet promoted the adoption of working from home, while lower delivery delays, easy navigation of websites/applications, infrequent order cancellations, and high quality products increased online shopping intentions ( 33 , 36 ).

Among sociodemographic characteristics, gender influences teleactivity adoption choices. This could be because females tend to take higher responsibility for household-related tasks which encourages them to engage in teleactivities (both working from home and online shopping) as compared with men ( 10 ). Younger individuals showed a higher propensity toward more frequent teleactivity engagement, which could be because of greater appreciation of ICT usage and a tendency to adapt easily to rapid digitalization ( 37 ). Among household-level attributes, higher-incomes showed higher accessibility to ICT-related devices (e.g., smartphones and desktops) which increased their propensity to the adoption of teleactivities ( 38 ). However, the results are not the same for low-income households. Households with children (aged less than 16 years old) revealed a similar trend, which may be because of childcare responsibilities or that such households have higher needs and have to make more trips which can be more inconvenient if they involve children ( 39 ).

On the other hand, access to travel modes (i.e., access to a vehicle, owning a driving license, owning a transit pass) might have a negative impact on teleactivities. Several studies revealed that individuals owning cars are less likely to engage in online shopping ( 38 ), as they provide a higher level of travel flexibility and the opportunity to travel by their preferred mode. This contrasts with the results for households with lower vehicle ownership which showed their dependency on ICT to perform daily activities. Lastly, the built environment plays a significant role in the adoption of teleactivities ( 40 ). For instance, individuals living in suburban settings (far from the central business district) are more likely to adopt working from home given the higher commute distances and travel times ( 41 ). In addition, studies suggest that online shopping is more prevalent among individuals living in urban areas as they tend to have better computer skills and access to ICT infrastructure ( 42 ). However, some studies suggest that even rural residents follow the same trend because of the lower accessibility to shops, making the need for online shopping for daily needs all the greater ( 43 , 44 ).

The contribution of this study is to investigate how preferences toward online grocery shopping, online food ordering, and working from home will shape up after the COVID-19 pandemic. One of the key features of this study is to capture the intricate relationship that might exist among these activities, along with exploring the plausible endogeneity. Methodologically, a MVOP model is adopted to account for unobserved factors that affect preferences toward these activities, while also controlling for the effect of multiple explanatory and endogenous variables simultaneously. Furthermore, this study investigates how activity-travel behavior before and during the pandemic might influence the post-pandemic frequency of teleshopping and telecommuting. Lastly, the effect of sociodemographic, mobility tool ownership, and built environment attributes are also considered in this study. Some of the questions this study aims to answer in the post-pandemic context are: Is there an intricate relationship among activities such as teleshopping and telecommuting? Does working from home increase or decrease the likelihood of engaging in grocery shopping online or vice versa? Does working from home increase or decrease the likelihood of engaging in online food orders or vice versa? How do engagement rates in teleactivity before the pandemic and during the early and later stages of the pandemic affect post-pandemic choices? How do experiences with teleshopping and telecommuting during the pandemic affect post-pandemic choices?

Data Source

This study uses data from a web-based survey conducted in the Central Okanagan region of British Columbia, Canada. This region consists mainly of five cities which include Kelowna, West Kelowna, Vernon, Lake Country, and Peachland. The survey was conducted from November 2020 to January 2021. The survey was distributed through Facebook. Information collected was for four different periods: pre-pandemic (i.e., before March 2020), early stages of pandemic (i.e., during March 2020), later stages of pandemic (i.e., during the survey which was November 2020 to January 2021), and post-pandemic (no travel restrictions/social distancing measures). The post-pandemic section assumes a future scenario in which all restrictions related to the COVID-19 pandemic are lifted and is the only stated preference component in the survey. The data collected on the other three time points (i.e., pre-pandemic and early and later stages of the pandemic) revealed preference data from the respondents. Respondents were asked to provide feedback on the activity engagement frequency using options such as every day, a few times a week, a few times a month, a few times a year, or never. In the case of the later stage of the pandemic (i.e., during the survey which was November 2020 to January 2021), respondents were asked to report their teleactivity engagement frequency in the seven days immediately before completing the survey. To capture these patterns objectively in the seven-day period different options were used, varying from one day to five or more days, or not at all. This period is considered as the later stage of the pandemic because at this stage some of the COVID-19 related restrictions had been eased but some were still in place such as “avoid all non-essential travel,”“restrict large gatherings,” and others ( 45 ). Moreover, cases with emerging COVID-19 variants were also reported in the region in January 2021 ( 46 ).

The information collected is related to travel behaviors and experiences including themes such as travel patterns, telecommuting, telelearning, teleshopping, and others. More specifically teleactivities considered were: online shopping for grocery and medical supplies, online ordering of food from a restaurant, working from home, among others. In addition to work and home location, the survey captured socioeconomic and demographic information such as age, gender, education, employment status, occupation, driver’s license and transit pass ownership, as well as household-level attributes which include annual income, dwelling type, and vehicle ownership. Secondary data sources such as open data portals for the Okanagan cities for land use and transportation network information, point of interest information from Desktop Mapping Technologies Inc. (DMTI), and the 2016 Canadian Census, were also used in this study. Built environment and land use–related attributes were generated for 500-m buffer areas from the home location of the respondents using ArcGIS (v.10.6.1). Transportation infrastructure attributes include length of sidewalks and bike lanes. Locations of multiple destination points were used to generate accessibility measures such as distance from home to nearest activity points using ArcGIS. Further, to derive the land-use index, land-use characteristics including the percentage of residential, commercial, and industrial land uses, among others were used, following the method proposed by Bhat and Gossen ( 47 ). The land-use index ranges from 0 to 1, where 0 represents perfectly homogeneous land use and 1 perfectly heterogeneous land use. The following equation is used to calculate the land-use index.

$$\mathrm{Land}\mathrm{use}\mathrm{index}=1-\left\{\frac{|\frac{r}{T}-\frac{1}{3}|+|\frac{c}{T}-\frac{1}{3}|+|\frac{m}{T}-\frac{1}{3}|}{\frac{4}{3}}\right\}$$

(1)

where

T is the total land area,

r is the residential land area,

c is the commercial/industrial land area, and

m is the miscellaneous land area.

Following data collection and cleaning, to reasonably represent the Okanagan population a validation exercise was performed by comparing the survey data with the 2016 Canadian census. More specifically, the survey sample was weighted using an iterative proportional fitting technique (IPF) ( 48 ). The validation results reveal that the majority of the sample characteristics lie within a few percentage points from the census data (Table 1) and represent the Okanagan population to some extent. For example, females have a share of 51.5% in the survey and 49.1% in the census. In the case of age, individuals aged 18–29 are over-represented by 4%. However, the data under-represents the older population aged 75 years and above. Furthermore, individuals in management-related occupations have a share of 11.3% and 10.7% in census and survey data, respectively, which shows under-representation by only 0.6%. For household-level attributes, household size of 5 or more is over-represented by only 0.3%. Lastly, the share of individuals with an annual household income of $50,000–$99,999 are under-represented by 1.9%, whereas, individuals with an annual household income of $200,000 and above are over-represented by 3.3%. The final surveying sample was 226, of which 197 had all the information required for the model.

Table 1.

Comparison of Survey Data with Census Data

Attributes	Census (%)	Survey (%)
Gender
Male	48.5	49.1
Female	51.5	50.9
Age
18 to 29 years	12.6	16.6
30 to 34 years	7.2	8.5
35 to 39 years	6.7	12.1
40 to 44 years	6.7	10.3
45 to 49 years	7.3	5.4
50 to 54 years	9.5	15.7
55 to 59 years	10.9	11.2
60 to 64 years	10.4	8.5
65 to 69 years	9.6	6.7
70 to 74 years	6.9	4.0
75 years and above	12.2	1.0
Annual household income
$ 0–$49,999	34.4	32.3
$50,000–$99,999	34.2	32.3
$100,000–$149,999	18.3	15.0
$150,000–$199,999	7.7	11.5
$200,00 and above	5.5	8.8
Household size
1 person	26.5	24.3
2 people	43.2	38.9
3 people	13.2	15.0
4 people	11.0	15.5
5 or more people	5.9	6.2
Occupation
Management occupations (e.g., business, finance, and administration)	11.3	10.7
Natural and applied sciences and related occupations	5.4	6.2
Health occupations	8.5	6.7
Education, law and social, community and government services	8.9	10.2
Art, culture, recreation and sport related occupation	2.5	3.1
Sales, and service occupations	25.2	26.7
Trades, transport and equipment operators and related occupations	17.7	14.2
Natural resources, agriculture and related production occupations	3.1	3.6
Occupations in manufacturing and utilities	3.2	3.6
Other	14.3	15.1

Data Description

Figure 1 illustrates the proportion of respondents engaging in online grocery shopping, online food orders, and working from home a few times a week/everyday during the pre-pandemic (i.e., before March 2020), in the early stages of the pandemic (i.e., during March 2020), and in the post-pandemic period. It is evident that a substantial rise in telecommuters was observed during the early stages of the pandemic as compared with the pre-pandemic era. With regard to the post-pandemic situation, more individuals engaged in working from home, showing a continued preference toward telecommuting. One of the reasons for higher share of respondents working from home in the early and later stage of the pandemic is the COVID-19 related measures during this time period. More specifically, in March 2020 the Okanagan region experienced significant lockdowns and social distancing measures forcing workers to adopt working from home strategies. However, in the later stage of the pandemic some of the COVID-19 related restrictions eased but some were still in place such as avoiding all non-essential travel and restricting large gatherings ( 45 ). Moreover, cases with emerging COVID-19 variants were also reported in the region in January 2021 ( 46 ). Such unprecedented scenarios and self-awareness about maintaining physical distances and avoiding exposure to crowds and public spaces led to the adoption of a working from home approach in the region. For online food orders, a sudden increase in consumers was noticed during the early stages of the pandemic as compared with the pre-pandemic period. However, for the post-pandemic situation, a reduction in online food orders is noticed as compared with the later stages of the pandemic, indicating people’s desire to return to dining-in and socializing. Lastly, an increase in online grocery shopping was observed in the early stage of the pandemic as compared with the pre-pandemic period. Furthermore, Figure 2 depicts the distribution of individuals engaging in telecommuting and teleshopping activities in the later stage of the pandemic. Here they reported their frequency of teleactivity participation in the seven days immediately before responding to the survey. The largest participation was noticed for individuals working from home, with almost 25% of the respondents telecommuting five or more days per week. By contrast, 17.3% of the individuals preformed online grocery shopping once a week, whereas the share of individuals performing online food orders at the same rate was 18.5%, as expected.

Figure 1.

Distribution of individuals who engaged in online grocery shopping, online food ordering, and working from home a few times a week/everyday.

Figure 2.

Distribution of individuals who engaged in online grocery shopping, online food ordering, and working from home in the later stages of the pandemic (in the seven days before completing the survey).

Methodology

This study uses the MVOP framework to investigate the post-pandemic frequency of online grocery shopping, online food orders, and working from home. The model deals with three dependent variables measured on an ordinal scale: 0 (never/a few times a year), 1 (a few times a month), and 2 (a few times a week/every day). Let i be an index for individuals (i = 1, 2, …., I) and q be the index for ordered variables (q = 1,2…., Q), where Q is the total number of dependent variables for an individual i. Let the observed level for individual i and variable q be $Y_{\mathrm{iq}}$ , and let the number of ordinal levels for the dependent variables be $K_q.$ Instead of modeling the frequency directly, ordered-response models assume that there is an underlying latent continuous variable Y* depicting online grocery, online food ordering, and working from home propensity with respect to frequency ( 36 ):

$${Y_{\mathrm{iq}}}^{*}={\mathrm{\beta }}_q{X}_{\mathrm{iq}}+{\epsilon }_{\mathrm{iq}},Y_{\mathrm{iq}}=k\mathrm{if}{\mu }_q^k<{Y_{\mathrm{iq}}}^{*}<{\mu }_q^{k+1}$$

(2)

where

${\mathrm{\beta }}_q$ is a vector of parameters associated to $X_{\mathrm{iq}}$ ,

$X_{\mathrm{iq}}$ is a vector of explanatory variables,

${\epsilon }_{\mathrm{iq}}$ is the unobserved error terms, and

${\mu }_q^k$ is the lower bound threshold for frequency category k of the dependent variable q.

In this study, the model allows correlation in the ${\epsilon }_{\mathrm{iq}}$ terms across the variables q for each individual i. ${\epsilon }_i$ is multivariate normally distributed if ${\epsilon }_i=({\epsilon }_{i1},{\epsilon }_{i2},\dots ..{\epsilon }_{\mathrm{iQ}}),$ with a mean vector of zero and correlation matrix $\sum$ . The off-diagonal terms of $\sum$ capture the error covariance across the underlying latent continuous variables of the different ordinal variables which can be formulated as ( 49 ):

$$\sum =\left[\begin{array}{ccc}1\hfill & {\rho }_{12}\hfill & {\rho }_{13}\hfill \\ \multicolumn{1}{c}{{\rho }_{21}}& 1\hfill & {\rho }_{23}\hfill \\ \multicolumn{1}{c}{{\rho }_{31}}& {\rho }_{32}\hfill & 1\hfill \end{array}\right]$$

(3)

They account for the effect of common unobserved factors influencing the propensity of each teleactivity engagement. This study also used the simultaneous equation modeling approach to test the endogeneity as exhibited in Kim and Wang ( 38 ). In this regard, simultaneous endogenous effect can be determined as follows:

$${Y_{i1}}^{*}=\mathrm{\beta }{\text{'}}_1{Y_{i2}}^{*}+{\mathrm{\beta }}_1{X}_{i1}+{\epsilon }_{i1}$$

(4)

$${Y_{i2}}^{*}=\mathrm{\beta }{\text{'}}_2{Y_{i1}}^{*}+{\mathrm{\beta }}_2{X}_{i2}+{\epsilon }_{i2}$$

(5)

Further, the parameter vector of the multivariate ordered probit model can be represented as:

$$\delta =({\mathrm{\beta }}_1,{\mathrm{\beta }}_2,\dots {\mathrm{\beta }}_Q;{\mu }_1,{\mu }_2,\dots .,{\mu }_Q;R)$$

(6)

where ${\mu }_q=({\mu }_q^1,{\mu }_q^2,\dots .,{\mu }_q^{K_q})$ for q = 1, 2,…., and Q and R denote the off-diagonal terms of the error correlation matrix. Assume the actual observed response level for individual i and dependent variable q is $m_{\mathrm{iq}}.$ Therefore, the likelihood function (L) for an individual can be represented as:

$$L_i\left(\delta \right)=\mathrm{Pr}(Y_{\mathrm{iq}}=m_{i1},y_{i2}=m_{i2},\dots .,y_{\mathrm{iQ}}=m_{\mathrm{iQ}})$$

(7)

This study accounts for the correlation by estimating the MVOP model through CMP module in Stata 15.0 ( 50 ). The estimation of the model requires a simulation approach using the Geweke–Hajivassiliou–Keane (GHK) simulator ( 51 ). In this context, 200 Halton draws are used in the GHK simulator ( 52 ). Lastly, the descriptive statistics of the variables retained in the final model are provided in Table 2.

Table 2.

Descriptive Statistics of the Explanatory Variables

Variables	Definition	Mean/distribution	SD
Previous online activity engagement frequency
Pre-pandemic × Early stage of pandemic × Later stage of pandemic
Online grocery shopping
Never × Participation × Participation	If individual never participated before pandemic, but participated in early and later stages of pandemic = 1, else 0	4.56%	na
Participation × Participation × Participation	If individual participated before and in early and later stages of pandemic = 1, else 0	18.27%	na
Online order of food
Participation × Participation × Participation	If individual participated before and in early and later stages of pandemic = 1, else 0	31.97%	na
Working from home
Never × Participation × Participation	If individual never participated before pandemic, but participated in early and later stages of pandemic = 1, else 0	11.67%	na
Participation × Participation × Participation	If individual participated before and in early and later stages of pandemic = 1, else 0	25.88%	na
Previous online activity engagement experience
Sporadically website is difficult to navigate	If individual sporadically faces difficulty in website navigation = 1, else 0	31.97%	na
Working from home is highly productive	If individual finds working from home highly productive = 1, else 0	11.16%	na
Always experience poor internet	If individual always experiences poor internet = 1, else 0	35.53%	na
Sociodemographic attributes
Age 18 to 34	If individual’s age is 18 to 34 = 1, else 0	23.35%	na
Age 40 to 54	If individual’s age is 40 to 54 = 1, else 0	31.97%	na
Age 65 or above	If individual’s age is 65 or above = 1, else 0	12.69%	na
Annual household income $50,000–$80,000 × Single_Detached_Dwelling	If individual’s annual household income $50,000–$80,000 and lives in a single detached dwelling = 1, else 0	6.59%	na
Annual household income $80,000–$100,000 × Gender_Female	If individual’s annual household income $80,000–$100,000 and is a female = 1, else 0	10.15%	na
Household size of 2 children or more	If a household has more than 2 children = 1, else 0	5.07%	na
Education level_Bachelors_and_Above × Occupation_Management	If individual’s education level is bachelor’s or above, and works in management related occupations = 1, else 0	3.04%	na
Education level_Bachelors_and_Above × Occupation_Natural_and_Applied_Science	If individual’s education level is bachelor’s or above, and works in natural and applied science related occupations = 1, else 0	5.58%	na
Accessibility to travel modes and built environment
No vehicle in a household	If there is no vehicle in a household = 1, else 0	4.56%	na
Own vehicle × own house	If individual owns a house and vehicle = 1, else 0	69.03%	na
Land-use index	Land-use mix index within a 500 m buffer from the home location	0.47	0.24
Length of sidewalk (in km)	Available length of sidewalk within 500 m buffer of home location	5.58	4.15

Note: SD = standard deviation; na = not applicable.

Model Results

This section provides a brief discussion of the goodness-of-fit measures considered in the study and the parameter estimation results for the proposed model. Table 3 provides parameter estimation results of the separate univariate models and the proposed joint model. It can be clearly observed that there is not much difference in the relationship between explanatory and dependent variables for both independent and joint models. However, the joint model provides additional information through error correlation structure and captures endogeneity. To empirically demonstrate that the proposed approach with error correlation outperforms the model assuming zero correlation across error terms, a likelihood ratio (LR) test is conducted. The LR value can be calculated as follows:

$$\mathrm{LR}=2(L{L}_{\mathrm{MVOP}}-L{L}_{O{P}_1}-L{L}_{O{P}_2}-L{L}_{O{P}_3})$$

(8)

where $L{L}_{\mathrm{MVOP}}$ is the log likelihood of the proposed MVOP model, and $L{L}_{O{P}_1},L{L}_{O{P}_2},L{L}_{O{P}_3}$ represent the log-likelihood values of the independent ordered probit models corresponding to three dependent variables. The LR value is found to be 31.50 which is significantly higher than the critical chi-squared value for five degrees of freedom at a 99% level of confidence. This confirms that there is a need to adopt a joint equation modeling system to model the preferences toward online grocery shopping, online food orders, and working from home, while accounting for correlation across the error terms. Therefore, the joint model is used for further discussion. Moreover, to test for multicollinearity among the variables retained in the final model, a variance inflation factor (VIF) is used. In this regard, values ranging from 0 to 5 depict no significant collinearity, while significant collinearity exists if the value is above 5 ( 53 ). The VIF results confirm that no such collinearity exists among the variables retained in the final model, as the maximum VIF value was found to be 2.02. Lastly, this study extensively tests the impact of past teleactivity participation behavior and experiences along with sociodemographic, access to travel modes, and built environment attributes.

Table 3.

Parameter Estimation Results of Joint and Independent Model

		Joint model			Independent models
		Post-pandemic online grocery shopping	Post-pandemic online ordering of food	Post-pandemic working from home	Post-pandemic online grocery shopping	Post-pandemic online ordering of food	Post-pandemic working from home
Explanatory variables	Definition	Coefficient (t-stat)			Coefficient (t-stat)
Previous online activity engagement frequency
Pre-pandemic × Early stage of pandemic × Later stage of pandemic
Online grocery shopping
Never × Participation × Participation	If individual never participated before the pandemic, but participated in early and later stages of pandemic = 1, else 0	1.17 (2.71)	na	na	1.26 (2.47)	na	na
Participation × Participation × Participation	If individual participated before and in early and later stages of pandemic = 1, else 0	1.82 (4.78)	1.25 (3.42)	na	2.31 (7.30)	na	na
Online ordering of food
Participation × Participation × Participation	If individual participated before and in early and later stages of pandemic = 1, else 0	na	1.34 (4.64)	1.03 (4.55)	na	1.00 (3.14)	1.09 (4.24)
Working from home
Never × Participation × Participation	If individual never participated before pandemic, but participated in early and later stages of pandemic = 1, else 0	na	na	0.20 (1.10)	na	na	0.14 (0.42)
Participation × Participation × Participation	If individual participated before and in early and later stages of pandemic = 1, else 0	na	na	0.39 (1.76)	na	na	0.59 (1.93)
Previous online activity engagement experience
Sporadically website is difficult to navigate	If individual sporadically faces difficulty in website navigation = 1, else 0	0.98 (3.36)	na	na	0.83 (3.35)	na	na
Working from home is highly productive	If individual finds working from highly productive = 1, else 0	na	na	1.08 (3.00)	na	na	1.59 (4.08)
Always experience poor internet	If individual always experiences poor internet = 1, else 0	na	−0.13 (−0.57)	na	na	−0.28 (−1.03)	na
Sociodemographic attributes
Age 18 to 34	If individual’s age is 18 to 34 = 1, else 0	−0.56 (−1.92)	0.70 (2.37)	na	−0.19 (−0.67)	1.05 (3.27)	na
Age 40 to 54	If individual’s age is 40 to 54 = 1, else 0	na	−0.57 (−1.88)	−0.52 (−2.16)	na	−0.40 (−1.40)	−0.57 (−2.08)
Age 65 and above	If individual’s age is 65 and above = 1, else 0	na	−1.06 (−2.03)	na	na	−1.26 (−2.03)	na
Annual household income $50,000–$80,000 × Single_Detached_Dwelling	If individual’s annual household income is $50,000–$80,000 and lives in a single detached dwelling = 1, else 0	1.56 (2.82)	1.17 (2.76)	na	0.62 (1.08)	1.17 (2.68)	na
Annual household income $80,000–$100,000 × Gender_Female	If individual’s annual household income is $80,000–$100,000 and is a female = 1, else 0	na	0.53 (1.46)	0.62 (1.90)	na	0.10 (0.30)	0.81 (1.95)
Household size of 2 children or more	If a household has more than 2 children = 1, else 0	na	2.61 (3.02)	2.83 (5.56)	na	2.64 (4.04)	1.65 (3.36)
Education_level_Bachelors_and_Above × Occupation_Natural_and_Applied_Science	If individual’s education level is bachelor’s or above, and works in natural and applied science related occupations = 1, else 0	na	2.03 (2.96)	1.98 (3.32)	na	1.08 (2.32)	6.59 (0.02)
Education_level_Bachelors_and_Above × Occupation_Management	If individual’s education level is bachelor’s or above, and works in management related occupations = 1, else 0	na	na	1.42 (2.89)	na	na	6.33 (0.01)
Accessibility to travel modes and built environment					na	na	na
No vehicle in a household	If there is no vehicle in a household = 1, else 0	na	0.93 (1.83)	1.05 (2.27)	na	0.13 (0.32)	2.06 (3.38)
Own vehicle × Own house	If individual owns a house and vehicle = 1, else 0	−0.57 (−2.13)	na	na	−0.59 (−2.27)	na	na
Land-use index	Land-use mix index within a 500 m buffer from the home location	na	−0.94 (−1.67)	na	na	−1.37 (−2.32)	na
Length of sidewalk within 500m of home location (in km)	Available length of sidewalk within 500 m buffer of home location	na	na	−0.03 (−1.67)	na	na	−0.06 (−2.16)
Endogenous variable
Future working from home	na	0.79 (4.23)	−0.37 (−1.77)	na	na	na	na
Threshold parameters
Threshold 1	na	2.00 (4.45)	0.54 (1.78)	0.86 (4.48)	1.45 (5.66)	0.42 (1.30)	0.87 (3.94)
Threshold 2	na	3.74 (5.15)	2.02 (4.93)	1.04 (5.31)	2.78 (8.50)	2.00 (5.40)	1.08 (4.90)
Error correlation
Post-pandemic online grocery shopping and post-pandemic online of food				0.24 (1.35)	na	na	na
Post-pandemic online grocery shopping and post-pandemic working from home				−0.57 (4.24)	na	na	na
Post-pandemic online of food and post-pandemic working from home				0.47 (2.09)	na	na	na
Goodness-of-fit measures
Log-likelihood value at convergence value for joint model				−261.41	na	na	na
Log-likelihood value at convergence value for independent models				na	−83.34	−99.42	−94.40

Note: na = not applicable.

Previous Online Activity Engagement Frequency and Experiences

The estimated MVOP model incorporates the effect of previous teleactivity frequencies and experiences on the post-pandemic preferences. For instance, individuals who never participated in online grocery shopping before the pandemic but started participating in the early and later stages of the pandemic are more likely to engage in online grocery shopping after the pandemic. This finding suggests that individuals with longer exposure to online grocery shopping during the initial stages of the pandemic are more likely to continue performing them after the pandemic as well. This indicates their long-term behavior and familiarity with online grocery shopping encourages them to continue doing it even after the pandemic. Furthermore, individuals who participated in online grocery shopping before the pandemic and in the early and later stages of the pandemic revealed a higher propensity toward online grocery shopping after the pandemic. This shows the effect of long-term behavior and familiarity with the ICT usage making them predominant volunteers of online grocery shopping even after the pandemic. Interestingly, engagement in online grocery shopping throughout the three timeframes showed a higher magnitude as compared with participation in just the early and later stages. This reaffirms that consumers with longer exposure to teleactivities are likely to continue to shop online in the future at a higher rate as compared with consumers with recent exposure. Similar results were observed for online food ordering and working from home, reaffirming that individuals with higher exposure to performing these activities are likely to continue performing them even in the post-pandemic world.

In the case of cross-effects, results reveal that individuals who participated in online grocery shopping before the pandemic and in the early and late stages of the pandemic had a positive relationship with online food ordering after the pandemic. Similarly, individuals exposed to online food ordering in the past are more likely to work from home in a post-pandemic world. With regard to experiences related to teleshopping and telecommuting, individuals who only sporadically faced difficulty in website navigation are more likely to continue online grocery shopping. On the other hand, individuals with a poor internet connection revealed a lower propensity toward online food ordering, reaffirming the importance of ICT for online activity engagement ( 54 ). Lastly, results showed that those employees who were extremely productive working from home during the pandemic are more likely to continue working remotely. This reaffirms that working from home may boost their productivity by reducing family–work conflicts, social isolation, and distracting environment ( 35 ).

Sociodemographic Attributes

This study confirms the effects of sociodemographic attributes such as gender, age, education, household income, household size, and dwelling type. The results reveal that younger individuals (18–34 age group) have a higher propensity toward online food ordering, which is consistent with the past literature ( 55 ). There is an understanding that younger individuals are more tech savvy and familiar with ICT usage that links them with more frequent teleactivity participation. Interestingly, younger individuals still show a lower propensity to grocery shopping online in the post-pandemic context, which may seem counter-intuitive based on the existing literature ( 56 – 58 ). This may be related to the unavailability of cost-effective grocery delivery services in the study area compared with other parts of the world. For example, additional costs such as annual subscription fees, service charges, and delivery fees associated with available delivery services in the region leads to higher grocery bills which might not be applicable in metropolitan cities ( 59 ). This additional cost burden may deter younger individuals from frequent online grocery shopping. Furthermore, middle-aged individuals (38, 40–53) revealed lower preferences for working from home and online food ordering after the pandemic, indicating that the adoption of this new lifestyle of teleactivities at this age adds time constraints which reduce their overall productivity ( 10 , 11 ). Similarly, older individuals revealed a lower preference for online food ordering in the post-pandemic period. This might be a result of their unfamiliarity with ICT usage and online food orders eliminating the opportunity to socialize with people, which is an essential component of one’s life. In the case of education and occupation, individuals with higher education (bachelor’s or above) working in the natural and applied science sector revealed a higher propensity toward working from home for the post-pandemic period. Similar findings were observed for individuals working in the management sector with a bachelor’s degree or above. This is presumably linked to their familiarity and adaptability with ICT uses or perhaps the effect of flexibility associated with their job given their high income.

Among household-level attributes, individuals with an annual household income of $50,000–$80,000 living in single detached dwellings showed a higher propensity toward online grocery shopping and online food ordering after the pandemic. This reaffirms an understanding that medium-income groups residing in single detached suburban areas have higher expenditure capabilities and that living farther from grocery stores and restaurants encourages them to engage in these activities remotely ( 60 ). Interestingly, females residing in households with an annual income between $80,000 and $100,000 are more likely to work from home in the post-pandemic period. This reflects that females residing in medium/high income households take higher responsibility for household-related activities such as childcare and prefer jobs that are flexible in nature, allowing them to work from home. Similarly, they revealed a positive relationship with online food ordering in the post-pandemic period, which might be again a result of household-related responsibilities and preferences toward socializing among family/friends at home. Household size was found to be a significant factor affecting post-pandemic choices. For instance, it was determined that individuals living in households with a higher number of children are more likely to engage in online food ordering and working from home. This finding shows that households with more children prefer to eat at home together and socialize among themselves and it can be very challenging to make trips with younger ones ( 39 ). Moreover, their propensity toward working from home might be a result of childcare responsibilities such as drop/pickup from daycare/schools encouraging one to choose the flexible option of telecommuting.

Accessibility to Travel Modes and Built Environment

Access to tools such as a driver’s license or transit pass and household vehicles that provide an opportunity to travel is referred to as mobility tool ownership ( 61 ), allowing individuals to travel with their preferred mode. Access to these transport-related resources can have a significant impact on teleshopping and telecommuting behavior. For instance, vehicle owners residing in owned (non-rented) dwellings exhibited a lower propensity for online grocery shopping after the pandemic. This might reflect the travel flexibility associated with transport-related resources, which allows them to choose their preferred mode to perform an activity. Interestingly, the absence of a vehicle in a household encourages working from home even after the pandemic. This suggests that vehicle deficient individuals prefer to choose employment positions that are flexible in nature (allowing working from home) ( 62 ). Similarly, results were observed for online food ordering. Among built environment variables, the land-use index depicts the share of residential, commercial, and other areas, which includes industrial, in the neighborhood. This index ranges from a value of 0 to 1, where values closer to 0 represent homogeneous land mix and values closer to 1 represent heterogeneous land mix. Interestingly, a negative relationship is observed between the land-use index and online food ordering. This suggest that individuals residing in urban areas are often equipped with multi-modal travel opportunities, well-connected active transportation networks, and easy access to transit stops, which make it easier to travel to restaurants closer to their residences. Similarly, individuals with a higher available length of sidewalk within 500 m of their home location are less likely to work from home, which again indicates the effect of urban characteristics.

Endogeneity Effects and Error Correlation

This study reveals a complex inter-relationship between telecommuting and teleshopping choices. Looking at endogenous effects, the preference for working from home post-pandemic positively influences the preference for online grocery shopping after the pandemic. This indicates that decisions of one life domain such as working from home can influence other domain decisions such as online grocery shopping. Given the adoption of working from home, individuals are more likely to spend time at home and reduce daily mobility, which has a spillover effect on other daily travel decisions such as grocery shopping. Past literature has suggested that such individuals who are spending more time at home appreciate the convenience of online shopping (e.g., scheduled delivery, low exposure to public spaces, ability to purchase high-demand items during such crises, and avoiding long queues at stores) and therefore prefer online grocery shopping ( 13 ). On the contrary, a negative relationship was observed for online food orders. This could be triggered by their willingness to “go out” to compensate for increased sedentary tendencies that come with working from home and performing other activities such as grocery shopping remotely. Such individuals consider dining-in at a restaurant as an opportunity to travel outside and socialize with friends/family/colleagues or prefer cooking at home instead. However, other endogenous relationships such as the effect of online food ordering after the pandemic on post-pandemic online grocery shopping and vice versa turned out to be statistically insignificant and was dropped from the final model estimation.

Furthermore, this study accounts for unobserved factors that may jointly influence post-pandemic decisions such as frequency of online grocery shopping, online food orders, and working from home. The error correlation terms are found to be statistically significant, showing that it is appropriate and more efficient to estimate the three equations jointly instead of developing them separately. A positive error correlation between post-pandemic online grocery shopping and post-pandemic online food ordering reveals that unobserved attributes that increased the propensity to engage in online grocery shopping are positively correlated with unobserved attributes that increased the propensity to engage in online food ordering. Similarly, post-pandemic online food orders from restaurants and post-pandemic working from home revealed a positive error correlation term. On the contrary, a negative relationship is revealed between online grocery shopping and working from home. This indicates that the unobserved factors have an inverse relationship, which suggests unobserved factors that increase the propensity for online grocery shopping decrease the propensity toward working from home, and vice versa.

Marginal Effects

To further understand the impact of various attributes retained in the MVOP model on telecommuting and teleshopping choices, marginal effects were estimated (Table 4). For instance, the variable indicating long-term past exposure to online grocery shopping shows an impact of 16% on post-pandemic online grocery shopping (for a few times a month). This suggest that individuals’ long-term exposure to online grocery shopping increases the probability of participating in online grocery shopping after COVID-19 by 16%. Similar results were observed for online food orders, where past exposures increase the probability to perform them at least few times a month post-pandemic by 11%. In addition, working from home pre-pandemic and in the early and later stages of the pandemic increases the probability of working from home post-pandemic a few times a week/everyday by 2%. Furthermore, older individuals revealed an 11% reduction in their probability of engaging in online food ordering a few times a week/every day after the pandemic, as expected. Interestingly, females living in households with an annual household income of $80,000–$100,000 show an increase of 8% in their probability of working from home in the post-pandemic era a few times a week/everyday. The presence of a higher number of children in a household increases the probability of performing online food orders and working from home a few times a week/everyday by 16% and 4% respectively, in the post-pandemic era. A similar lack of vehicles in a household increases the propensity to work from home after the pandemic a few times a week/everyday by 5%. Lastly, urban characteristics such as the presence of mixed land use and access to sidewalks have a substantial impact on telecommuting and teleshopping choices in the post-pandemic context.

Table 4.

Marginal Effects of the Variables Retained in the Multivariate Ordered Probit Model

Variables	Post-pandemic online grocery shopping			Post-pandemic online ordering of food			Post-pandemic working from home
	Never/few times a year	Few times a month	Few times a week/every day	Never/few times a year	Few times a month	Few times a week/every day	Never/few times a year	Few times a month	Few times a week/every day
Never × Participation × Participation_ Online Grocery Shopping	−0.16***	0.10***	0.06***	na	na	na	na	na	na
Participation × Participation × Participation_ Online Grocery Shopping	−0.26***	0.16***	0.09***	−0.28***	0.14***	0.13***	na	na	na
Participation × Participation × Participation_ Online Ordering of Food	na	na	na	−0.21***	0.11***	0.10***	−0.01**	0.0001	0.01*
Never × Participation × Participation_ Working from Home	na	na	na	na	na	na	0.03**	−0.0002	−0.03*
Participation × Participation × Participation_ Working from Home	na	na	na	na	na	na	−0.02*	0.0001	0.02*
Sporadically website is difficult to navigate	−0.14***	0.08***	0.05***	na	na	na	na	na	na
Working from home is highly productive	na	na	na	na	na	na	−0.008	−0.0001	0.008
Always experience poor internet	na	na	na	0.03***	−0.01***	−0.01***	na	na	na
Age 18 to 34	0.08***	−0.05***	−0.03***	−0.15***	0.08***	0.07***	na	na	na
Age 40 to 54	na	na	na	0.08***	−0.04***	−0.03***	−0.02**	0.0001	0.02*
Age 65 and above	na	na	na	0.23***	−0.12***	−0.11***	na	na	na
Annual household income $50,000–$80,000 × Single_Detached_Dwelling	−0.22***	0.13***	0.08***	−0.26***	0.13***	0.12***	na	na	na
Annual household income $80,000–$100,000 × Gender_Female	na	na	na	−0.06***	0.03***	0.03***	−0.08***	0.0004	0.08***
Household size of 2 children or more	na	na	na	−0.34***	0.18***	0.16***	−0.04***	0.0002	0.04**
Education level_Bachelors_and_Above × Occupation_Natural_and_Applied_Science	na	na	na	−0.28***	0.15***	0.13***	−0.04**	0.0002	0.04**
Education level_Bachelors_and_Above × Occupation_Management	na	na	na	na	na	na	−0.11***	0.0006	0.11**
Own vehicle × Own house	0.08***	−0.05***	−0.03***	na	na	na	na	na	na
No vehicle in a household	na	na	na	−0.12***	0.06***	0.05***	−0.05**	0.0003	0.05**
Land-use index	na	na	na	0.21***	−0.11***	−0.10***	na	na	na
Length of sidewalk within 500m of home location (in km)	na	na	na	na	na	na	−0.04***	0.0002	0.04***

Note: * 90% confidence level; ** 95% confidence level; *** 99% confidence level; na = not applicable.

Discussion

The findings of the study clearly suggest that unprecedented scenarios like the COVID-19 pandemic and recent developments in ICT have undoubtedly affected activity-travel behavior and choices. For instance, the findings of the study suggest that individuals are likely to continue engaging in teleshopping in the future. This may have benefits and present challenges that cities need to prepare for. For example, in the case of benefits, increased teleactivities may indicate a reduction in personal travel (if teleactivities replace in-person activities) which can reduce traffic congestion and traffic-related emissions. However, some studies have shown a complementary effect of teleactivities, indicating an increase in activities rather than reduction ( 63 ). In addition, depending on the delivery mode, online shopping may increase freight vehicles on the road. So, increased teleshopping may result in an overall increase in traffic and related emissions. Keeping this in mind, it is necessary to recognize the possible benefits and challenges associated with online shopping in the planning and policymaking process. To do so, one of the major requirements is collecting more data on teleshopping including the collection methods of online deliveries such as pickup and delivery, and their travel modes. Such data could be useful to update travel demand models including expanding their scope to include freight vehicles resulting from online shopping. Such models will allow the planning process to reflect the future travel demand more accurately and incorporate policies that address the wholesome effects of teleshopping, associated freight travel, and in-person shopping travel, and consequently create more sustainable transport systems in the region.

Moreover, the study suggests that individuals exhibited a strong desire to frequently work from home after the pandemic. In this regard, the findings of the study also suggest that individuals finding telecommuting highly productive are likely to do it more frequently. This implies that employees may have better control over their time (e.g., savings in time spent commuting) and can sustain a better work–life balance when telecommuting, resulting in more productivity in work. In addition to travel time savings, telecommuters may be able to cut down their expenses such as fuel and parking costs. Although many employees find telecommuting to be beneficial, it is not without challenges. For instance, working from home limits interaction with colleagues resulting in challenges associated with new member integration, team building, collaboration, and working together on complex tasks. From an employer’s perspective, telecommuting may allow them to cut down on their expenses including office space requirements. On the other hand, it may present challenges such as arrangements associated with remote working, supervision and monitoring of working hours, and performance ( 64 ). Moreover, telecommuting also has an impact on the local economy such as office building rentals, transit ridership, and restaurants and coffee shops near workplaces.

As we come out of this pandemic and with ICT continuing to develop to address many challenges associated with telecommuting, we can see a preference for hybrid working arrangements such as telecommuting for a few days in the week. The collected data also show a preference for telecommuting for a few days a week after the pandemic; for example, 60.2% of the respondents revealed a preference for telecommuting a few times a week, whereas the share of individuals willing to work from home every day was only 26.2%. In this regard, more data collection is required to continue monitoring the preferences and changes. If such flexible working arrangements continue, it is important to recognize that they may present transportation planning and policymaking challenges. For example, one of the fundamental elements for transportation planning is the travel survey, which traditionally collects data for a weekday assuming that all weekdays may represent the same demand/behavior. Given the flexible working arrangement, a typical weekday may not exist anymore. Rather, it may be a typical work week, demonstrating the need for weeklong data collection. Moreover, given the close interaction between in-home/teleactivities and in-person activities such as their substitution and complementary relationship, future data collection needs to collect time use information on how individuals spend time at home and outside. Such data will be useful to develop next-generation travel demand models encompassing both in-home and out-of-home activities, and predict travel demand with higher accuracy. In addition to employee data, future efforts should also focus on collecting data from the employers to understand their perspectives.

Several more inferences can be drawn from this study to increase the adoption of teleactivities after the pandemic. For example, enhancing consumer experience with website navigation and promoting digitalization/ICT usage among consumers of all age groups and business sizes are some of the ways of accomplishing this. Furthermore, for socioeconomic attributes, results reveal that individuals with vehicle ownership are less likely to engage in online grocery shopping after the pandemic. This indicates that these individuals use travel flexibility associated with vehicle ownership. Therefore, future policies and investments are required to improve the level-of-service (LOS), particularly investing in alternative transportation infrastructures such as walking and cycling facilities. This could significantly improve the LOS of roads as it is likely to attract individuals to use alternative travel modes and will free up space in traffic lanes. In a similar line of investigation related to land-use attributes, results suggest that individuals residing in urban areas are more likely to dine-in at a restaurant. Therefore, policies need to target mixed land-use developments offering residential and commercial establishments to attract individuals for both living and engaging in activities at close-by destinations.

Conclusions

This study investigates individuals’ preferences toward teleshopping and telecommuting in the post-pandemic era. In particular, it investigates the frequency of online grocery orders, online food orders, and working from home. A MVOP framework is adopted to account for the correlation and endogenous effects among the three choices of teleactivities. This study uses data from a web-based survey conducted in the Central Okanagan region of British Columbia, Canada. Evidence from this study provides essential input on how post-pandemic activity-travel patterns will shape up and to what extent travel behavior before and during the pandemic influences them. In addition, this study also tests the effect of sociodemographic features, accessibility to travel modes, and built environment attributes.

Furthermore, the endogenous variable effect is one of the most important findings of this study. Results reveal that one life domain decision made by an individual, such as working from home in the post-pandemic era, will positively influence another life domain decision such as online grocery shopping. On the other hand, working from home in the post-pandemic era revealed a negative relationship with online food ordering after the pandemic. Through this line of investigation, we confirmed the hypothesis that individuals are likely to continue engaging in online grocery shopping, online food ordering, and working from home even after the pandemic. For example, results suggest that individuals engaging in teleactivities before the pandemic and during its early and late stages are more likely to continue performing them online as compared with individuals with recent exposures. Furthermore, for socioeconomic attributes, results reveal that individuals with higher annual household incomes, the presence of children in a household, and higher education levels are more likely to engage in teleactivities. However, results were not the same for older individuals, individuals who own a vehicle, and urban dwellers.

Nevertheless, this study has certain limitations. Modeling with such a small sample size is challenging and might affect the parameter estimation results. However, the exercise followed in the study and statistically significant model results provide behaviorally plausible insights into the activity-travel behavior changes linked with the post-pandemic context. Furthermore, this study serves as a baseline for future studies to compare evolving activity-travel behavior of individuals in the post-pandemic context. However, one of the agendas for future studies is to achieve comprehensive data collection for such modeling purposes which is often challenging. In this line of investigation, variables measuring accessibility to different activity points (e.g., distance to restaurants/shopping centers/grocery stores from home location) were also tested. However, they were dropped from the final model because of statistical insignificance. This may reflect the constraints associated with low sample size. Future studies could therefore concentrate on data collection and test extensively the effect of built environment and land-use measures on preferences for teleactivities. Moreover, the survey was distributed through Facebook. The adoption of convenience sampling and reliance on social media, may have introduced some bias into the data. For example, the sample, over-represents younger individuals and under-represents older individuals, indicating that the participants of the survey may be more technology savvy and skilled in ICT usage such as web-browsing, smartphones, and computers, among others. This may depict respondents’ higher inclination toward teleactivities, and one may interpret results accordingly. Therefore, readers should interpret the results with caution. In addition, the data used in this study were collected during late 2020 and early 2021. Although the timing of survey might have a limited impact on the survey responses as at this time period some of the COVID-19 related measures were lifted in the region. There is a possibility that individuals’ attitudinal preferences and perceived risks toward activity-travel participation may have changed especially in the post-pandemic context because of availability of vaccines, higher vaccination rates, lower transmission rates and the easing of COVID-19 related measures. To capture such preferential changes more effort is required to collect data that will provide information about post-pandemic activity-travel choices in future. Keeping this in mind, currently an activity-travel survey is being administered in the region by our research group to capture such choices. Overall, the findings of this study provide insights into how teleshopping and telecommuting will evolve in the post-pandemic world and assist in developing policies to promote sustainable goals for communities.