Table 2.

Candidate explanatory variables.

Label	Definition	Related studies
Area	1: Prefectures where the state of emergency was declared on April 7 0: Prefectures where the state of emergency was declared on April 13	The environmental conditions may be different between areas 1 and 0.
Age	Age (log)	The possibility of a digital divide due to age has been discussed by Ramsetty and Adams (2020), and age has also been identified by Elena-Bucea et al. (2020) as one of the main factors in information technology adoption.
Gender	1: Female 0: Male	Elena-Bucea et al. (2020) also identified gender as a factor in information technology adoption. Leung, Sharma, Adithipyangkul, and Hosie (2020) examined the macrolevel relationship between gender gap and COVID-19 infections.
Job	1: employed 0: homemaker	Workers may have changed their information technology usage behavior due to remote work (Venkatesh, 2020; Neely 2020).
FamIncome	Family (household) income	Individual income and economic status can influence behavior related to usage of digital devices (Beaunoyer et al. 2020; Elena-Bucea et al., 2020).
DispIncome	Individual disposable income
PerIncome	Personal income
Edu	Number of years of education (9: junior high school, 12: high school, 16: university (college), 18: graduate school)	Kim et al. (2020) examined the relationship between social class and the impact of COVID-19. In addition, Elena-Bucea et al. (2020) found that education is one of the factors that affects the adoption of information technology.
FamSize	Number of family members	In Campbell (2020), cases of domestic violence under the influence of COVID-19 were examined, and whether a person lives with a family or alone, and the number of family members, is thought to affect behavioral change.
Child	0: no children 1: with child(ren)	Consumer behavior is altered when children's schools are closed (Maria et al., 2020). The presence or absence of children and the age of the children are also thought to affect behavioral changes based on studies that considered the effects of COVID-19 on children (Van Lancker & Parolin, 2020).
YngChild	Age of the youngest child (if there are no children or the child is over 20 years old, the value is 20)
House	0: Rental house 1: Own house	Home ownership is one indicator of economic status and can also be interpreted as a factor explaining residential location.
Car	0: Do not own cars 1: Car owner	Consumers who own their own cars are assumed to have a lower risk of infection from using public transportation (e.g., Hu et al. 2020), and their behavior is also expected to change.
Use19	Average usage time for all time bands on weekdays during 2019 ($T^{-1}\sum _{z\in 1}^T{\beta }_{i0}^z$).	Frequent users and infrequent users have different knowledge levels and attitudes toward mobile devices. This difference may also be related to differences in behavior change (e.g., Beaunoyer et al. 2020)