Abstract
Historical changes in the use of time have long attracted researchers’ attention. Less attention has been paid to the changing configuration of time use on different days of the week. This article uses time use data collected by Statistics Canada from 1981 to 2005, and experience sampling data collected in 1985 and 2003 in Ontario (Canada) to examine changes in the allocation of time to paid work, housework, shopping, child care, personal needs, and free time on weekdays and Sundays. It examines further whether changes in the behavioural make-up of Sundays have contributed to what is referred to as ‘Sunday blues’ or emotional discomfort at the doorstep of a new week.
Introduction
Historical changes in the use of time have long attracted researchers’ attention (Robinson and Converse, 1972; Juster, 1985; Gershuny, 1992; Zuzanek and Smale, 1997; Bittman, 1998; Sayer et al., 2004; Fisher et al., 2006). Less attention, however, has been paid to the changing configuration of time use on different days of the week, particularly Sundays, as opposed to Saturdays and weekdays. Yet, flexibilization of working schedules, deregulation of the operating hours in the service sector, Sunday shopping, and conflicting work schedules of spouses that have intensified during the past two decades have contributed to the ‘de-synchronization’ of daily lives in modern societies. Sometimes subsumed under the heading of the ‘24-hour society’ (Glorieux et al., 2008), the above trends have affected not only the amount of time allocated to different activities but also its re-configuration between different days of the week (Kellerman, 1991).
This article uses time use data collected by Statistics Canada in 1981, 1992, 1998, 2005, and experience sampling data collected in 1985 and 2002–2003 by the Research Group on Leisure and Cultural Development at the University of Waterloo to examine changes in the allocation of time to paid work, housework, shopping, child care, personal needs, and free time on weekdays and Sundays. It examines further whether changes in the behavioural make-up of different days of the week have affected their emotional profiles and contributed to a growing emotional discomfort at the doorstep of a new week, sometimes referred to as ‘Sunday blues’.
Calendar rhythms: Conceptualization and research findings
Interest in the changing time use structure and emotional texture of different days of the week is conceptually grounded in studies of subjective or social time that go back to the end of the nineteenth and the beginning of the twentieth centuries (Henri Bergson, Emile Durkheim, Maurice Halbwachs). An important part of the debate over the nature of ‘inner’ or ‘social’ time as opposed to mechanical or astronomical time was the conceptualization of the weekly cycle as a social phenomenon. According to Durkheim (1912), the week is a socio-cultural rather than natural artefact. Temporal segregation of different days of the week reflects a need for ‘pulsating’ alternations between the ordinary and the extraordinary, the sacred and the profane.
Durkheim’s notion of the weekly cycle was elaborated by Sorokin. The following statement from his Sociocultural Causality, Space and Time (1943) may well serve as a springboard for the analyses reported in this article. ‘Imagine for a moment’, Sorokin wrote, ‘that the week suddenly disappeared. What a havoc would be created in our time organization, in our behaviour, in the coordination and synchronization of collective activities and social life, and especially in our time apprehension. Many of us would certainly mix our appointments, shift and change our activities, and fail many times to fulfil our engagements. If there were neither the names of the days nor the weeks, we would be liable to be lost in an endless series of days – as grey as fog – and confuse one day with another.’ The week is, in Sorokin’s words, one of the most important points of our ‘orientation in time and social reality’ (1943: 420).
Interesting observations about behavioural and emotional attributes of the weekly cycle can be found in Zerubavel’s Hidden Rhythms (1981) and The Seven Day Circle (1985). According to Zerubavel, there are distinct expectations of behaviours and attitudes associated with different days of the week. Saturdays are used to catch up on personal services, shopping, and other family or household obligations. Sundays are reserved for worship, sleep, increased leisure involvement and family contact. Fridays serve as a transition from weekdays to weekends, while Mondays are marked by an emotional downturn associated with a return to the routines of the workweek (‘blue Monday’).
Issues addressed by Zerubavel from a historical and conceptual perspective were examined empirically by researchers studying uses of time. Findings reported by Szalai (1972), Chapin (1974), Katz and Gurevitch (1976), Robinson (1977), and Zuzanek and Smale (1992) by and large corroborate Zerubavel’s observations, showing that, for the employed population, levels of paid work peak in midweek, housework and shopping activities surge on Saturdays and free-time activities and family contacts dominate daily life on Sundays.
Differences in the emotional profiles of different days have also attracted researchers’ attention. An interesting attempt to empirically examine the emotional texture of the weekly cycle has been undertaken by Larsen and Kasimatis (1990). In their study of undergraduate students, the ‘hedonic effect,’ measured as a differential score of positive and negative feelings, peaked on Saturdays, followed by Fridays, Thursdays, and only after that by Sundays. Zuzanek and Mannell’s (1993) analyses of the emotional pulse of the week showed that employed adults experienced higher positive feelings (affect) and lower negative feelings (anxiety) on weekends rather than on weekdays, with Saturdays marginally outperforming Sundays. The lowest levels of affect and the highest levels of anxiety were reported on Mondays. In a recent study of employed adults, Ryan et al. (2010) also registered higher emotional well-being ratings on weekends (attributed by the authors to Sundays’ greater behavioural autonomy and ‘relatedness’), but they found no support for the notion of ‘blue Monday.’
Unlike weekly rhythms, across-the-day rhythms of human life attracted less research attention. Some factual information about the distribution of time use across the day was collected as part of applied and marketing studies. First attempts to examine the sequencing of daily behaviours was undertaken by mass media organizations that were interested in establishing viewing preferences of their audiences at different times of the day in an attempt to identify ‘prime media time.’ Ward’s 1954 study for the Mutual Broadcasting System in the United States, time budget studies of life-styles and media habits conducted by the Tokyo Radio and Television Culture Research Institute (1960–61), or the 1965 BBC time-budget study in Britain, may serve as examples of this type of research.
Across-the-day variations in emotional states, stress levels and performance also attracted the attention of researchers. Because of methodological differences and an interest in different substantive issues, it is difficult to summarize the findings of these studies. Studies of social pathological rhythms in Boston and across-the-day distribution of distress calls in Newcastle, Australia have shown that telephone self-reports of stress, drug related problems and distress peak around 3:00 pm. A survey of married couples in the UK also recorded the highest levels of self-estimated stress around 3:00 pm (Parkes, 1974). Studies of the across-the-day distribution of moods among adolescents suggest that Friday and Saturday evenings stand as emotional climaxes of the week (Larson and Richards, 1995), whereas on Sundays positive feelings begin to subside in the afternoons (Csikszentmihalyi and Hunter, 2003).
While we have reasonably reliable information about general time use trends in industrial societies and valuable information about across-the-week distribution of time and emotional states, there is little if any information about historical changes in the configuration of activities and moods across the week and across-the-day. If the days of the week are becoming behaviourally more alike, does this affect their emotional profiles? Are Sundays becoming increasingly ‘grey as fog’? And if the latter is the case, when does the Sunday ‘greying’ begin to thicken – in the mornings, afternoons or in the evenings? These are some of the questions that are addressed in this article.
Research problem: The charm of diversity and the importance of the rite
Analyses reported in this article are statistical, but the issue at its centre may have been best formulated not by social scientists but by a writer – Antoine Saint Exupéry, the author of ‘The Little Prince,’ a fairy tale written over half a century ago that remains one of the most telling manifestos of human empathy (Saint Exupéry, 1943/2000). When the main character of the tale, the Little Prince, who landed on earth from another planet, meets the fox in the desert, whose main concern is how to escape the hunters, they touch upon a problem that is central to this article – the charm of diversity and the importance of the rite! ‘One must observe the proper rites,’ said the fox. ‘What is a rite?’ asked the little prince. ‘Those are actions too often neglected,’ said the fox. ‘They are what makes one day different from other days, one hour from other hours. There is a rite, for example, among my hunters. Every Thursday they dance with the village girls. So Thursday is a wonderful day for me! I can take a walk as far as the vineyards. But if the hunters danced at just any time, every day would be like every other day, and I should never have any vacation at all.’ (p. 60) Did Sundays become, over the past two decades, behaviourally more like the rest of the week? Did the emotional profile of weekend days, particularly Sundays, change during the observed period? Did the across-the-day distribution of moods on different days of the week change since mid-1980s?
Data and method
Data
Findings reported in this article are based on Canadian national time use surveys conducted in 1981, 1992, 1998 and 2005 by Statistics Canada and Experience sampling surveys (ESM) conducted in 1985 and 2002–3 by the Research Group on Leisure and Cultural Development at the University of Waterloo, under the direction of Jiri Zuzanek (principal investigator) and Roger Mannell (Zuzanek and Mannell, 1993).
The 1981 National Time-Use Pilot Study was conducted in 14 urban centres across Canada over a period of three months (September to November). It sampled adult population aged 18 and older (n = 2680). The 1992, 1998 and 2005 General Social Surveys (GSS) sampled populations aged 15 years and older and were conducted over the entire year. Their respective samples were 8996 in 1992, 10,749 in 1998, and 19,597 in 2005. 1
The two ESM surveys were conducted in the Kitchener-Waterloo and larger Metropolitan Toronto areas of Ontario. The 1985 ESM survey sampled 167 employed respondents in Kitchener-Waterloo (n = 5955 self-reports). The 2002–3 survey was part of a larger study of Adolescent Time Use and Well-Being and sampled 214 parents of teens attending school (n = 10,453 parental self-reports).
In ESM surveys, respondents carried electronic pagers or pre-programmed wrist watches that were signalled (beeped) randomly during wake-up hours, usually in two-hour intervals, for a period of one week. At the time of the beep respondents were asked to record what they were doing, where and who they were with, as well as their subjective feelings. The signal response rates in the 1985 and 2002–3 ESM surveys were 74% and over 80%, respectively.
While time use information collected by ESM surveys is not as complete and detailed as in time diary surveys, they provide a much more detailed picture of the changing emotional texture of daily life.
Although the 1985 and 2003 ESM surveys used similar survey instruments and sampling procedures, matching the two data sets for trend analyses required some adjustments. For the purposes of comparison, respondents who did not have children were excluded from the 1985 sample and the 2003 sub-sample was limited to employed parents living exclusively in the Kitchener-Waterloo area. As a result, the 1985 sub-sample was reduced to 72 respondents (2567 self-reports) and the 2003 sub-sample to 125 respondents (6299 self-reports).
Method
Both GSS and ESM surveys collected time use and emotional well-being information but they collected it differently.
In GSS surveys, respondents were asked to recall all activities conducted during the day preceding the interview. 2 Time was computed by summing the durations of all activity episodes performed during the survey day. This information was grouped into larger activity categories that encompassed typical daily behaviours of the entire 24-hour day.
ESM surveys, unlike time diaries, did not record the duration of the activities. Nor did they cover the entire 24-hour day, but rather only its wake-up part. Time use information from the ESM surveys was therefore computed as a percentage of wake-up episodes grouped into activity categories similar to those used in time diary surveys. 3
Questions about subjective well-being, such as life satisfaction, perceived time pressure and stress were asked in Canadian GSS, but these questions captured ‘generalized’ assessments of respondents’ emotional well-being rather than the emotional texture of specific situations, days of the week or times of the day. Differently than in the time diary surveys, ESM respondents were not asked to assess their generalized feelings of happiness or stress, but rather to report how they felt at the moment when they were beeped, for example how happy, worried, stressed or pressed for time they felt. By monitoring daily life as a sequence of instantaneous experiences precisely anchored in time, ESM data provide researchers with a unique opportunity to examine the emotional and health effects of time use when it happens (Csikszentmihalyi and Larson, 1987; Larson and Richards, 1995). The mean ratings of singular or composite well-being measures can be used to assess emotional texture of different groups of activities, different days of the week and different periods of the day, as well as emotional profiles of individual respondents or different social-demographic groups. 4
Experiential states of particular interest for the analyses reported in this article include: the amount of behavioural choice at the time of the beep, respondents’ enjoyment of and interest in activities performed, their wish to do something else, as well as composite measures of affect and anxiety summarizing positive and negative feelings. Affect is a composite of feeling happy, good and cheerful (Alpha = 0.84 in 1985 and 0.84 in 2003). Anxiety sums up feelings of being worried, upset and tense (Alpha = 0.89 in 1985; 0.84 in 2003).
Analyses
Analyses reported in this article focus on historical changes in time use and subjective well-being. Data used for these analyses were collected as part of disjoined surveys conducted at different times. The confidence in the substantive significance of the observed trends rests on representative sampling of Canada’s population by GSS surveys and the uses of standard weights. Posthoc Bonferroni controls were used in the analyses of within-survey differences.
Analyses of the changes in subjective valuations of well-being recorded in 1985 and 2003 by ESM surveys are more debatable because of the smaller size of the ESM samples. Yet both of these surveys were conducted by the same team of researchers, using similar survey instruments, within similar geographic areas, and for the analyses reported in this article their samples were standardized with regard to employment status, marital status and parenthood (see p. 12). It is therefore our opinion that the two surveys provide an intriguing and unique opportunity to examine not only changes in the use of time but also its emotional connotations.
Changing time use patterns on Sundays and weekdays: GSS and ESM findings
Sunday time use in Canada: GSS 1981 to 2005. Adult population aged 20 to 65 (minutes per day)
Paid and domestic work, personal needs, and free-time include travel associated with these activities.
In 1981, 45.6% of women aged 20 to 65 reported having at least one child under the age of 15 as part of the household. Corresponding percentages in 1992, 1998 and 2005 were 40.9%, 35.6% and 34.2%, respectively.
Time allocated to child and family care did not change much between 1981 and 2005 for the adult population (42 and 38 minutes, respectively) due to a decline of parenthood. 5 If, however, child and family care were computed for families with children under the age of 15, the amounts of Sunday child and family care would rise – from 69 minutes in 1981 to 87 minutes in 1992, and to 90 minutes in 1998 and 2005. In families with children under the age of 5, child care on Sundays increased from 100 minutes in 1981 to 146 minutes in 2005, i.e. by 46%.
The amount of time spent during Sundays on personal needs declined between 1981 and 2005 by almost 50 minutes, with much of the decline accounted for by a drop in time spent on meals at home and personal care. 6
Time use changes on weekdays and Sundays: Canada 1981 to 2005; adult population aged 20 to 65 (percent of change)
Changes of Sunday time use by gender: 1981 to 2005 (percentage change)
Mothers with at least one child under the age of 5.
Weekday and Sunday time use: ESM 1985 and 2003; employed parents (percentage of wake-up time)
findings are based on 2,567 self–reports collected from 72 respondents. 2003 findings are founded on 6,299 self-reports from 125 respondents. Bolded means signify significant Sunday-weekday differences (Posthoc Bonferroni alpha <.05).
Time use and its emotional correlates: ESM 1985 and 2003; (employed parents; standardized betaa)
Standardized beta, controlling for age, gender and education, are statistically significant
(p < 0.05). Parenthesized betas are marginally significant (< 0.1).
Affect = composite of feeling happy, good, and cheerful (alpha = 0.87).
Socio-demographically, the two sub-samples, used for the analyses reported in this article, closely resemble one other. All respondents were employed parents living in the Kitchener-Waterloo area. Women constituted 60% of the 1985 sub-sample and exactly the same percentage in 2003. Twenty-six percent of respondents in 1985 reported having high school or lower education; 40% completed college or some university, and 34% had a BA or higher university degree. The corresponding figures in 2003 were 30%, 37% and 33%. Eighty-three percent of respondents in 1985 were married and 17% divorced or single. In 2003, the respective figures were 84% and 16%. The only noticeable difference between the two sub-samples was respondents’ age. The mean age in 1985 was 37, while in 2003 it was 45.
ESM data collected in 1985 and 2003 show that the share of time allocated on Sundays to paid work increased during the observed period from 4.9% to 6.4%. Sunday episodes of domestic obligations also grew from 33.0% in 1985 to 36% in 2003. Much of this increase arises from the surge in Sunday shopping, house upkeep and child care. The frequency of cooking episodes has, on the other hand, fallen, thus paralleling the GSS findings. The share of free-time episodes in 2003 was lower than in 1985, 35.0% compared to 45.4%. Similar to the GSS findings, ESM data show that Ontario parents spent in 2003 less time socializing with friends and relatives, reading, pursuing manual hobbies and listening to the radio, while increasing their involvement in computer related activities that were virtually absent in 1985.
ESM data show, as well, that paid and domestic work expanded during the observed period faster on Sundays than on weekdays. Sundays’ paid workload in 2003 was 31% higher than in 1985, while on weekdays it grew by only 12%. On weekdays domestic workload contracted during the observed period by 7%, but on Sundays it expanded by 7%. The number of shopping episodes on Sundays was three-fold higher in 2003 than in 1985, yet on weekdays it declined by 25%. In 2003, there were 23% fewer free-time episodes reported on Sundays than in 1985, but only 9% fewer free-time episodes on weekdays.
Overall, analyses of GSS and ESM data produce similar findings with regard to time use trends from the early 1980s to 2005. At the beginning of the 21st century Sundays became busier and resembled weekdays behaviourally more than they did two decades ago.
Changing emotional texture of weekdays and weekends: 1985 and 2003 (ESM)
Data collected in 1985 and 2003 by ESM surveys allows us to compare the emotional texture of weekdays and weekends two decades apart.
Table 5 shows that positive emotional outcomes were associated in both 1985 and 2003 on weekdays and weekends primarily with free-time activities, such as socializing with friends, physically active leisure, or meals at home, while paid work and domestic obligations (in particular, housekeeping) were accompanied by negative emotional connotations. So how did the broadening of the spectrum of negatively shaded activities on Sundays affect the experiential profile of the day that was hailed since the Middle Ages as “dearly loved but one day betwixt a Saturday and Monday”? (Carey, 1979/1722).
Emotional textures of different days of the week: ESM 1985 and 2003
Bolded means: Posthoc Bonferroni alpha for Sunday/weekday differences is significant at 0.05 level.
There may have been a historical shift in the emotional valuation of Sundays compared to Saturdays. In the past, Sundays have been lauded as emotionally more attractive than Saturdays (Zerubavel, 1985), but this assertion is not supported by ESM findings. In 1985, the levels of affect and anxiety were virtually the same on Sundays and Saturdays, and so was the desire to engage in an activity other than the one pursued at the time of the beep. 8 This was even truer in 2003. If in 1985 Sundays were rated as experientially freer than Saturdays, then the reverse was true in 2003. Questions about how interested or bored respondents were at the time of the beep were not asked in 1985 and therefore cannot be used for a comparison with the 1985 findings, but the virtually identical levels of interest in things done on weekdays or Sundays, as well as almost identical levels of boredom across the entire week, probably signal a continuous erosion of Sundays’ special emotional status.
Finally, there is some evidence that across-the-day distribution of positive and negative feelings on weekdays and weekends was shaped differently in 2003 than in 1985. Figure 1 shows that in 1985 perceived behavioural discretion (freedom to choose what one was doing) peaked on Sunday afternoons, before declining in the evenings. On the contrary, in 2003 it fell in the afternoons. On Sunday afternoons in 1985 respondents wished to do something else the least. In 2003, they felt so the most (Figure 2). Sunday afternoons in 1985 were relatively free of anxiety, while in 2003 the levels of anxiety rose steadily through most of the day (Figure 3). In short, in 1985 positive feelings peaked on Sunday afternoons, while in 2003 Sunday afternoons were marked by an emotional downturn, and the across-the-day distribution of positive and negative feelings resembled weekday patterns to a greater extent than in 1985.
How free were respondents to choose what they were doing: 1985 and 2003 Sundays and weekdays compared (means). How much did the respondents wish to do something else: 1985 and 2003 Sundays and weekdays compared (means). Across-the-day distribution of anxiety: 1985 and 2003 Sundays and weekdays compared (means).


All of the above having been said, did the changes in the allocation of time – more time working for pay, added work around the house, more shopping, less free time and less socializing – change the emotional texture of Sundays? Our analyses offer rather limited support for such a conclusion. Sundays’ greater sense of experiential constraints did not seriously affect their emotional valuation. Composite measures of affect and anxiety in 2003 did not differ much from the levels recorded in 1985. Nor was there a noticeable change in respondents’ desire to do something else other than what they were doing at the time of the beep (Table 6). If there were a shift in the experiential make-up of the weekly cycle, it was relatively small and the similarities between the emotional colouring of different days in 2003 and 1985 are probably more striking than the differences.
The question is: Why are the differences so small – why don’t we see clearer emotional effects of the visible behavioural changes? There are two possible answers to this question. One of the answers is that the samples of the two surveys are relatively small and hence deficient for the purposes of establishing general well-being trends. While there is some truth in this observation, we do not think that findings about relatively small changes in the Sundays’ emotional texture are due to sampling. The fact that most of the ESM behavioural findings parallel GSS conclusions suggest that ESM findings about the emotional make-up of Sundays should probably be taken at their face value and their interpretation sought in factors other than sampling.
Discussion and conclusion
Empirical findings rarely provide a definitive answer to conceptual issues raised in theoretical debates. Calendar and daily rhythms in the lives of the employed parental populations did not initially constitute the research focus of GSS or ESM surveys. It was, therefore, gratifying to find out that in spite of the potential ‘noise’ often encountered in survey research as a result of imperceptible ambiguities in the wording of questions, innocuous differences in the operationalization of variables or unanticipated changes in the research focus, the basic expected relationships between the days of the week, daily behaviours and moods, advanced by previous research, were mostly supported by our findings. With regard to the changes of subjective well-being on Sundays reported in this article, the findings may be tentative but they can certainly serve as an invitation for further analyses of the changing emotional make-up of the weekly cycle.
Our analyses of across-the-week distribution of daily activities indicate that weekly behavioural rhythms are strongly affected by the weekday–weekend pulse and that this pulse has changed during the period from 1981 to 2005. Sundays became busier and behaviourally closer to weekdays than they were at the beginning of the 1980s. This change would probably become even more obvious if one were to go back to the older data sets, such as for example Chapin’s 1968 data from Washington, DC, where Sunday meals at home occupied more than 90 minutes and shopping only 8 minutes.
Economic and social effects of the deregulation of shopping and service hours were debated intensively by economists in Canada and other industrial counties (Laband and Heinbuch, 1987; Lanoie et al., 1994). Most of this debate focused on the economic efficiency of deregulation policies, their impact on labour demand or the redistribution of social welfare. Arguments were made that deregulation of working and shopping hours will help create new jobs, ease traffic and parking congestion, increase competition between retailers and possibly lower prices (Kay and Morris, 1987; Burda and Weil, 2005; Skuterud, 2005). The liberalization of work schedules and shopping hours, we were told, would also accommodate the changing lifestyle demands of working women who had limited access to shopping on weekdays. The general well-being effects of new time arrangements have, however, not been examined systematically.
If the biological function of the week resides in the release of physiological fatigue and adaptation to physical stress (Larsen and Kasimatis, 1990), 9 and its economic function lies in the synchronization and coordination of economic life (Zerubavel, 1985), then the socio-cultural week cycle helps integrate us into the social fabric of the family, the community and the society of which we are part. Calendar rhythms served throughout history as instruments of social cohesion and are particularly precious at times when established ways of life undergo radical transformation.
Our analyses indicate that changes in the behavioural profiles of Sundays have not been paralleled by changes in their emotional texture. Respondents surveyed in 2003 continued to enjoy Sundays at a level not very different from 1985. Behaviourally, Sundays are today more constrained and more like the rest of the week, but we do not seem to mind. Survey research has shown that, generally, people tend to enjoy the things they are doing (Michelson, 2010).To make the best of one’s life is natural. In a way, this serves as a self-defence mechanism. Life becomes miserable when we continuously dislike what we are doing. 10 Economic analyses suggest that the net gainers of shopping and service deregulation were large retail chains, but the removal of shopping restrictions on Sunday met with popular support in most countries. 11 And if the question were asked today, most people, including women, would probably not want to return to the working and shopping arrangements of the 1980s. Convenience has its merits. The question is: Does it make us happier?
Unlike the fox in the Exupery’s story about the Little Prince, we do not seem to be overly concerned with the ‘colonization’ of the Sundays by weekdays. Nor do we seem to fully understand and appreciate the kindred bondage between diversity and the rite that was so well mastered by the fox. Perhaps not only the days of the week became ‘grey,’ but so did we.
Footnotes
Acknowledgement
The author would like to thank James T. White and Alexander Graham for their assistance in preparing this manuscript.
