A multicountry perspective on gender differences in time use during COVID-19

Significance We find pervasive gender differences in time use during COVID-19. Surveys of diverse samples with over 30,000 respondents reveal that women—especially mothers—spent more time on necessities such as childcare and chores. In turn, time spent completing household chores was linked to lower well-being. This research reveals persistent time-use differences between women and men in household responsibilities during the COVID-19 pandemic.

Note. The symbol "--" indicates that the variable was not assessed in that study. a Across all studies (unless otherwise indicated as omitted), respondents reported their age and gender. b In these studies we recorded household size (i.e. whether respondents live alone or with others in the household). c Respondents were nationally representative in terms of age, gender, ethnicity, and occupation status (i.e., full-time employed) for the US. d Respondents were nationally representative in terms of age, gender, ethnicity, and occupation status (i.e., full-time employed) for Canada. e Respondents were nationally representative in terms of age, gender and ethnicity. Sample size is based on all available data in each sample. 18.31% (10.14) --Note. The symbol "--" indicates that the variable was not assessed in that study. Descriptive statistics for the time-use measures are reported as episode-weighted statistics (i.e. the percentage of time that respondents reported spending on each activity is weighted by the total amount of time they spent in all other activities measured within that sample). a In these studies we measured time-use on work, active, and passive leisure with 1-item. b Overall leisure is a composite of active and passive leisure across all studies. The composites differ per study and the exact items are presents in the tables below c Passive leisure in these studies was measured with 1-item only. d In these datasets, active and passive leisure were measured with 1-item each. Note. N = 440. SD = Standard deviation. In this dataset, we did not record times spent on necessities. Note. N = 840. SD = Standard deviation. In this dataset, we did not record time spent on passive leisure or necessities.  (exercising, meditating, watching TV, reading, relaxing, etc.) 11.43 (9.94) Active leisure % Spending time alone with your partner 7.41 (7.57) % Socializing with others (e.g., talking in-person or virtually with your friends and family) 6.16 (7.46) Necessities % Doing household chores (e.g., preparing meals, doing laundry, cleaning, etc.) 12.13 (9.81) % Taking care of your children (e.g., homeschooling, reading, playing with them, etc.) 23.54 (16.70) Other activities % Sleeping 20.86 (13.71) % Other 1.57 (6.65) Note. SD = Standard deviation. In this dataset, we did not record passive leisure separately from overall leisure.  Note. SD = Standard deviation. a This item can be treated as both active leisure (spending time with family) and a necessity (taking care of family). We treat it as a necessity given that COVID-19 has likely placed additional demands on people's time, including the need to spend more time with family that is not necessarily viewed as leisure time.   Note. SD = Standard deviation. a This item can be treated as both active leisure (spending time with family) and a necessity (taking care of family). We treat it as a necessity given that COVID-19 has likely placed additional demands on people's time, including the need to spend more time with family that is not necessarily viewed as leisure time. Note.Happiness was measured as a single-item question: "Taking all things together, how happy would you say you are? 0 = not at all to 10 = extremely" in Samples 4 to 9; "In general, to what extent do you feel happy these days? 0 = very unhappy to 10 = very happy" in Samples 1 to 3; and "When compared to before the COVID-19 pandemic, how happy are you? 1 = much less happy to 5 = much more happy". Tables S3a to S3e describes the sample specific items that were used to calculate the time-use composites.  Note. These sensitivity analyses illustrate the effect size (Cohen's d) we would be able to detect given each sample size at α = .05. Note. These sensitivity analyses illustrate the effect size (Cohen's d) we would be able to detect given each sample size at α = .05. In sample 3 we collected data among working parents. In Samples 8-9, we collected data among students and did not measure parental status.   .04 -.20 .13 -1.14 .253 Note. CI = confidence interval. PI = prediction interval. Necessities is a composite of chores, caretaking, and/or family time. Overall leisure is a composite of active and passive leisure. Work is a composite or paid work and school work (Samples 8-9). Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, weekly work hours, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, weekly work hours (apart from models with time-use work), education level, number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched. Note: For the necessities (composite) effect, we included Samples 3-7 and Sample 9. For the overall leisure effect, we included Sample 1 and Samples 3-9. For the active leisure effect, we included all samples. For the passive leisure effect, we included Sample 1 and Samples 4-9. While the meta-analytic effect for necessities and overall leisure was significant, it was only for necessities that the prediction interval did not include zero (see Table S17a for details and covariates included).

Meta-Analyses: Gender Differences in Time-Use Across Samples (with Covariates)
Supplementary Information for Time-Use and Subjective Well-Being during COVID-19 | p. 14 Note. Necessities is a composite of chores, caretaking, and/or family time. Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, weekly work hours (apart from models with time-use work), education level, number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched. Note. Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, education level, weekly work hours (apart from models with time-use work), number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched. Note. Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, weekly work hours (apart from models with time-use work), education level, number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched.

Fig. S2d | Meta-Analysis: Gender Differences in Time Spent on Caretaking (Samples 3-4) During COVID-19.
Note. Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, education level, weekly work hours (apart from models with time-use work), number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched.

Fig. S2e | Meta-Analysis: Gender Differences in Time Spent on Caretaking/Family Time (Samples 5-7) During COVID-19.
Note. Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, education level, weekly work hours (apart from models with time-use work), number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched.

Fig. S2f | Meta-Analysis: Gender Differences in Time Spent on Overall Leisure COVID-19.
Note. Overall leisure is a composite of active and passive leisure. Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, weekly work hours (apart from models with time-use work), education level, number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched.

Fig. S2g | Meta-Analysis: Gender Differences in Time Spent on Active Leisure During COVID-19.
Note. Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, weekly work hours (apart from models with time-use work), education level, number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched.

Fig. S2h | Meta-Analysis: Gender Differences in Time Spent on Passive Leisure During COVID-19.
Note. Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, education level, weekly work hours (apart from models with time-use work), number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched.

Fig. S2i | Meta-Analysis: Gender Differences in Time Spent on Paid Work + School Work (Samples 8-9) During COVID-19,
Note. Necessities is a composite of chores, caretaking, and/or family time. Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, weekly work hours (apart from models with time-use work), education level, number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched.   .339 Note. CI = confidence interval. PI = prediction interval. Given that caretaking among parents and non-parents was included only in Sample 4, we could not run a meta-analysis on this individual item as we did with gender differences in time-use Necessities is a composite of chores, caretaking, and/or family time. Overall leisure is a composite of active and passive leisure. Work is a composite or paid work and school work (Samples 8-9). Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, weekly work hours (apart from models with time-use work), education level, number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched.. Supplementary Information for Time-Use and Subjective Well-Being during COVID-19 | p. 26

Fig. S3a | The meta-analytic effect of gender (1=women) and gender by parental status (1=yes) based on models with covariates.
Note. For the necessities (composite) effect, we included Samples 4-7. For the overall leisure effect, we included Sample 1 and 4-8. For the active leisure effect, we included Samples 1-2, 4-7. For the passive leisure effect, we included Sample 1 and 4-7. For the work effect, we included Samples 1-2, 4-7. The most robust metaanalytic effect emerged for necessities among parents as indicated by the confidence and prediction interval that did not cross zero (see Table S19c      Covariates in Sample 4 are: age, household income, employment status, weekly work hours (apart from models with time-use work), household size, education level, number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched.

Meta-Analyses: Gender and Gender X Parental Status Differences in Happiness Across Samples (with Covariates)
Supplementary Information for Time-Use and Subjective Well-Being during COVID-19 | p. 30

Fig. S5 | Gender differences in happiness.
Note. The mega-analyses were conducted without covariates. The meta-analyses and the sample-specific effects are based on models with covariates as follows. Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, weekly work hours (apart from models with timeuse work), education level, number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched.  Note. β = standardized regression beta; se = standard error; r = r-partial. Necessities is a composite of chores, caretaking, and/or family time. Overall leisure is a composite of active and passive leisure. Work is a composite or paid work and school work (Samples 8-9  .312 Note. CI = confidence interval. PI = prediction interval. Necessities is a composite of chores, caretaking, and/or family time. Overall leisure is a composite of active and passive leisure. Work is a composite or paid work and school work (Samples 8-9). Covariates in Sample 1 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 2 are: age, employment status, marital status, number of children, and days since the survey was launched. Covariates in Sample 3 are: age, household income, employment status, marital status, and number of children. Covariates in Sample 4 are: age, household income, employment status, household size, weekly work hours (apart from models with time-use work), education level, number of children, and days since the survey was launched. Covariates in Sample 5-7 are: age, household income, household size, education level, number of children, and days since the survey was launched. Covariates in Sample 8 are: age and days since survey was launched. Covariates in Sample 9 are: age, race, socio-economic status (composite of parental education and income), household size, and days since the survey was launched. Note.We tested the null hypothesis that there is no significant difference between the models i) a random intercept model of the IV predicting DV and ii) a random intercept and slope model of the IV predicting DV. The rejection of the null (p-value < .05) suggests that adding the random slope to the random intercept model improves the fit of the model. Unstructured covariances were used in all random slope models except for the model with active leisure where an independent covariance structure was used due to the convergence issues which are common with unstructured covariances (Boedhoe et al., 2019). Note. CI= confidence interval. Beta estimates represent coefficients in standard deviation units. Pooled data from 9 samples were used in the analysis. Necessities, chores and caretaking were not measured in Samples 1, 2 and 8. Passive leisure was not measured in Samples 2 and 3. Overall leisure is a composite of active and passive leisure. Work is paid work except for Samples 8-9 where it is paid work and schoolwork. Caretaking/family time was measured as 'taking care of family (kids or elderly) in Samples 3 and 4, and 'taking care of others or spending time with family' in Samples 5, 6, and 7.Random slope models were used in mega-analysis of pooled data for necessities, overall leisure, active leisure, passive leisure, chores, caretaking/family time and work. Covariances were unstructured (except for active leisure where an independent covariance was specified).The main effects for necessities remain robust after applying Bonferroni corrections, that is, using an alpha/5 in evaluating the statistical significance of 5 time use models that we preregistered (necessities, overall leisure, active leisure, passive leisure, work). Note. CI= confidence interval. Beta estimates represent coefficients in standard deviation units. Pooled data from 9 samples were used in the analysis. Necessities, chores and caretaking were not measured in Samples 1, 2 and 8. Passive leisure was not measured in Samples 2 and 3. Overall leisure is a composite of active and passive leisure. Work is paid work except for Samples 8-9 where it is paid work and schoolwork. Caretaking/family time was measured as 'taking care of family (kids or elderly) in Samples 3 and 4, and 'taking care of others or spending time with family' in Samples 5, 6, and 7. Random slope models were used in mega-analysis of pooled data for necessities, overall leisure, active leisure, passive leisure, chores, caretaking/family time and work. Covariances were unstructured (except for active leisure where an independent covariance was specified). Estimates for parents=1 or parents=0 were retrieved from the interaction model using the postestimation commands (marginsplot) in Stata 15.1. Note. CI= confidence interval. Beta estimates represent coefficients in standard deviation units. Pooled data from 9 samples were used in the analysis. Necessities, chores and caretaking were not measured in Samples 1, 2 and 8. Passive leisure was not measured in Samples 2 and 3. Overall leisure is a composite of active and passive leisure. Work is paid work except for Samples 8-9 where it is paid work and schoolwork. Caretaking/family time was measured as 'taking care of family (kids or elderly) in Samples 3 and 4, and 'taking care of others or spending time with family' in Samples 5, 6, and 7. The caretaking analysis was restricted to Sample 3 to capture the pure measure of taking care of family. Guided by ikelihood ratio tests, random slopes were used in all models and random intercept model was used when necessities was a predictor.. Covariances were unstructured (except for active leisure where an independent covariance was specified). The main effects for chores remain robust after applying Bonferroni corrections, which involve using an alpha/5 for testing 5 time use models that we preregistered (necessities, overall leisure, active leisure, passive leisure, work) in evaluating the statistical significance. Beta estimates represent coefficients in standard deviation units. Random slope models were used in mega-analysis of pooled data for necessities, chores and taking care of others. Covariances were unstructured. Random intercept model was used for the mega-analysis of happiness. Linear regression analysis was used in sample-specific analaysis. Taking Care of Others V1 was measured by a question that assessed 'taking care of others' in Samples 3 and 4, and a question that assessed 'taking care of others or spending time with family' in Samples 5, 6, and 7. Beta estimates represent coefficients in standard deviation units. Random slope models were used in mega-analysis of pooled data for total leisure, active leisure, passive leisure and work. Covariance was unstructured for total leisure, passive leisure, and work; and independent for active leisure. Linear regression analysis was used in sample-specific analaysis. Taking Care of Others V1 was measured by a question that assessed 'taking care of others' in Samples 3 and 4, and a question that assessed 'taking care of others or spending time with family' in Samples 5, 6, and 7.  Note. CI= confidence interval. Beta estimates represent coefficients in standard deviation units. Pooled data from 9 samples were used in the analysis. Necessities, chores and caretaking were not measured in Samples 1, 2 and 8. Passive leisure was not measured in Samples 2 and 3. Overall leisure is a composite of active and passive leisure. Work is paid work except for Samples 8-9 where it is paid work and schoolwork. Caretaking/family time was measured as 'taking care of family (kids or elderly) in Samples 3 and 4, and 'taking care of others or spending time with family' in Samples 5, 6, and 7. The caretaking analysis was restricted to Sample 3 to capture the pure measure of taking care of family. Guided by ikelihood ratio tests, random slopes were used in all models and random intercept model was used when necessities was a predictor.. Covariances were unstructured (except for active leisure where an independent covariance was specified). The main effects for chores remain robust after applying Bonferroni corrections, which involve using an alpha/5 for testing 5 time use models that we preregistered (necessities, overall leisure, active leisure, passive leisure, work) in evaluating the statistical significance. Note. Age ranged from 18 to 25. Education was measured as the average of 7 educational levels (from some grade school to post-grad degree) for parents 1 and 2. The mean and median of the income was equal. Range for no of cohabitants was 0 to 10 and for no of days since survey launch was 0 to 44. Note. Global happiness was measured with the question: "Taking all things together, how happy would you say you are?" rated from 0 (Not at all) to 10 (Extremely). Positive affect was measured with three positive feelings experienced last week, and negative affect with three negative feelings experienced last week (1 = Very rarely, 5 = Very often/always). Work was captured with 3-items (work, study, commuting). Overall leisure was the sum of active and passive leisure. Active leisure was the sum of time spent praying/worshipping/meditating, socializing, exercising, intimate relations, going outdoors, hobbies. Passive leisure was the sum of time spent watching tv, napping/resting, relaxing, doing nothing. Necessities was measured as the sum of time spent shopping, personal hygiene, preparing food, doing housework. For each composite, an episode weighted statistics was computed where the amount of time that respondents reported spending on each activity is weighted by the total amount of time spent in all measured activities: work, active leisure, passive leisure, necessities and other measured activities (phone/computer use, eating, waiting +other, social media use). Note. Work was captured with 3-items (work, study, commuting). Overall leisure was the sum of active and passive leisure. Active leisure was the sum of time spent praying/worshipping/meditating, socializing, exercising, intimate relations, going outdoors, hobbies. Passive leisure was the sum of time spent watching tv, napping/resting, relaxing, doing nothing. Necessities was measured as the sum of time spent shopping, personal hygiene, preparing food, doing housework For each composite, an episode weighted statistics was computed where the amount of time that respondents reported spending on each activity is weighted by the total amount of time spent in all measured activities: work, active leisure, passive leisure, necessities and other measured activities (phone/computer use, eating, waiting +other, social media use). The weighted timeuse necessities were log transformed as skewness was greater than 3. Covariates were: age (range = 18 to 25), gender (female=1, male=0), race (dummy variables for being White, Black, Asian, Hispanic and Other), SES (education and income were entered as a composite or individually depending on the model), number of days since survey launch, number of co-habitants (range = 0 to 10).

Table S12d
Regression analyses predicting differences in SWB at Time 1 by socio-demographic status (Sample 9) Model 1 SWB Note. Work was captured with 3-items (work, study, commuting). Overall leisure was the sum of active and passive leisure. Active leisure was the sum of time spent praying/worshipping/meditating, socializing, exercising, intimate relations, going outdoors, hobbies. Passive leisure was the sum of time spent watching tv, napping/resting, relaxing, doing nothing. Necessities was measured as the sum of time spent shopping, personal hygiene, preparing food, doing housework For each composite, an episode weighted statistics was computed where the amount of time that respondents reported spending on each activity is weighted by the total amount of time spent in all measured activities: work, active leisure, passive leisure, necessities and other measured activities (phone/computer use, eating, waiting +other, social media use). The weighted time-use necessities were log transformed as skewness was greater than 3. Covariates were: age (range = 18 to 25), gender (female=1, male=0), race (dummy variables for being White, Black, Asian, Hispanic and Other), SES (education and income were entered as a composite or individually depending on the model), number of days since survey launch, number of co-habitants (range = 0 to 10).

Table S12e
Regression analyses predicting differences in SWB at Time 2 by socio-demographic status (Sample 9) Model 1 SWB Note. Work was captured with 3-items (work, study, commuting). Overall leisure was the sum of active and passive leisure. Active leisure was the sum of time spent praying/worshipping/meditating, socializing, exercising, intimate relations, going outdoors, hobbies. Passive leisure was the sum of time spent watching tv, napping/resting, relaxing, doing nothing. Necessities was measured as the sum of time spent shopping, personal hygiene, preparing food, doing housework For each composite, an episode weighted statistics was computed where the amount of time that respondents reported spending on each activity is weighted by the total amount of time spent in all measured activities: work, active leisure, passive leisure, necessities and other measured activities (phone/computer use, eating, waiting +other, social media use). The weighted time-use necessities were log transformed as skewness was greater than 3. Covariates were: age (range = 18 to 25), gender (female=1, male=0), race (dummy variables for being White, Black, Asian, Hispanic and Other), SES (education and income were entered as a composite or individually depending on the model), number of days since survey launch, number of co-habitants (range = 0 to 10).

Table S12f
Regression analyses predicting differences in SWB at Time 2 by socio-demographic status -controlling for SWB at Time 1 (Sample 9) Model 1 SWB Note. Work was captured with 3-items (work, study, commuting). Overall leisure was the sum of active and passive leisure. Active leisure was the sum of time spent praying/worshipping/meditating, socializing, exercising, intimate relations, going outdoors, hobbies. Passive leisure was the sum of time spent watching tv, napping/resting, relaxing, doing nothing. Necessities was measured as the sum of time spent shopping, personal hygiene, preparing food, doing housework For each composite, an episode weighted statistics was computed where the amount of time that respondents reported spending on each activity is weighted by the total amount of time spent in all measured activities: work, active leisure, passive leisure, necessities and other measured activities (phone/computer use, eating, waiting +other, social media use). The weighted time-use necessities were log transformed as skewness was greater than 3. Covariates were: age (range = 18 to 25), gender (female=1, male=0), race (dummy variables for being White, Black, Asian, Hispanic and Other), SES (education and income were entered as a composite or individually depending on the model), number of days since survey launch, number of co-habitants (range = 0 to 10). Note. Work was captured with 3-items (work, study, commuting). Overall leisure was the sum of active and passive leisure. Active leisure was the sum of time spent praying/worshipping/meditating, socializing, exercising, intimate relations, going outdoors, hobbies. Passive leisure was the sum of time spent watching tv, napping/resting, relaxing, doing nothing. Necessities was measured as the sum of time spent shopping, personal hygiene, preparing food, doing housework For each composite, an episode weighted statistics was computed where the amount of time that respondents reported spending on each activity is weighted by the total amount of time spent in all measured activities: work, active leisure, passive leisure, necessities and other measured activities (phone/computer use, eating, waiting +other, social media use). The weighted time-use necessities were log transformed as skewness was greater than 3. Covariates were: age (range = 18 to 25), gender (female=1, male=0), race (dummy variables for being White, Black, Asian, Hispanic and Other), SES (education and income were entered as a composite or individually depending on the model), number of days since survey launch, number of co-habitants (range = 0 to 10). Note. Work was captured with 3-items (work, study, commuting). Overall leisure was the sum of active and passive leisure. Active leisure was the sum of time spent praying/worshipping/meditating, socializing, exercising, intimate relations, going outdoors, hobbies. Passive leisure was the sum of time spent watching tv, napping/resting, relaxing, doing nothing. Necessities was measured as the sum of time spent shopping, personal hygiene, preparing food, doing housework For each composite, an episode weighted statistics was computed where the amount of time that respondents reported spending on each activity is weighted by the total amount of time spent in all measured activities: work, active leisure, passive leisure, necessities and other measured activities (phone/computer use, eating, waiting +other, social media use). The weighted time-use necessities were log transformed as skewness was greater than 3. Covariates were: age (range = 18 to 25), gender (female=1, male=0), race (dummy variables for being White, Black, Asian, Hispanic and Other), SES (education and income were entered as a composite or individually depending on the model), number of days since survey launch, number of co-habitants (range = 0 to 10).

Table S12i
Regression analyses predicting individual time-use differences at Time 1 by socio-demographic status ( Note. Covariates were: age (range = 18 to 25), gender (female=1, male=0), race (dummy variables for being White, Black, Asian, Hispanic and Other), SES (education and income were entered as a composite or individually depending on the model), number of days since survey launch, number of co-habitants (range = 0 to 10).

Exploratory analyses with recalled pre-COVID-19 time-use
In the surveys of remote workers who were recruited primarily from the US and Brazil (n = 24,327), respondents also indicated what their time allocation looked like in a typical day prior to the pandemic. Below, we report results when controlling for these recalled pre-COVID-19 time-use measures. Additional results are present in Tables S13a-4c.

Sample 5
In models without covariates, gender differences in time-use did not hold when controlling for recalled pre-COVID-19 time-use. However, in models with our preregistered covariates, gender differences in time spent on necessities remained significant when controlling for these measures (Mwomen = 16.81, SD = 10.08 vs. Mmen = 15.66, SD = 7.83, d = .12, p = .004; see Table S14a).  Table S14a for these exploratory analyses.

Sample 6
Unlike Sample 5 where gender differences in time-use during COVID-19 were no longer significant when controlling for recalled pre-COVID-19 time-use, in this study these exploratory analyses revealed that the following gender differences were significant: overall possibly due to having a larger sample size to detect such differences (see Table S14b).
Next, we ran exploratory analyses controlling for recalled pre-COVID-19 time-use and found a significant interaction between gender and parental status on overall leisure  The differences in time-use were significant among non-parents (d = .23, p < .001) and among parents (d = .42, p < .001), suggesting that while both parents spent more time on necessities during the pandemic than before, mothers might have experienced a greater increase in time spent on necessities (see Fig. S6, Panel B).

Sample 7
In this study, gender differences in time-use were not significant when running exploratory analyses controlling for the recalled pre-COVID-19 time-use (see Table S11h).
Next, we ran exploratory analyses controlling for these recalled time-use measures. Unlike Studies 5-6 results, the interaction between gender and parental status on necessities was not significant even in models with covariates (F[1, 729] = .114, p = .735), suggesting that in this sample the potential increase in time spent on necessities did not differ by gender among parents, possibly due to the diversity of countries included in this sample. See Table S14c for detailed results. Note. For models without covariates we ran paired t-tests. For models with covariates we ran repeated measures models with time as a within-subjects factor. Descriptive statistics for the time-use measures are reported as episode-weighted statistics (i.e. the percentage of time that respondents reported spending on each activity is weighted by the total amount of time they spent in all other activities measured within that sample). Work is a composite of 2 items (i.e. working productively, working unproductively); overall leisure is a composite of 1 item capturing active leisure like going outdoors, and 1 item capturing passive leisure like watching TV. Necessities is a composite of 2 items (i.e. doing errands/household chores and taking care of/spending time with family). Overall leisure is a composite of active and passive leisure. a Given that the skewness value of the time-use work composite was above 8.32 (thus above the pre-registered cutoff point of 2), we log transformed this variable.

Table S13b
Descriptive statistics for recalled pre-COVID time-use and mean differences with during COVID time-use (Sample 6) During-COVID Note. For models without covariates we ran paired t-tests. For models with covariates we ran repeated measures models with time as a within-subjects factor. Descriptive statistics for the time-use measures are reported as episode-weighted statistics (i.e. the percentage of time that respondents reported spending on each activity is weighted by the total amount of time they spent in all other activities measured within that sample). Work is a composite of 2 items (i.e. working productively, working unproductively); overall leisure is a composite of 1 item capturing active leisure like going outdoors, and 1 item capturing passive leisure like watching TV. Necessities is a composite of 2 items (i.e. doing errands/household chores and taking care of/spending time with family). Overall leisure is a composite of active and passive leisure. Note. For models without covariates we ran paired t-tests. For models with covariates we ran repeated measures models with time as a within-subjects factor. Descriptive statistics for the time-use measures are reported as episode-weighted statistics (i.e. the percentage of time that respondents reported spending on each activity is weighted by the total amount of time they spent in all other activities measured within that sample). Work is a composite of 2 items (i.e. working productively, working unproductively); overall leisure is a composite of 1 item capturing active leisure like going outdoors, and 1 item capturing passive leisure like watching TV. Necessities is a composite of 2 items (i.e. doing errands/household chores and taking care of/spending time with family). Overall leisure is a composite of active and passive leisure.  (under 18), and number of days since survey launch were entered as continuous variables. Gender (1=female) was dummy coded. In this study, household size was measured as a binary variable (1=living with others; 0=living alone). All respondents in this dataset were employed adults. We did not capture weekly work hours in this dataset.

Table S14b
Regression analyses predicting time-use differences in necessities by gender and parental status controlling for recalled pre-COVID-19 time-use (Sample 6) Gender ( (under 18), and number of days since survey launch were entered as continuous variables. Gender (1=female) was dummy coded. In this study, household size was measured as a binary variable (1=living with others; 0=living alone). All respondents in this dataset were employed adults. We did not capture weekly work hours in this dataset.

Table S14c
Regression analyses predicting time-use differences in necessities by gender and parental status controlling for recalled pre-COVID-19 time-use (Sample 7) Gender ( (under 18), and number of days since survey launch were entered as continuous variables. Gender (1=female) was dummy coded. In this study, household size was measured as a binary variable (1=living with others; 0=living alone). All respondents in this dataset were employed adults. We did not capture weekly work hours in this dataset.

Fig. S6 | Interaction between gender and parental status on time spent on necessities during COVID-19
Panel A

Panel B
Note. In these analyses we control for recalled time spent on necessities pre-COVID-19 as well as for ur preregistered covariates: age, income, number of children, education, number of people in the household, and days since survey launch. Necessities is a composite of household chores and taking care of others. Error bars are CI95%. Panel A is the pattern observed in Sample 5 and Panel B is the patterned observed in Sample 6.

Additional Analyses Sample 4: Shared Time on Chores and Caretaking Responsibilities
Additional measures. As part of a different project, in Sample 4 respondents who indicated living with someone else as a couple were also asked to report: 1. How many hours did they and their partner spent together on household tasks, 2. How many hours did they spend on household chores alone without their partner, and 3. How many hours did their partner spend on household chores alone without them (slide scale from 0 to 100, with max 100 across all three questions).
In addition, among these respondents those who further indicated having children were asked to report: 1. How many hours did they and their partner spend together on caretaking tasks, 2.
How many hours did they spend on caretaking tasks alone without their partner, and 3. How many hours did their partner spend on caretaking tasks alone without them (slide scale from 0 to 100, with max 100 across all three questions).
Additional results. We followed the same procedures as for these additional time-use measures. In line with our pre-registered steps for our main analyses, we regressed gender on these time-use outcomes. These exploratory analyses revealed no gender differences in amount of time spent together on household tasks or caretaking tasks. However, these analyses showed a consistent pattern whereby women (vs. men) reported spending more time alone than their partner on both household and caretaking (see Figure S below). Similarly, men (vs. women) reported that their partner spent more hours alone completing both household and caretaking tasks (see Figures S ).
Next, we regressed happiness on the household time-use measures and on the caretaking time-use measures respectively and found that spending time together on both types of necessities was positively associated with happiness. By contrast, completing either of these necessities alone was negatively associated with happiness. Notably, there was no association between how many hours one's partner spends on caretaking responsibilities and happiness (see Table S15 below).
Supplementary Information for Time-Use and Subjective Well-Being during COVID-19 | p. 62

Fig. S7a | Gender differences in time spent alone or with partner on household tasks (Sample 4).
Note. N = 520 for items capturing time spent on household tasks. N = 340 for items capturing time spent on caretaking.
Covariates: age, gender, employment status, education, monthly household income, household size, number of children, and number of days since survey launch. Age, monthly household income, education, household size, number of children, weekly work hours, and number of days since survey launch were entered as continuous variables. Gender (1=female) and employment status (1 = employed) were dummy coded.

Fig. S7b | Gender differences in time spent alone or with partner on caretaking tasks (Sample 4).
Note. N = 520 for items capturing time spent on household tasks. N = 340 for items capturing time spent on caretaking.
Covariates: age, gender, employment status, education, monthly household income, household size, number of children, and number of days since survey launch. Age, monthly household income, education, household size, number of children, weekly work hours, and number of days since survey launch were entered as continuous variables. Gender (1=female) and employment status (1 = employed) were dummy coded. Notes. N = 520 for items capturing time spent on household tasks. N = 340 for items capturing time spent on caretaking. Covariates: age, gender, employment status, education, monthly household income, household size, number of children, and number of days since survey launch. Age, monthly household income, education, household size, number of children, weekly work hours, and number of days since survey launch were entered as continuous variables. Gender (1=female) and employment status (1 = employed) were dummy coded.