TIME USE RESEARCH

Time provides the organizational key to action at the level of individuals, groups, and institutions. It also defines a normative framework that regulates interpersonal relationships and allows synchronized operations in different parts of society. In the concept of time, structural as well as symbolic facets assume significance. In fact, as it is conceived in sociological theory, time is a means of social coordination as well as a dimension that assigns value to action schemes in a system assuring social order (Pronovost 1989; Sue 1994).

The empirical study of the temporal organization of human action reveals the functional characteristics of social roles and the societal division of social tasks. For instance, indicators of inequality and social exclusion can be derived and compared by referring to estimates of the amount of time involved in gender-related activities such as market work and housework. The time use of populations or subpopulations is studied mostly by means of diary procedures that assess individual "time budgets:" the sequence, timing, and duration of activities performed by individuals over a specified period. It is misleading to think that the aim of time budget research is time as either a physical or a subjectively perceived entity. As stressed by the major postwar promotor of this area of study, the Hungarian sociologist Szalai, the object of studying time is to discover the use people make of their time, or "the arrangement and the fit of people's activities in a temporal frame of reference, the temporal order and structure of everyday life" (1984, p. 20).

From this point of view, time is a special kind of resource. As with material goods and more symbolic commodities such as money, people have a "fund" of time at their disposition and make decisions on how to use, "spend," or "invest" it. However, time is a far more democratic resource in that every person deals with the same basic "stock" of it, such as the twenty hours in a day and the seven days in a week. This means that a shared reference of differential time allocation patterns can facilitate coherent comparisons and meaningful interpretations. It is this particular aspect that has made time use an important topic in quantitative social research. Both academic scholars and national statistical offices have shown a growing interest in time-budget data because those data permit policy-oriented microanalyses of changing lifestyles at the individual or household level as well as macroanalyses of social and economic inequalities in the context of cross-national comparative studies.

In contrast to the physical notion that attributes equivalent temporal resources to all people and therefore facilitates systematic accounts, from a subjective point of view, the length of a day is not always the same. Some people seem to have more time than others. This phenomenon reveals the sociopsychological dimension of time in which concepts such as stress and alienation are relevant. In a world where the quality of life depends largely on what one gets out of the time at one's disposal, coping with time and compressing multiple activities into the same time slot have become important skills. Furthermore, it is a universal observation that boring periods during a day, week, or year can seem long, whereas other, quantitatively equivalent but exciting periods of time are perceived as a passing moment. However, in studying social time, time-budget research adopts an activity-oriented approach and focuses only indirectly on subjective experiences. In substance, this means that observable patterns of behavior are selected as primary evidence and, in methodological terms, standard time units are chosen for measurement purposes.

THE TIME USE RESEARCH TRADITION

Although social time is not intrinsically quantitative, the use of standard time units for the purposes of analyzing the structure of everyday life seems legitimate, since the transfer of human work from agriculture to artificially controlled industrial environments and the subsequent changes in civilization have largely transformed natural time (tied to seasonal conditions and biological needs) into conventional, rational time (Elias 1988). As a corollary to this process, social life has become dominated by timekeeping. This chronometric function is characterized by a universally accepted "time" language that coordinates rhythms of action in the public and private spheres (Zerubavel 1982). Therefore, time budgets, which are concerned with different kinds of schedules for structuring the flow of events, are a key to the systematic investigation of the complex interdependencies and trade-offs of modern life.

The historical origins of the study of social time lie in both sociological theory and empirical research. With regard to theory, the early French school of sociology was interested in this phenomenon from the point of view of the historical and anthropological dimensions of social change (Pronovost 1989). Around the turn of the century, Hubert, Mauss, Durkheim, and Halbwachs conceived of social time as an intrinsically qualitative phenomenon that was relevant for the characterization of the sacred-profane symbolic dichotomy in the evolution of the collective consciousness or the formation of a collective memory. Their focus was a macro one, and their main interest was to explain long-term cultural change.

In contrast, Mead in the early 1930s chose an individualistic, micro-oriented approach that offered a philosophical rationale for studying time in the present. The present was conceived of as the context for the emergence and assimilation of various social time systems in interplay with the definition of different notions of "the self." Thus, the French and American approaches represent polar opposites, with one conceiving time within the matrix of historical societal relationships, and the other from the perspective of mutable configurations of symbolic interactions in small groups. The somewhat later work of Sorokin reflects both approaches. He presents the functionalist idea that the plurality of individual time schemes requires extensive synchronization to achieve social cohesion and that time expresses the sociocultural "pulsation" of a society.

In the 1960s, Gurvitch gave new impetus to the study of time after a considerable period of neglect. He was the first to conceive of time as an important source of contradictions and potential conflict. In particular, he stressed the hierarchically diversified aspects of the phenomenon (e.g., social time, time in organizations, time in special social groups) and raised the issues of power and legitimacy. More recently, Merton introduced the concept of "socially expected durations" that highlight the normative aspects of the embeddedness of time in social structures. In contrast, Elias stressed the role of time as a symbolic means of social regulation but also of increasingly unpredictable individual self-expression.

The historical origins of empirical investigations of social time are even older, going back to the middle of the nineteenth century. Three lines of research are significant: the research conducted by Friedrich Engels on the English working class, where the temporal organization of daily life was the issue; the studies undertaken by Frédéric LePlay, in which the economic "family-budgets" of workers in several European countries were assessed (similar to what "time budgets" do today); and the experimental work on time and motion performed by Frederick Taylor, based on carefully collected chronometric data. Taylor's aim was to introduce strategies of scientific time management in industry. In regard to the adoption of time-budget methodologies, there were Bevans's pioneering studies (1913) of how workers spent their spare time and early Soviet inquiries by Strumilin into time use as the basis for rational social programming. Lundberg and Komarovsky's research into the organization of time within the realm of community research was conducted along the lines of American cultural anthropology.

Of more enduring interest, however, were two studies published in the 1930s. Jahoda et al.'s Marienthal: The Sociography of an Unemployed Community (1933) was a substantial contribution to the study of time use. It explores changes in the meaning of time for German working-class families when work, as a dominant regulating and legitimizing criterion for time use, has disappeared. Male workers tended to become severely disoriented and alienated after losing their work-based prestige, whereas their wives were successful in mastering that situation because they had much more positive attitudes based on more complex sources of social recognition. This research also shows how important it is to be aware that similar circumstances can assume diverse meanings for different groups.

The other study, whose importance lies in its methodological ideas, is Sorokin and Berger's Time-Budgets and Human Behavior (1939). Here the aim was to explore meaningful criteria for decision making conducive to different time structures. Information on motivations and the kinds of experiences associated with certain practices and future projects was collected to acquire a deeper understanding of how people deal with their time. Even more important, this research raised the crucial epistemological question of how to divide essentially continuous strings of behavior into activity segments that, beyond commonsense classifications, can be grouped into homogeneous and mutually exclusive categories (Kurtz 1984). The difficulty was that some activities that from an external viewpoint seemed identical could assume unequal functions in the eyes of those concerned, or conversely, that substantially different activities could assume similar functions. This means that any classification of activities presupposes an interpretive act. With this fundamental problem in mind, these authors were forced to approach earlier and purely descriptive assessments of time allocation with skepticism and spell out the methodological issues in their research.

After World War II, research took different directions in accordance with divergent political ideologies. The assessment of living conditions, which involved obtaining background data for economic planning and monitoring centrally initiated social change, continued to be of pivotal importance in research in eastern Europe's communist countries. Prudensky's time-budget studies in the Soviet Union in the 1960s not only followed the direction of Strumilin's work but also reflected these ideological concerns. The need to broaden data gathering to obtain effective guidance for public policy at the national level led Hungary's statistical office to begin the first microcensus research in this field.

In capitalist societies, time use research was concerned principally with mass media and leisure culture. Pioneering time use studies of audiences were undertaken by the BBC as well as NHK, the Japanese radio and television system, using large-scale survey techniques. From 1960 onward, NHK conducted regular five-year follow-up rounds of research to obtain time series statistics that showed long-term longitudinal development. This was a useful strategy because it produced an interesting account of how, in terms of time use, traditional ways of life are supplanted by innovative, primarily television-centered styles.

This was the situation in 1963, when the idea of conducting a Multinational Comparative Time Budget Research Project emerged. This was an ambitious sociological initiative in light of the organizational and data-processing difficulties of those years. Launched by a group of scholars directed by Szalai, sponsored by UNESCO's International Social Science Council, and coordinated by the Vienna Centre, this project attempted to obtain an interculturally valid body of knowledge that would shed light on regularities or variations in the functioning of human societies with regard to time use. This information was to be derived from a database of twelve different countries by using methodological instruments that assured a high level of analytic precision (Szalai 1977). In organizing the initiative, the basic concern was to avoid the emergence of a single central vantage point regarding the collection, elaboration, and interpretation of information. Therefore, research sites had considerable autonomy in studying the uniform data sets collected by means of strictly standardized survey instruments from probability samples of urban populations in the twelve countries under investigation.

From a positivist point of view, the focus on chronometric evidence and on an array of "hard" time use indicators enhanced the scientific character of the study and facilitated the collaboration of teams from such culturally and sociopolitically different environments as the United States and the Soviet Union. Of course, collaboration entailed the acceptance of common working hypotheses such as the expected influence of the major independent variables of industrialization and urbanization on the modalities of the division of market work and nonmarket work in households. By contrast, time for leisure was thought to be correlated with superior levels of modernization and democratization. These hypotheses clearly reflected the research traditions of the day, and so to connect the ideologically distant worlds of the 1960s, it was necessary to choose highly conventionalized and neutral time indicators as empirical evidence.

The Twelve Country Project, characterized by a strong belief in the "scientific and social import of cross-national comparative research" (Szalai 1977), did not go without criticism. Some thought that it was most important for cross-national research to contribute findings on general theoretical problems (Przeworski and Teune 1970). However the promoters were convinced that the discovery of the empirical peculiarities of cultural settings was at least as important as the verification of a priori hypotheses on common characteristics and trends. That the pragmatic point of view prevailed meant that the problem of a lack of reliable, relevant, and usable data had to be overcome.

In retrospect, it seems that this project did not contribute much to general theory, but it did produce an elaborate methodology whose essential lines are applied to basic and official survey research in many countries today. In fact, as Szalai hoped, the homogenization of time-budget methods now permits the drawing of "maps" of collective daily activity schemes at different levels of definition that have proved to be useful diagnostic elements for many policymakers and grassroots organizations. When economic indicators are insufficient, statistical information regarding the use of time can open up new policy perspectives and guide substantial change, especially when gaps in the quality of life manifest themselves and corrective action is needed to improve the conditions of disadvantaged social groups.

METHODOLOGICAL ASPECTS OF TIME-BUDGET RESEARCH

Since the work of Szalai, the methodology of time use research has been further refined. Under the auspices of the International Association of Time Use Research, in particular under the guidance of Harvey (1984, 1993), who has repeatedly codified the best practices, statistical bodies have reached a consensus on the format of official survey research.

The task of discovering the temporal order and structure of everyday life by means of time-budget methods involves fax more complex activities than gathering simple answers to questions of who does what, when, where, and with whom. In the design of a time-budget study, methodological issues such as the scope and scale of the research, the population from which the sample is to be drawn, the format of the data-gathering instruments, the classification and coding of activities, the choice of basic indicators, and the validity and reliability of the data must be resolved.

Defining the scope of a study also means fixing its scale. In fact, a time use survey may deal with a special group of persons, such as working women or teenagers, or with comprehensive national populations, perhaps excluding preschoolers. Or it may focus on a daily activity such as housework and child care or leisure. Alternatively, it may attribute equal weight to all everyday pursuits. Finally, the study may be aimed at discovering how the time of a special kind of day is spent, may take into account the rhythm of the week by distinguishing workdays from Sundays, or may be interested in longer periods such as the year with its seasonal differences and, in the extreme case, the life cycle (which of course would have to be studied on the basis of long-term recollections). Theoretically grounded sociological research is, for economic reasons, more likely to have circumscribed objectives, whereas national statistical offices have all-encompassing multipurpose datasets available. However, in both cases, the predominant tendency is to focus on the twenty-four hours of one or more single days in the life of the respondent. When these days are distributed over the week, month, or year, the average profiles of the period can be synthetically reconstructed. Such profiles most often refer to uninterrupted sequences of nonoverlapping main or "primary" activities. When there is interest in "secondary" activities or, more precisely, in contemporaneous activity episodes, the respondent usually is asked to designate the elements that represent the principal flow, which typically covers the 1,440-minute arc of the day. In fact, leaving secondary activity out of focus furnishes an unduly simplified picture of what is going on, since it ignores efficiency strategies that enable people who are short of time to deal simultaneously with multiple jobs. This frequently criticized weakness is compensated for by the heuristically valid fact that strict 24-hour accounts produce agile descriptive models. With such models, whatever time is saved on one kind of activity is strictly accredited, using zero-sum logic, to one or more other activities. Therefore "time set free" and the equivalent "time gained" concept furnish concise indicators of social change. Tracing the balance of the two magnitudes gives a dynamic slant to the analysis of time use and sheds light on the spectrum of strategic options.

Depending on the scope of the study, populations and samples are variously defined. From this point of view, the most significant difference between time use studies regards the choice of the sampling unit, which may be the individual or the household. In earlier studies, individual time use was of primary concern, and so estimates were obtained by classifying persons by their sociodemographic characteristics. More recent research, however, has looked more into how different types of families, as molecular units, manage time allocation with regard to income generation as well as work in the sphere of home and child care. Statistical offices now use very large probability samples of households to be able to generate cross-tabulated data on specific territorial areas and particular social groups. In Italy, for instance, the last national time use survey, conducted by Istat in 1988 and 1989, consisted of more than 38,000 persons belonging to almost 14,000 households. A survey conducted in Germany in 1991 and 1992 by the Statistisches Bundesamt included 7,200 households. One of the most difficult problems in time-budget research is the sample units' frequent refusal to respond once they see how much time is involved. In fact, nonresponse rates tend to be high and in some official surveys amount to almost 30 percent. This problem is easier to handle in smaller-scale studies, which often use quota sampling.

Data gathering in time use research begins with an interview (Scheuch 1972) to record the characteristics of the respondent and his or her family, contractual work arrangements, normal labor supply, and housing or other assets and to inquire into irregularities in the day designated for collecting the time-budget information. The time budget itself is registered in a protocol, a diary, or modular display where the beginning and the end of each activity can be indicated together with other information. The resulting datasets show for each day and respondent (1) the number of different activities performed and the frequency of each activity in separate episodes (for instance, the series of daily meals or the periods passed in front of the television set) and (2) the timing, duration, and sequence of activities or activity episodes. Most often, the interviewees register activities by using their own words. A grid of minimal time intervals is given (the "fixed interval" solution), where the task is to fill each interval with an activity, or the interviewee is asked to specify the exact time points of his or her schedule (the "open interval" solution). To obtain the essential elements of the interviewee's context, there is usually room to indicate contemporaneous activities (for instance, reading while using public transport or listening to music while doing homework); participation in activities with family members, neighbors, friends, and colleagues; and where the activity takes place. The least expensive method of data collection is the condensed telephone interview, which explores time use on the previous day. For field studies, there are other procedures, such as single face-to-face interviews and two personal interviews. In the first case, the person is asked to recall what he or she did the preceding day. This procedure is complicated when a lot of detail is required. The second procedure involves two personal interviews. During the first, background information is collected and the time use diary is left behind, to be filled in the next day. During the second interview, on the day after the respondent's observation of his or her time use, the interviewer checks and refines the registrations. In Scandinavian countries, people were asked to return diaries by mail. This saves a second visit, but it is advisable only when intelligent and conscientious collaboration can be assumed.

The greatest methodological challenge in time use research is the choice of the scheme of classification of activities in terms of which the structure of everyday life is represented. Sorokin started to tackle this problem, but convincing theoretical or empirical criteria for constructing typological keys have not been found, and using conventional categories of ordinary language is not entirely satisfactory. Normative and/or contractual work arrangements suggest a fairly unambiguous specification of "market work," but there are some activities in the home that, according to circumstances, can be classified as either housework or leisure. This difficulty could be overcome if the respondent did the coding himself or herself, but usually the log of daily routines is described in the respondent's own words and codification is done by someone else, following criteria that exclude personal and/or subjective meanings. An even more fundamental issue is whether current classifications can be assumed to be meaningful in cross-cultural terms. Time use studies distinguish the minimal basic activity groupings of personal needs, formal work or education, household work, and leisure. The hidden dimension that is postulated by such groupings is obviously a reflection of the Western opposition between necessity and freedom of choice. It places market work immediately after biological needs and before domestic work, which is placed near leisure. This implicitly individualistic and work-oriented, contractual rationale probably is not well suited to representing the more solidarity-oriented temporal orders of everyday life in traditional societies (Bourdieu 1963).

Another difficulty concerns the level of specificity at which a common array of activities is reported at the collective level. The daily pursuits of persons who lead a busy life can be meaningfully recorded in great detail, whereas those of persons tied to the home usually have much less texture. One way to approach this difficulty is to construct hierarchical coding frameworks in which the first column in a multiple-digit code divides the day in terms of major classes of activities. Additional columns focus on increasingly more complex but exhaustive time accounts. The time use project coordinated by Szalai identified in its time-budget protocols ninety-six activity categories. For some purposes, these were reduced to thirty-seven and, for others, to the following ten main groups: work, housework, child care, shopping, personal needs, education, organizational activity, entertainment, active leisure, and passive leisure. Today the coding schemes for official statistical surveys often include many more basic activity categories because they have to accommodate the heterogeneity of lifestyles across gender groupings, generations, occupational categories, and rural versus urban residential environments.

Once time-budget data have been collected and coded, decisions about data processing and indicator construction can be made. According to the complexity of statistical data elaborations, there are three different levels of analysis (Stone 1972, pp. 96–97). First, activity arrays in terms of frequencies or durations, possibly taking company or locations into account, are cross-tabulated with the sociodemographic characteristics of the actors. Second, single activities and their positioning during the course of the day are studied. Finally, stochastic activity sequences are analyzed, focusing on the structure and rhythm of chronological daily routines. Since the 1960s, the following set of indicators generally have been used in computations: (1) the generic average duration of an activity, where the numerator refers to the total sample, disregarding whether there was involvement in the activity, (2) the rate of participation, or the percentage of interviewees who were involved in an activity, and (3) the specific average duration of an activity, where the denominator includes only those who have engaged in it.

Another issue in these surveys is the validity and reliability of data sets. Questions of validity can be raised by difficulties in recall and incorrect identification of activities among respondents or by possible alterations of spontaneous behavior after observation and the consequent distortions in reports or by research instruments that inadequately reflect the specificities of the observed sociocultural context. Research directed at data quality ( Juster 1985; Niemi 1993) has shown that results obtained by means of time budgets present at the aggregate level a high correlation with those obtained by means of other forms of observation, such as interviews, workplace or school statistics, and telephone surveys. Moreover, the hypothesis that the desirability or social prestige of certain activities or lifestyles could influence time use reports has not been confirmed. In general, it seems that the twenty-four-hour frame of reference helps reduce such effects and brings informal and often undeclared work commitments to light. Certainly, activities that are assumed to be of secondary importance, such as conversations and listening to the radio, are under represented in current summary tables that restrict the attention to "primary" time allocations. However, this cannot be considered an invalidating shortcoming. Nevertheless, to assure validity, time budgets presuppose the concept of rational time. If a population does not live by the clock, any calculation of time budgets is meaningless.

CONTRIBUTIONS OF TIME USE RESEARCH

Time use research has gained momentum because of interest on the part of international agencies in comparing the functioning of societies in their national settings, the need to connect demographic change and social development, the need to focus on gender-related or generational variables to understand the changing role of the family, the awareness that economic variables reflect wealth and well-being only in very partial ways and that household and care activities must be brought into focus, and the need to construct articulated databases for decision making about social policy.

Since the late 1980s, nationally representative time use data sets have been available for several countries, but the evidence is not easy to compare because activity classifications do not always coincide. Therefore, with the hope that many European Union countries will participate in the very

expensive data-gathering process, Eurostat is preparing a standardized survey. Up to the present, only Japan has truly comparable five-year time series data to indicate trends and changes in lifestyle since 1970 (NHK 1991). In terms of Monday-to-Friday behavior, for instance, sleeping time and housework have decreased while market work has not. Over the years, leisure on all weekdays, especially hobbies, private lessons, and sports activities have grown steadily. Television viewing time reached a peak in 1975 (probably because of the advent of color television) but returned in 1990 to the 1970 level. The illustrations confirm the rule that trend-setting evidence of new life styles is found not so much in the main activity categories but in apparently marginal activities.

To demonstrate the interest of time-budget data in a comparative assessment of the different logics of time structurization, it is best to choose an example from the uniform data set gathered in the Twelve Country Study.

In Table 1, the patterns of daily urban time allocation ascertained in 1965–1966 for the United States, France, and Hungary are presented in the most synthetic form. The data refer to an average weekday and compare time use for personal needs (mostly sleep and meals), market work or study, household work, leisure, and non-work-related travel among adults aged 18–65, subdivided by gender and employment status.

When one compares the starkly different research contexts of those days (market economy, welfare state, state-controlled system), a set of clear-cut differences in time use emerge from these different ways of life; at one extreme the United States and at the other Hungary, with France in between. The average duration of market work was much shorter in American cities than in Hungarian ones; this was due mainly to the fact that employed women were more likely to hold part-time jobs in the United States, while in Hungary they held full-time jobs. Across the three countries, household work and leisure show systematic secular trends. Employed American men enjoyed one hour more of leisure and contributed half an hour less to housework than did their Hungarian counterparts. For employed American women, housework lasted one hour less, and leisure lasted one and a half hours more than was the case for employed Hungarian women. Finally, household work among unemployed American women required two hours less time and leisure benefits lasted two hours longer than was the case for the corresponding group in Hungary. The uneven availability of household appliances and unequal access to leisure amenities (in particular, television) were the causes of these differences in lifestyles. In addition, the table reveals well-known inequalities in gender and employment status. If one combines the time invested in market work and housework, it appears, though less significantly in the west than in the east, that women in the labor force contributed a considerably larger share of work and enjoyed much less leisure than did employed men.

Table 2 shows that gender differences are clearly implicated in the discrepancies in the number of hours of economic activity and housework per week between men and women, indicating that there are analogous patterns of inequality in developing countries such as Bangladesh, India, and Nepal. In these countries, women are more likely to spend their time in subsistence activities, whereas men tend to have a monopoly of paid jobs. In addition, women's time investment in housework is six times that of men in Bangladesh and three times that of men in the other two countries. However, the gap in overall work hours, though disadvantaging women, is less evident. In fact, in Bangladesh they contribute 54 percent of the total micro-productive time input, in India 55 percent, and in Nepal 58 percent.

Data from the French national survey (Insee 1989) on the effect of cumulative social roles among women are much more analytic. In France, for mothers with husbands under 45 years of age and at least one child younger than 25 years old, increasing from one child to three and more children means, if they are unemployed, an increase of one hour and fifteen minutes of house work and, if they are employed, an increase of fifty-two minutes. Where does this extra time come from? In the case of especially pressured employed mothers, the time investment in human capital in the form of caring for additional children implies a reduction of one hour and thirteen minutes in the duration of market work, including a seven-minute reduction in free time and fourteen minutes less of sleep. Especially for women with more than one child, this negatively affects their competitive position in the professional world. Data such as these should be of interest to policymakers.

This example also shows that considerable caution is required in interpreting time use data. In general, estimates of differential time allocations for men and women in certain activities do not reflect only gender differences. Demographic variables, especially family composition, the structure of the labor force, and the availability of household help, intervene in causal links between gender and time use. Particular attention must be paid to these influences in longitudinal analyses such as that of Gershuny and Robinson (1988), which analyzed U.S. and British time-budget data over three decades. Statistically controlling for female labor force participation, male unemployment, and declining family sizes, those authors concluded that in the 1980s, women did substantially less housework while men did a little more than in the 1960s. In another study that analyzed data from repeated surveys in eight Western countries, a general reduction in time dedicated to all kinds of work was found, along with a convergence of time use models among males and females and a growing international similarity in the patterns of the division of time between work and leisure (Gershuny 1992).

In recent decades, considerable progress has been made in representing the multidimensionality of time use phenomena because official data, instead of regarding samples of randomly chosen individuals, have been collected from all members of households. This has made it possible to observe how husbands' time management affects their wives and vice versa and to determine what it means for families if both husband and wife are employed and if children come into the family nucleus.

In Table 3, pertinent data on couples from the national time use survey conducted in Germany in 1990–1991 are presented. It shows time use models for types of families defined by employment status and the presence of children. Assuming the operation of compensatory mechanisms, it also

takes the weekly rhythm of time use into account by distinguishing workdays (Monday to Friday) from weekends (Saturday and Sunday).

One model regards more traditional couples in which only husbands are employed and wives do most of the housework. Husbands increase their market work when there are children, but regardless of the presence of children, they defend their daily leisure and contribute to housework mostly on weekends. The other model concerns couples in which both partners are employed. In this case, it is not surprising that the wives' market work is considerably shorter than the husbands', but what is important is that in the presence of children, both partners increase their market work by more than one hour each. The housework of mothers increases on all days, whereas fathers limit increases in their domestic chores to the weekends.

From a micro perspective, what the German example shows are the implications of decisions made within the family. Here the family is seen as the institutional arena where partners search for a suitable compromise in their interlocking role definitions. From a macro perspective, the implications of gendered time use arrangements for the changing division of labor and the growing interaction between the market sector and the household sector are important. Sociologists and economists have often criticized the fact that mostly female domestic and caring activities, mostly male "do-it-yourself" repair initiatives, and voluntary and other socially useful work done by both men and women go unrecorded in labor statistics and national accounts. These are "productive activities" insofar as they can be delegated to persons other than those who benefit from them.

Table 4 shows a selection of the results of a United Nations Development Programme analysis of a posteriori standardized time budget data from the most recent national surveys conducted in fourteen different countries (Goldschmidt-Clermont and Pagnossin-Aligisakis 1995). This analysis distinguishes between market oriented System of National Accounts (SNA) activities considered in the UN System of National Accounts and non-SNA activities, and introduces the necessary controls for the demographic structures of the populations.

Despite the nonhomogeneous social structures and value systems of France, Germany, Great Britain, and the United States, everywhere statistically unrecorded (non-SNA) activities absorb about as much labor time as do recorded (SNA) activities. Furthermore, total economic time allocations (SNA plus non-SNA) tend to be equal among men and women. Although this demonstates social equality in general terms, it can be seen that very strong gendered divisions of tasks prevail in all cases. In fact, in these four countries, women contribute only one-third of total market-oriented productive time, whereas they contribute two-thirds of total non-market-oriented productive time.

The availability of comparable time budget data is a prerequisite for official statistics that aim

to include the production value of nonmonetarized activities, through "satellite accounts," in their quantitative frameworks. The most recent national time use surveys have been conducted with these applications in mind.

THE RELEVANCE OF TIME-BUDGET DATA SETS FOR SOCIAL POLICY

Sociologists have often been skeptical about the utility of time-budget research. Time budgets are thought to provide data that are "broad but shallow" (Converse 1972, p. 46) and offer no more than static, tendentially commonsense descriptions of only the manifest aspects of everyday life. From a theoretical point of view, it is argued that there is a lack of explanatory hypotheses and relevant concepts that could bring norms, experiences, attitudes, and values to the fore. On the methodological side, the main criticism is that classification schemes of activities are imprecise, are unevenly general or detailed, and have barely changed since the 1920s (Pronovost 1989, pp. 78–80). Implicit in these observations is the dilemma Szalai faced earlier: whether priority should be assigned to the testing of hypotheses or to multipurpose database construction. With the latter option comes the question of how to reconcile, in designing the studies, cross-national and longitudinal comparability and adherence to sociocultural settings and historically changing conditions.

Time-budget research is applied research that has increasingly been aimed at the design and evaluation of social policy. Its relevance in this context derives from the fact that in modern affluent societies, citizens often value scarce time more than material or monetary resources; thus, time use rationalization and efficient time management in the personal, family, and public sphere have become matters of general concern. Contingencies curtailing time use evidently are distributed unequally in the social world. For instance, health checks in public institutions often involve waiting times that private medical care does not, and not owning a means of transportation makes long commuting times unavoidable. Hence, social policies and their provisions try to make circumstances or opportunities more equal for everyone. Part-time employment, flexible worktimes, and compressed workweeks have been introduced mostly for pressured working mothers with small children.

Shop-opening hours and office schedules have been changed to permit effective coordination and a better reconciliation of tasks. The effectiveness these measures can be monitored with the help of time-budget procedures.

The complexity of time-budget data sets has often been insufficiently exploited. Initially this was due to limitations in handling enormous data sets, but increased technological resources and new multivariate statistical techniques have opened new frontiers. Almost exclusive attention has been given to average durations and frequencies of primary activities, but the study of configurations emerging from an association of these activities with other contemporaneous activities promises a better understanding of modern time regimes. Until now, not much research has been done on routinized rhythms or the strategic sequencing of activities. Also, the collaborative or conflicting interface of the various schedules of family members, the reconstruction of networks of participative personal contacts (a topic of great significance in regard to lonely children, the ill, and the elderly), and the relationship of the use of urban spaces to time use (a problem studied by human geographers) are all interesting areas for future research because of the greater availability of important data sets.

Activity classifications will have to undergo critical study to better reflect changes in the activity patterns of everyday life. Statistical offices are presently reconceptualizing their taxonomies. Paid work might be broken down into its constituent parts and examined analytically, but other activities need redefinition. For example, some kinds of domestic work have been absorbed by the market; care activities now regard the elderly more than children; dealing with service bureaucracies has become a time-consuming task; there is a new spectrum of voluntary forms of participation at the social, political, and cultural levels; and leisure behavior has changed in relationship with new media.

However, the problem is not just a technical one regarding exclusively descriptive coding schemes. Time use data assume importance only when they provide a valid epistemological key for the interpretation of social change. As has already been pointed out, earlier lines of research identified two central components of time use: market work and leisure. Today, time use studies based on data from household samples may help identify other valid criteria of time use to better understand how families cope with growing structural unemployment and increasing social insecurity.

references

Bevans, George Esdras 1913 How Workingmen Spend Their Spare Time. New York: Columbia University Press.

Bourdieu, Pierre 1963 "La Société Traditionelle: Attitude à l'Égard du Temps et Conduite Économique." Sociologie du Travaille, 5:24–44.

Converse, Philip E. 1968 "Time Budgets" In David L. Sills, ed., The International Encyclopedia of Social Sciences, vol. 16. New York: Macmillan—Free Press.

Elias, Norbert 1988 Time. New York: Basil Blackwell.

Gershuny, Jonathan 1992 "La répartition du temps dans les sociétés post-industrielles." Futuribles, Mai-Juin, 215–226.

——, and John P. Robinson 1988 "Historical Changes in the Household Division of Labor." Demography 25(4):537–552.

Goldschmidt-Clemont, Luisella, and Elisabetta Pagnossin-Aligisakis 1995 Measures of Unrecorded Economic Activities in Fourteen Countries, Occasional papers no. 20. New York: UNDP.

Harvey, Andrew S., Alexander Szalai, David Elliott, Philip J. Stone, and Susan M. Clark 1984 Time Budget Research: An ISSC Workbook in Comparative Analysis. Frankfurt: Campus.

—— 1993 "Guidelines for Time Use Data Collection." Social Indicator Research 30:197–228.

INSEE 1989 "Les Emplois du Temps des Français." Economie et Statistique 223.

Jahoda, Marie, Paul Lazarsfeld, and Hans Zeisel 1971 Marienthal: The Sociography of an Unemployed Community. Chicago: Aldine (German orig., 1933).

Juster, F. Thomas 1985 "The Validity and Quality of Time Use Estimates Obtained from Recall Diaries." In F. T. Juster and F. P. Stafford, eds., Time, Goods, and Well-Being. Ann Arbor: Institute of Social Research, University of Michigan.

Kurtz M.-F. D. 1984 "Les Budget-Temps: Réflexion Épistémologique," L'année Sociologique 34:9–27.

NHK 1991 Japanese Time Use in 1990. Japanese Broadcasting Corporation, Public Opinion Research Division.

Niemi, Iris 1993 "Systematic Error in Behavioural Measurement: Comparing Results from Interview and Time Budget Studies." Social Indicators Research 30:229–244.

Przeworski, Adam, and Henri Teune 1970 The Logic of Comparative Inquiry. New York–London: Wiley Interscience.

Pronovost, Gilles 1989 "The Sociology of Time." Current Sociology 37(3):1–129.

Scheuch, Erwin K. 1992 "The Time-Budget Interview." In Alexander Szalai, ed., The Use of Time. The Hague–Paris: Mouton.

Sorokin, P. A., and C. Q. Berger 1939 Time-Budgets and Human Behavior. Cambridge, Mass.: Harvard University Press.

Statistisches Bundesamt 1995 Die Zeitverwendung der Bevoelkerung, Tabellenband I. Wiesbaden.

Stone Philip J. 1972 "The Analysis of Time-Budget Data." In Alexander Szalai, ed., The Use of Time. The Hague–Paris: Mouton.

Sue, Roger 1994 Temps et Ordre Social. Paris: Presses Universitaires de France.

Szalai, Alexander, ed. 1972 The Use of Time: Daily Activities of Urban and Suburban Populations in Twelve Countries. The Hague–Paris: Mouton.

—— 1977 "The Organization and Execution of Cross-National Survey Research Projects." In Alexander Szalai and Riccardo Petrella, eds., Cross-National Comparative Survey Research. Oxford, UK: Pergamon.

—— 1984 "The Concept of Time Budget Research." In Andrew S. Harvey et al., eds., Time Budget Research. Frankfurt: Campus.

United Nations 1995 The World's Women 1995: Trends and Statistics. No. E.95. XVII.2.

Zerubavel, Eviatar 1982 "The Standardization of Time. A Sociohistorical Perspective." American Journal of Sociology 88 (1):1–23.

Elke Koch-Weser Ammassari