Cover Image
close this bookResearch Methods in Nutritional Anthropology (United Nations University, 1989, 201 p.)
close this folder4. The relevance of time-allocation analyses nutritional anthropology: The relationship of time and household organization to nutrient intake and status
View the document(introduction...)
View the documentIntroduction
Open this folder and view contentsMethods of data collection
View the documentProcessing and interpreting data
View the documentProcedures for collecting and analysing time-activity data efficiently and effectively
View the documentConclusions
View the documentReferences
View the documentAppendix 1. Messer, mitla field notes, 4 September 1980

(introduction...)

Introduction
Methods of data collection
Processing and interpreting data
Procedures for collecting and analysing time-activity data efficiently and effectively
Conclusions
References
Appendix 1. Messer, mitla field notes, 4 September 1980

ELLEN MESSER
World Hunger Program, Brown University, Providence, Rhode Island, USA

Introduction

The purpose of this paper is to indicate how studies of time allocation can be combined with nutritional studies to help illuminate household and general socio-economic conditions that impinge on nutrition and health.

The organization of time is a significant feature of any life-style. The ways people classify intervals in time, schedule activities according to these intervals, and fit such activity schedules into time comprise aspects of their time sense, which varies from culture to culture.

What people do in time and how they time what they do are aspects of cultural identity. Usually a distinction is made between Western industrialized societies in which "time is of the essence" and non-industrialized societies in which "productivity" is evaluated in terms of social relations and cultural values in addition to material output "per unit of time." Yet even in nonindustrialized societies, work demands at particular times may create time conflicts as different activities, such as food production and household maintenance, compete for time and create bottlenecks in individual and household schedules. The priorities people set in expending time, the urgency with which people complete tasks, and the ways in which people perceive their seasonal cycles of production and consumption are all features of their strategies for survival.

Time budgets and time-activity patterns are useful analytical units for understanding how people produce income and meet household needs in any socio-cultural group. Time and activity data can, for example, provide significant measures showing how the organization of individual or household activities, along with other aspects of the environment, contribute to the food provisioning and the nutritional and health maintenance of individuals, households, and communities. Studies of the organization of time, from scientific and folk perspectives, can identify many of the factors that affect what, when, and how often people eat. Additional comparisons of activity patterns among individuals (categorized by age, sex, or occupation) and households (classified according to occupational type, income level, numbers and ages of children, or family structure) can also further our understanding of why some households or cultural patterns succeed and others fail nutritionally under apparently the same environmental conditions.

Some Applications of Time - activity Data for Nutritional Studies

Anthropologists, psychologists, and economists have been collecting time-activity data for years to answer questions within their disciplines. Time-activity data have become increasingly important for nutritional studies as (a) functional measures have been increasingly emphasized in nutrition and (b) as the research perspective has broadened beyond studies of the individual to comparisons of the effects of different individuals on the nutritional status of others in their households. Nutritionists as well as anthropologists are now looking at nutrition in its entire social context to understand the important linkages between social and household variables and nutritional well-being. These questions, which form the basis for an ongoing study of the relationship between nutrient intake and function (Calloway et al., 1979), are essential if we are to understand the relationships between nutrition, socio-cultural organization, and various categories of work and leisure performance. Projects have been organized in terms of the relationships between nutrient intake and: disease-immune response; reproductive function; work capacity, i.e. physical function; psychological (cognitive) function; and social function (Calloway et al., 1979). To understand these relationships, one needs data on nutrient intake as well as (a) typical household tasks, household organization of tasks, typical times it takes to complete certain sets of tasks, and activity schedules of the focal female and others around her; (b) meal scheduling and the manner of acquiring, preparing, and serving food; (c) sources of infection; and (d) patterns of child care in households. Additionally, one needs to know where to look for linkages between social organization, time use, and biological well-being.

Some of the approaches anthropologists and other social scientists have used to study the relationships between time allocation and nutritional status include careful assessment of seasonal and household differences in time use. Sharman's (1970) investigation of the relationships between food beliefs and practices, income, and nutritional status in rural Uganda, for example, shed light on several pertinent issues. She concluded that social organization of household labour, including how women expended time and other resources, appeared to be a significant factor in predicting the nutritional outcome of children. In this society, management of resources, not just their availability soon after the harvest, determined whether there would be sufficient food for household members throughout the year.

Sharman found it was crucial to examine co-operation within and between household units. She noted that the generosity, wastefulness, and frugality of the various household managers were critical factors, as some women spent much of their time and substantial food resources early in the post-harvest season preparing beer for sociability groups. The expenditures depended on their desire for this form of sociability and on the amount of household labour available for their organization. Also, Sharman found that men and women who spent most of their time working other people's fields ate fewer meals per day, and used nutritionally poorer ingredients that were quicker to prepare. Moreover, their children tended to scavenge more. The nutritionally poorest staple and relishes (vegetable-protein complement) were, furthermore, those that required the least time to prepare.

"Mothercraft" - time management and household child management - has also been suggested as a significant variable in other nutritional studies. Evaluating effects of a mother-child health programme in Cali, Colombia, Franklin and Vald(1979) quantified the relationship between mother's time and abilities and children's nutritional status. They considered mothers' use of time and food (measured in calories) as complementary scarce resources, which together could produce more or less acceptable levels of child nutrition. Employing a regression model, their data support the hypothesis that a mother's time and abilities, not just food availability, significantly influence child nutrition.

Time allocation of focal females may also affect nutritional well-being via an interaction between nutrition and infection. Kumar (1977), in an investigation of the relationships between occupations, incomes of women, and children's nutritional status, studied how seasonal conflicts in time allocation between agricultural tasks and child care could periodically subject children to infection and nutritional stress. She noted how, during peaks in seasonal work in the fields, women were inhibited from spending adequate time providing food and health care for children, who were, in turn, cared for by others, usually older siblings. Infants and very young children subjected to this periodic neglect while their mothers joined harvesting teams in the fields might not be adequately fed or protected from the sun, and often became ill. As a result, children of those mothers who spent the most time working in the fields and, therefore, earned higher incomes were often worse off nutritionally than those of mothers working shorter hours and earning less. The results of time-allocation observations could document, then, how women's work, given the social organization of child care, is related to child nutrition and health maintenance. Such studies might explain the seemingly skewed negative relationship between mother's income and children's nutritional and health status in some socio-cultural situations.

Studies of household labour organization and time allocation by all members have also furthered understanding of the variables intervening between levels of income and nutritional outcomes of a household (e.g. Nag, White, and Peet. 1978; Quizon-King, 1978). Such timeallocation records can indicate (a) how the labour of all household members, including children, contributes to household maintenance, income, and nutritional and health status, and (b) how time allocation by women for work outside and inside the home interferes with child feeding and child care. Time-allocation records can be particularly useful in examining the health and nutritional consequences of women's seasonal labour patterns and surrogate-mother child care. Women working outside the home have less time for breast-feeding and child care but more income to buy food: are their households better or worse off?

In a study by Tripp (1981) of Nigerian women, households of women working outside the home were better off. Those women who worked as traders had children who showed fewer signs of malnutrition than women not so employed. He concluded that women who earned their own income invested it mainly in food for their children, which accounted for their children's superior nutritional levels.

In contrast, Popkin and co-workers (Popkin and Solon, 1976; Popkin, 1980), in a series of studies, suggested that women who work as traders in the Philippines have children who are less well-nourished in terms of vitamin A status than their nonworking counterparts. They suggest that the reason for the vitamin A deficit is that working mothers have less time to prepare leafy green vegetable sauces that are rich in vitamin A to go with the dietary staple, rice, than those who spend their time gathering such ingredients for subsistence. Thus, as women enter the cash economy, their purchased diet may be inferior to the home-cooked meals they no longer have the time to prepare. However, none of the studies by Popkin and co-workers provides dietary and time data sufficient to support such a view. Such data would include when work took place and when income was a available over periods of a day, a season, and a year and would provide more adequate information than occupation and income categories.

An important use of time studies is in evaluating the actual or potential nutritional impact of socio-economic development policies. For example, some policies can be anticipated to have adverse effects on women by increasing their workloads, which may already average more than ten hours a day (Scrimshaw, 1982). Where men are lured into non-agricultural occupations, women may be left to run household farms completely alone. Superbreeds of cows, pigs, and poultry, which require more forage and water, make more work for women who often must carry these items. Fertilized crops grow more weeds, and women must pull them. Insecticides require water, and women must draw it (Fortmann, 1981, pp. 211-212). Precise calculations of such increases in "drudgery" and their impact on women's ability to care for and feed their households can only be obtained through time-allocation studies.

Water and food-processing technologies can also be evaluated in relation to their effects on women's time allocations. In what seasons would such labour-saving devices as mills be most desired? How have traditional social arrangements managed to decrease the time needed for food preparation in certain high labour-demand seasons? What are the effects of technological changes on such social arrangements?

Hemmings-Gapihan (1981), in her study of the potential impact of a solar energy unit in Upper Volta, used small samples of women to observe the amount of time they spent drawing water and pounding grain in different seasons - tasks that would be facilitated by the introduction of a solar energy cell. Time allocation by women to household and food-production activities, as well as leisure, was recorded to determine the effects of modernization on household structure, the social organization of labour, and women's work. Such examples of time-allocation analysis are important for understanding the relationships between household organization and nutrient intakes of members.

For advancing the productivity of women, preliminary studies of women's time allocation and patterns of household decision-making regarding all aspects of food provisioning can yield significant information about potential food increases that could result from introducing new forms of income production. If the labour time to clean, water, and feed household animals, as well as the daily rations of time and cash to supply them with food, could be shown not to interfere with women's time allocated to feeding and providing health care for human beings (including their own health and well-being), then development policies that included animal husbandry might be beneficial. Such policies might be particularly important in instances in which women's income seems to go directly and almost exclusively into providing more food for their households.

Time data on scheduling work and household tasks can also help in the development of programmes, including mother-child health and nutrition projects. Such projects might schedule activities to least interfere with household production routines. Preliminary background data for programme development involve observation of whole days and timed observation of selected activities in a limited number of households over short periods of time in different seasons.

As Nevin Scrimshaw, among other perceptive nutritionists, has noted, it is unwise to design health and nutrition programmes that will add yet another burden in time to the already overworked female running a household. He has suggested, as an example, that national and international agencies planning nutrition interventions in the form of supplementary foods consider a low-cost processed food. Such a food would not require elaborate food preparation that would compete with other activities in a mother's already full day (Scrimshaw, 1982). The appeal to health and nutrition planners to pay more attention to the value of women's time in planning their programmes and evaluating the reasons for low participation rates is customary in the anthropological literature. Again, selective observational methods, perhaps combined with selective interviews focusing on time constraints, could aid in the wise design of nutritional studies.

Literature Review

Collection of time-use-activity data is not new to anthropology, economics, socioeconomic planning, or developmental psychology. Within ecological anthropology, for example, human ecologists, focusing on the human population as their unit of analysis, have timed activities as an integral part of their investigations tracing the flow of energy and other nutrients through local and regional ecosystems. Using time-motion studies to estimate human energy expenditure in food production (Rappaport, 1967), or estimating time and energy spent in production relative to consumption of food (Lee, 1969), anthropologists have calculated the relative effort expended in the food quest under varying conditions of resource availability and social organization.

Socio-ecological anthropologists, to construct or reconstruct "optimal foraging strategies" for human social groups in particular environments, have used the method of linear programming to calculate the hypothetical time and effort (energy) expenditure required to obtain resources of specific nutritional value (see Winterhalder and Smith, 1981). In explaining deviations of actual from expected values, they have often suggested that factors other than efficiency or ease of energy-calorie acquisition are valued by societies in their food quests. Tastes, especially for the fat in food, or political factors, such as the relative safety in times of social upheaval of sowing and harvesting crops with a given growing time and product return, are dimensions that can be useful in predicting and comparing food choices within and across cultures (Jochim, 1981).

Primatologists, focusing on the social group rather than the ecological population as their unit of analysis, have used time-frames to compare and contrast the frequency of feeding of adult males versus females or immature members or either sex in primate societies. Following single individuals, or censusing at regular intervals whole social groups, they have analysed how relative ranking affects the efficiency of feeding and the nutritive value of the intake of individuals or classes of individuals within primate groups. For a review of the methods of observation used in primate studies, their comparability, and relative merits, see especially Altmann (1974) and essays and appendices in Clutton-Brock (1977).

Similar methods have been used in socio-cultural studies. Sharman (1980), in a research design reminiscent of methods used in primate studies, analysed the "frequency and timing of eating events, the extent to which households eat together and eat the same thing; and the general format and content of meals." She examined the relationships between dietary choices and allocation of resources, both income and time, within households, with special attention to the nutrition of children under five, the effect of low income, and the impact of a women-infant-child feeding programme (WIC). Her preliminary study resulted in a series of research questions directed at improving the quality of reporting about the social context of nutrition and nutrition interventions: What are the complete eating patterns of household and individual members? What is the relationship of work schedule to eating patterns? What is the relationship of income, time, and shopping/food expenditures? Also, she suggests a route to dietary analysis in modern complex societies that would reflect factors such as the shortage of time, the dislike of cooking, the timing and frequency of meals, as dependent on domestic chores, visiting, and cash-work patterns on food choices, expenditures, and consumption patterns.

Economic studies by anthropologists and other social scientists have also involved research on time use. Particularly those studies investigating how people's time schedules affect their productivity, health, and quality of life, or, reciprocally, how productivity, health, and quality of life affect people's performance of various types of activities, have contributed data and methodologies. The pre-Second World War social anthropological study of the Bemba of Northern Rhodesia by Audrey Richards (1939), for example, was designed to answer the British colonial economist's questions about why natives did not work harder and whether it was possible to increase local food production.

Richards investigated how Bemba men, women, and children scheduled their activities, the time required for particular activities, and the relationships between social organization, work performance, and the local diet. On the basis of her observations of task performance by different women over several seasons, she was able to suggest how insufficient nutrient intake by adult women during the lean season, which was also the planting season, affected their ability to perform both productive tasks and household maintenance. She was also able to identify ways in which the nutritional deficiencies of adult women may have interfered with current and prospective production and nutrition of all household members. She blamed the situation on the British colonial interests, which had drawn men away from their home villages to the mines, and left the women with the entire burden of clearing new fields in addition to their usual gardening tasks.

Following Richards, more recent studies have addressed questions concerning the household factors, including organization of female labour, that affect production and distribution of household resources and, by extension, the nutrition and health of household members. Timing the performance of common subsistence activities across households forms a part of investigations (e.g. Sharman, 1970). Also, social anthropologists have tried to develop appropriate units in work and time by which to calculate the costs to various types of production, for example, how many persons with what technological instruments are needed to plough a certain class of field (Wallman, 1969).

Other anthropological studies drawing on time-activity data have addressed the questions whether people work "harder" (or, at least, longer) in some cultures (Gaulin and Konner, 1977) and how the industrialization of production lightens or increases workloads (Minge-Klevana, 1980). In related studies, time-allocation profiles of men, women, and children have also contributed to economists' understandings of socio-economic factors operative in women's fertility decisions and their decisions to enter the paid labour force (summarized in Evenson, 1981). Time use has also been a dimension in general social comparisons between societies (Szalai, 1972), and time-activity data have been collected to further recreation facility planning by showing how people prefer to allocate leisure time. The studies of Chapin (1974) on leisure-time activities in the greater Washington, D.C. area have added to the corpus of interview methods used in time-activity studies.

Psychological studies, in contrast, have employed time-frames to observe how parents interact with children, or, alternatively, how children "interact" with their environments as part of more general studies of child development (Whiting and Whiting, 1975). A few of these studies have investigated, in addition, child behavioural development in relation to nutritional supplementation of the mother and/or the child (see Herrera et al., 1980).

More sociologically oriented studies have dealt with related historical and comparative cultural questions of the changing perception and quality of time in different historical contexts (Moore, 1963; Thompson, 1967). The notion of time as an economic commodity to be rationed, spent, or wasted and the temporal dimensions of social and economic life are important qualifying factors for any quantitative study of time use. "Non-economic" allocations of resources, including time and energy allotted to ritual, social, or other economically "nonessential" activities must also be accounted for in arriving at an accurate depiction of the interrelationships between "work" and nutrient intake and nutritional status.

The differing time senses of industrial versus non-industrial societies have been dealt with in a number of historical studies (Thompson, 1967) and ethnographic reviews of societies and cultures undergoing technological transformations (Mead, 1955). These suggest that, outside Western industrial societies, one should expect no simple relationship between time allocated to work and household productive income, or, more to the point, no simple way to sample and measure "time use" to predict economic and nutritional well-being.

Pacing of activities over daily, weekly, seasonal, and yearly cycles may be determined by competing internal (household) and external (economic) demands, both natural and sociocultural in origin. For example, demands on women's time may be seasonally shaped not only by the need for female labour in planting and harvesting, but also by the cultural rules for the social organization of labour allocated to each task. How much effort should be expended by particular classes of individual - i.e. how hard and how long people work to meet culturally defined goals - is also a product of cultural training, expectations, and the cultural food system. These factors affect measurable physiological potentials of individuals to perform work, as well as standard tasks and standards for their performance.

Studies of ritual time are also important. For example, the ritual observance of one day of rest and heartier food consumption in every seven in many Jewish, Christian, and Moslem societies which are otherwise marginally nourished, and periods of ritual preparation and consumption in various societies, suggest that simple averaging of time (and activities) over a number of days will not yield a meaningful picture of how time-use patterns are related to the nutrition of individuals, households, and communities. Any study which proposed to use timeallocation analysis as a means to describe and compare social and economic conditions must, therefore, begin by collecting preliminary data on the "social value" of time and its implications before proceeding further.

With these caveats on qualitative time factors in mind, I will proceed in the next section to examine several possible ways to study time allocation quantitatively.

(introduction...)

Random or spot observations
Day-long observations

 

Observation, interview (respondent recall), and diary recording (respondent report) are the three basic methods by which to obtain time-activity data. Research designs may employ activity sampling techniques that are either opportunistic, random, or targeted. Whichever method or combination of methods is used, information must be recorded in standardized activity categories appropriate to the culture and research questions at hand. Whose time one follows, at what intervals and by what method depend on the initial research questions and the characteristics of the study population, particularly their "time sense" and willingness or ability to co-operate. The availability of research assistants to collect, process, and interpret the information, which can be collected by the various methods, and any additional purposes for which the data might be used, are also considerations.

The results of activity analysis can be used to supply information on the daily schedules of individuals and households, productive work inside and outside the household, how time is allocated to different work and leisure activities, and a number of other social and economic questions. Therefore, it is important to be as sure as possible of the range of one's research questions in advance. Certain methods, e.g. random sampling designs, usually preclude information on sequencing and co-ordination of activities. To chart activity profiles and levels of children, particularly very young children who may be the target of nutritional concern, probably only extensive observations or experimental observation frames can provide the information needed.

Furthermore, in all cases, research questions and settings determine whether one follows, for example (a) all members or social categories in a household or large social unit, (b) focal women or men, or (c) all children of a certain age. Also, the manner in which other types of information are collected, such as dietary intakes, energy or resource expenditures, and nutritional status - whether simultaneously with "time"activity data or subsequently - must also be carefully planned, so as to produce accurate information without upsetting normal household routines. Alternative strategies for accessing information on these and other questions will be considered in the following sections.

Preliminary Information

As mentioned above, observation and interview in the socio-cultural group should always precede the systematic investigation of time allocation among households. Preliminary study will help to establish the ecological, economic, and social context in which households operate and can contribute to determining significant criteria for classifying households and individuals for further study. In addition to background information, including demographic, health, sanitation, and socio-economic data, preliminary assessments should specify: (a) income-producing and other activities by men, women, and children of different ages or other social divisions; (b) the decision-making processes for allocating time and income for food acquisition and processing; (c) the decision-making processes for selecting food; (d) the social organization of food procurement and preparation; (e) the social organization and decision-making for partitioning food and eating; and (f) extra-household food-consumption patterns for various members. Seasonal factors in food supply, dietary patterns, activity schedules, and health should also be determined in advance to ensure adequate sampling. All these observations may form part of a more general ethnographic report on the community and contribute to related studies of physical activities, work output, social organization, and community functioning in relation to the food supply.

From these initial data should also emerge enough data on the time sense of the population, relative visibility of members performing an array of tasks at different times, culturally appropriate activity categories, and the relevant social unit or units for the study, so that one can proceed to select a workable method of data acquisition and a sampling design appropriate to the research questions and cultural setting.

Direct Observation

Direct observation is the method most commonly used by anthropologists to provide a description of the activity patterns, food habits, and social organization of a study population. If frequency, scheduling, and duration of tasks are the objects of investigation, direct observation is the only method that can be used among populations with little clock sense of time. Anthropologists often use the method of participant observation to describe the typical tasks, social relations, and nutrition and health components of a culture. More targeted observations may be needed for studies investigating the ranges of time and energy required to perform specific tasks, such as clearing gardens, weeding and harvesting, marketing, and breast-feeding. These studies may require focused, timed observations of selected individuals. Longer-term observations of particular households may be needed to answer research questions on such topics as residence rules and household dynamics that facilitate infection responsible at least in part for malnutrition.

Already cited above, the ethnographies of Lee (1969), Rappaport (1967), and Richards (1939) include examples of how time-activity data, collected by observation, can be analysed toward nutritional ends. Lee, working among the !Kung San of the Kalahari Desert, tried to demonstrate food-energy returns relative to labour-energy expenditure for food by computing the number of man-days of work as a percentage of man-days of consumption (table 1). For this calculation, he observed the daily activities of the San living at the Dobe waterhole for one season (medium dry, neither the best nor the worst) over a period of one month. Stratifying the population according to age and sex, he recorded adult activities - hunting, gathering, visiting, and resting at home - and found adults worked an average of 2.5 days out of 7, with a working day of about 6 hours in length.

While Lee's method is ingenious, he does not make clear how he counted "work" or, for that matter, "consumption." Children, for purposes of the study, are considered adults, yet time allotted by them to foraging and the food energy they acquire, which may make up a significant component of children's time and intake, are not specifically included in this calculation. Lee provides figures on how much time it takes women to crack some quantity of mongongo nuts, a major item in the diet, and how much time men spend on different types of hunting, but it is not clear whether he includes time spent in transit in this "work" figure.

Lee also notes that work patterns are erratic. For example, following a streak of bad luck, hunters may abandon the chase for a month or more and turn to visiting and dancing. He discusses the "cost" of short versus long-distance trips, but it is unclear whether this unlabelled unit of effort refers to energy expended, time, or inconvenience. Also, time spent going from camp to camp in search of sociability and food is not figured into the "work" of the food quest. Nor is there an adequate calculation of time given for and nutrient intake taken from "snacking," non-food sharing behaviour (see also Hayden, 1981). Thus, "work" in the food quest is probably underestimated. Work, in total, is underestimated still more, since it is not clear how Lee accounts for crafts or other activities. While he can discuss seasonal differences in food and water availability, and how these affect the food quest (time spent gathering food according to proximity, density, and diversity of foodstuffs), there are no systematic seasonal comparisons. Although the method of calculating how many days of work per week are needed to feed a certain number of people is suggestive, it is incomplete.

Rappaport's (1967) study of the Tsembaga sweet-potato gardeners/pighusbanders of highland New Guinea provides a second example of time-allocation calculations used in human ecological studies. Data were gathered on the entire population of 204 individuals, and the nutrient flow/ecological balance of people and the environment were analysed, in general, within a 20-year ritual cycle. To calculate dietary intake in energy and selected nutrients, Rappaport observed intensively daily consumption over a 10-month period in four households, composed of 16 individuals. To arrive at figures of energy expenditure in gardening, he observed the time it took to complete various activities, converted these figures into energy expenditure, and used the figures to calculate returns in energy per work hour.

Table 1a. Dobe work diary: a record of the activities at the Dobe camp for the 28-day period 6 July-2 August, 1964

      Adults (1) Children
(2)
Man-days of consumptiona
(3)
Man-days of work
(4)
Meat output (Ibs)
(5)
Week   Date          
1 July 6 18 9 27 9 -
7 14 9 23 6 92
8 15 9 24 2 -
9 15 9 24 3 12
10 16 9 25 7 -
11 18 11 29 3 -
12 18 9 27 7 -
2b   13 20 11 31 5 -
14 16 9 25 0 -
15 16 9 25 1 -
16 14 9 23 0 -
17 19 12 31 11 80
18 17 9 26 3 -
19 23 14 37 2 -
3   20 26 14 40 9 110
21 24 11 35 3 24
22 19 13 32 3 -
23 18 11 29 4 27
24 23 13 36 10 16
25 22 10 32 6 -
26 24 12 36 7 -
4 Aug. 27 22 13 35 12 7
28 27 13 40 12 80
29 26 13 39 9 10
30 24 11 35 16 12
31 22 10 32 4 20
1 24 11 35 8 -
2 22 11 33 16 -

a. Each entry in column 3 equals the sum of the entries in columns I and 2 for the given date.
b. Week 2 (13-19 July) shows an unusually low work output. The investigator contributed food on 12 and 17 July, resulting in a decreased subsistence effort for the seven-day period. Week 2 therefore has not been included in the final calculation of the S ratio.
Source: Lee,1969.

Table 1b. Summary of Dobe work diary

Week Mean
group size (1)
Adult-days (2) Total man-days of
consumption
(4)
Man-days of work (5) Meat (Ibs) (6) Work week (7) Index of
subsistence
effort (8)
1 (6-12 July) 25.6 114 179 37 104 2.3 .21
  (23-29)            
2 (13-19 July) 28.3 125 198 22 80 1.2 .11
  (23-37)            
3 (20-26 July) 34.3 156 240 42 177 1.9 .18
  (29-40)            
4 (27 July-2 Aug.) 35.6 167 249 77 129 3.2 .31
  (32-40)            
4-week totals 30.9 562 866 178 490 2.2 .21
Adjusted totals 31.8 437 668 156 410 2.5 .23

Source: Lee, 1969.

Time calculations were based on time-motion studies of small numbers of individuals. Two individuals were observed fence-making and seven individuals on three separate days were observed clearing underbrush for gardens. He carefully described the nature and pace of tasks, and, when performance was not uniform, he described his observations. He observed that the tempo, except in the case of the frail and elderly, was quite uniform; and the tempo of the same actors was also consistent. Men worked continuously at a set pace, with the only break a protracted stretch around midday. He then calculated the area cleared per hour.

He assumed that the more muscular men cleared more rapidly, and "eyeball impressions" confirmed this assumption (table 2). No systematic data were supplied on the time and scheduling of women's and children's activities, although he does suggest that the time is ripe for slaughtering pigs when they complicate women's activity schedules by getting into their gardens.

Audrey Richards' study (1939) of the Bemba of Northern Rhodesia provides a third example of the ways observational methods can be arranged to study native time use. Tables 3a to 3e provide an example of the methods she used to describe Bemba time allocations to different food system activities. Working in two villages over very short periods of time, she carefully planned to include data on how long it took to complete a typical task, such as building a fence or making a garden, average work days for men and women in different seasons, the amount of time it took to prepare food, including processing grain and making relishes, and the timing, structure, and content of "meals" and other food intake. She included these observations as part of a more general study of the social relations surrounding food production, distribution, and consumption. Interview and diary records were used to check her impressions based on observation.

Table 2. Clearing underbrush - time and motion study

Worker's name Sex Weight in Ibs. Time No. of strokes Time No. of strokes Time No. of strokes Comments
Akis M 88 10:37- 296 11:14- 250 12:14- 248 Only one 3-minute break during period. Next longest break: 15 seconds
      10:43   11:20   12:20    
Acimp F 85 10:55- 244 11:30- 209     No breaks longer than 20 seconds during working period
      11:01   11:36        
Avoi M 94 11 :20- 177 11 :50- 190     Slower than other workers because of short breaks, and slower strokes
      11:08   11:56        
Med> M 120 6 min. 233         Longer strokes than any of the others
Wale F 76 9:53- 246 10:58- 260     No breaks longer than 20 seconds during working period
      9:59   11:04        
Nimini M 96 6 min. 246          
Mer M 94 6 min. 316         Stated that he was in a hurry

Source: Rappaport. 1967.

Table 3b. Estimates of male and female labour required for each activity (in weeks, by calendar period)

Garden of present year to be cut 4 weeks, during May-September
Garden of present year to be fenced 4 weeks, during February-March
Garden of last year to be dug up for groundnuts or other crop 2 weeks, during August-October
Garden of three years ago to be burnt and prepared for third-year millet 1 week, during August-October
Garden of four years ago to be dug into mounds for beans or other crops 2 weeks, during October-January
Total 13 weeks as a bare minimum, of which 8 can only be done by male labour
In the village beds the labour expended can be roughly estimated as follows:
Digging of new village beds 2 weeks, usually during December-February
Complete rehoeing of old beds 3 weeks, usually during December-February
Or light hoeing of old beds 1 week, usually during September-February
Burning, weeding, etc. 1 week, usually during September-October
Total 4-6 weeks

Source: Richards, 1939, pp. 396 - 397.

Table 3c. Tasks in meal preparation for different staple foodsa

Porridge: millet Maize Cassava
Threshing/pounding Pounding Soaking
Sifting Repounding Chopping
Grinding Grinding Dry pounding
Cooking Cooking Cooking

a. Comments on time-attitude requirements: Millet and maize are prepared one day at a time, and it is not necessary to plan ahead. Cassava requires a 4- to 6-day routine, and one must plan ahead to have some roots soaking, some drying, and some being pounded into flour at any one time.

Source: Richards, 1939.

Table 3d. Time required for meal preparation

Task Comments Time required
Firewood-collection   1/2 hour
Water-fetching   1/2 hour
Relish-collection In gardens, bush 2-3 hours
Porridge-making Thresh/grind 3/4 to 1 hour
Relish-making Vegetables, nut sauce 1 to 11/2 hours
Total (for one daily meal)   Three hours

Source: Richards, 1939.

Table 3e. Timed processing of staple grain by seven women in Kampamba village (mins)a

  Flour in lb. Threshing Winnowing Grinding Total Time per lb.
1 43/4 12 14 35 61 12.8
2 71/4 10 13 30 53 7.4
3 111/4 14 15 55 84 7.5
4 91/4 10 15 42 67 7.2
5 12 8 12 45 65 5. 4
6 121/2 10 16 71 97 7.9
7 6 11 16 56 83 13.8

Average per Ib.: 7.4 minutes

a. Leaving out of account the dried legumes which have to be left to stew from four to six hours.

Source. Richards, 1939, p. 104

Using a small sample of seven women she calculated in an experimental situation the average time it took to thresh, winnow, and then grind a pound of flour (about 7.4 minutes a pound), and the total amount of time it would take to prepare six pounds of flour (about 45 minutes), the usual household expenditure. To this would be added the time it took to make the relishes (vegetables, spices, occasionally fish or meat) eaten with the flour (three-fourths to twice the time of the flour) for a total food preparation time of 21/2 hours. To this she added half an hour for fetching wood, to arrive at an average of three hours for the preparation of the evening meal, the only meal of the day.

Richards also notes the problems of housewives who were too tired to cook. Women in general prepared food after a full day in their gardens. Women too tired to go to the bush to find the ingredients for the relish to eat with the porridge simply did not cook, since without relish they said they could not eat the staple. Within the constraints of this food system, she also noted how children sometimes changed household compounds, and assiduously ingratiated themselves to the new eating group by contributing labour. Thus, the potential flexibility of the social organization to provide work and food was an important factor for individual food intake. Beyond this principle, there were also social obligations to share along kin lines, although she saw these being circumvented in times of dearth.Richards also noted that the pace of work was erratic: she was unable to specify the minimum time necessary to perform particular tasks, or, for that matter, the actual time natives spent, since they simultaneously spent time socializing. Interviews confirmed that natives did not conceptualize labour/time use in the European sense; most could not plan their work to obtain maximum results, could not calculate the labour required for any particular task, nor respond to her questions of how long tasks took. Nor did people eat according to any "regular" plan. Meals were scheduled in relation to agricultural activities, usually one meal at the end of the day. Seasonal routine and individual preferences, responsibilities, and resources altered the time of the main meal and actual daily intakes in the form of snacks. Women's daily gardening work ranged from 0 to 6 hours; an average figure might be 2 3/4 hours. People might work very hard for five hours in the garden, but then "collapse." In the slack season, old men in one village worked 14 days out of 20 and young men 7, while in the next village, in the busier season, men of all ages worked an average of 8 or 9 days out of 20. Average working days for men were calculated at 2- hours, and, for women, 2 hours gardening plus 4 hours of domestic work.

The foregoing three examples provide evidence of the value of general ethnographic observation for answering both questions about the time and energy expenditure demanded by activities in different seasons, as well as questions about social organization which may contribute to the adequate nutritional provisioning of households. Such reports have been criticized, however, as too impressionistic because they were not based on random sampling but were usually focused on a few subjects chosen for intensive study. Nor did they adequately sample the full range of activities over the course of a day or a year. To meet these criticisms, the method of random or spot observation has been developed.

Random or spot observations

The method of "spot observations" has been used to characterize the activities of groups of specific age, sex, and occupation within particular societies. These studies have been designed primarily to document labour time use for different societies. A "spot" record captures by observation the activities of all individuals present at the moment of entry into the household, and, by subsequent interview with those present, the activities of those not in view. The data are then coded in standard activity categories, such as work inside and outside the household and leisure time. Then the data are processed to determine the percentage of time spent on any particular class of activity by any individual or class of individuals. One simply calculates the number of instances of that activity as a percentage of all activities recorded (i.e. as a percentage of the number of visits) (Rogoff, 1978).

An early example of the use of spot observations was a study of the activities of males and females of different ages in a Mayo village in Sonora, Mexico, by Charles Erasmus and his wife in 1948. They used the activity data to compare working days of Mayos to those of semi-skilled labourers near the Washington, D.C. area (Erasmus, 1955). Although they did not use a random number table to schedule visits, they found that by checking the record charts they were keeping on every man, woman, and child by hours of the day, they were able to visit those individuals at hours of the day that were being neglected and fill out profiles for the 32 households and 200 inhabitants of this community. They were able to record a total of 5,000 observations in three months.

Erasmus then tallied activity observations, in culturally appropriate categories, into three general categories: "household activities," "economic activities," and "leisure." He presented each of the specific activities as a percentage of total observations for each sex, to show sexual division of labour, and then graphed the relative distributions of the three general activity categories (5 a.m. to 6 p.m.) for each sex (table 4). From the graphs, one can see at a glance the "work day," "siesta times," and the predominance of household over remunerative labour activities for women, in contrast to men (figs. la, lb). On the basis of his observations and analysis of spots, he was able to correct his initial erroneous "impressions" that Mayos in this community were "lazy" (i.e. spent a lot of time sitting around) and did not work very much. In fact, his data showed both men and women worked about nine hours and spent four hours at leisure in the course of the thirteen-hour day covered by the observers. Comparing his results to a timeallocation study of semi-skilled residents from the Washington, D.C. area, he found patterns of time use, in terms of household, economic, and leisure activities, to be roughly the same.

Table 4. Sex division of daily activities in a Mayo village

 

Percentage of observations

 

Men

Women

Economic activities  
Maguey fibre industries

13.9

0.5

Agriculture

6.9

0.2

Making or repairing tools

6.7

 
Buying and selling

5.6

1.4

Tending cattle

4.0

a

Tending other animals

3.8

1.6

Fishing

1.8

a

Working for others

1.6

1.8

Lime

1.6

 
Making fish nets for sale

0.8

a

Hunting

0.5

 
Collecting firewood for sale

0.5

 
Gathering

0.4

1.1

Carpentry

0.3

 
Preparing and spinning wool  

4.6

Weaving blankets  

1.5

Making bread, pastries, tortillas, or cheese for sale

0.3

1.8

Treating sick patients  

0.3

Subtotal

48.7

15.0

Household activities  
Preparing and serving food

0.9

24.3

Mending and sewing clothes  

5.4

Washing and ironing clothes  

4.0

Getting water for the house

1.3

3.9

Caring for children

1.0

3.7

Cleaning and arranging house or house furnishings

a

4.0

Repairing house or house furnishings

3.6

 
Milking

0.6

3.1

Errands

0.8

1.8

Collecting or chopping firewood for the house

1.6

0.3

Making herbal remedies  

a

Butchering

1.6

 
Eating

9.1

3.7

Subtotal

20.5 +

54.2 +

Leisure activities  
Lying down

10.9

4.1

Sitting

8.1

10.6

Chatting and visiting

8.0

9.1

Personal

2.3

6.1

Recreation

1.4

0.8

Subtotal

30.7

30.7

Total

100.0

100.0

a. Less than 0.1 per cent (only one observation).
Source: Erasmus, 1955.



Fig. 1a. Distribution of daily activities in a Mayo village: men (after Erasmus, 1955).



Fig. 1b. Distribution of daily activities in a Mayo village: women (after Erasmus, 1955).

Johnson (1975) also chose the spot observation method to document the work activities of a Machiguengo community in Peru as a more precise economical alternative to impressionistic recording. He and his wife collected spot observations over a ten-month period, while also gathering data for a broader study of the cultural ecology, social organization, and sex roles of this community. Using location and distance from a central point as a principle of selection, they visited all of the households within a 45-minute walking distance of their homestead, divided them into two groups (upriver/ downriver, to cut down on walking times), and set up a schedule of "random" visits for daylight hours, during which they recorded what everyone in the household was doing at the moment immediately before they became aware of the observers' presence.

They made spot observations in 13 households with a total of 105 members over 1,345 days, which resulted in 3,495 cases. Their hours of observation ranged from 6 a.m. to 7 p.m., during the daylight hours when it was convenient to travel, which were also the hours when the households they observed indicated they did not mind such unexpected visits. The data, coded according to person, time location, and activity, were punched directly onto IBM cards. (Probably today one would go directly from data-coding sheets to interactional data entries on computer tapes. ) After the observations were completed, the researchers simply calculated, for each social category of interest, the number of observations of each activity category as a proportion of the total, and converted these figures into the relative time spent on each activity. They were then able to compare activity rates by sex (table 5) and age class. As in the case of the Erasmus study, spot observation was combined with more traditional participant observation methods to acquire data on qualitative questions, such as cultural ecology, family organization, and sex roles.

As a third example, Minge-Klevana (1971) used spot observations to test the following hypothesis: as children allocate more of their time to education and less to the farm, parents make up for the forgone labour and offset the additional educational costs by allocating more of their own time to family-based labour. Working in a Swiss Alpine village, she studied time allocation of all family labour, the percentage performed by mother/wife, husband/father, and children, comparing those households where fathers did wage labour with those in which fathers stayed at home. From her original sample she selected 32 families for analysis because they had had more than 40 observations each. Although she used Johnson's method of "spotchecking" and accumulated 3,433 person-observations, it is not clear how she decided when to go where, nor is it clear from her descriptions how she calculated children's contributions to the household operations. Some of her difficulty in calculating children's work stems from the nature of defining "work" versus "leisure" activities for children who are ordinarily in school. For example, if children tend cows when the parents cannot both tend cows and harvest potatoes, should the anthropologist consider this "a holiday," as children themselves do?

While spot observations are more amenable to statistical analysis and can be considered less impressionistic than other forms of observation, the activity data so collected and analysed do not allow one to study how differences in household organization or individual capacity affect work effficiency. Nor is there an opportunity to observe and report on social interactions, qualitative differences, or sequencing in time use and performance. To overcome some of these deficiencies, some researchers combine spot-checking with 5-, 10-, or 15-minute narratives. This method provides them with a profile of standardized behaviours for selected social groups as well as a more qualitative record of interactions between people and their environments.

Table 5. Division of labour by sex (married adults)

Activity (%) Married men
(n= 15)
Married women
(n=20)
Eating 9.1 7.0
Food preparationa 1.5 18.1
Child-rearinga 0.1 8.8
Manufacture 10.4 15.9
Woodworka 6.7 0.6
Cotton clotha 0.1 13.5
Other 3.6 1.8
Wild foodsa 15.6 6.6
Collecting 2.9 2.5
Fishing 5.7 2.3
Huntinga 5.7 0.
Other 1.3 1.8
Garden laboura 18.5 6.6
Clearing, burning, plantinga 3.7 0
Weedinga 5.8 0.3
Harvest 6.1 5.1
Other 2.9 1.2
Idle 18.1 19.1
Hygiene 2.5 4.5
Visiting 8.0 5.8
Other 16.2 6.6

a. Differences are significant at p < .01 level (t-test).

Source: Johnson, 1975.

Whiting and Whiting's (1975) research team has used the method of spot observation with narratives extensively to observe children in six cultures and to chart their development in terms of a set of standardized behavioural categories. The final cross-cultural comparison of childrearing practices was based on observations of 67 boys and 67 girls, aged 3-11. Each child was observed at least 14 times. Five-minute narratives described in detail the behaviours during the observation period, including activities coded in standardized categories, and also details of interaction, conversation, and physical movement. With one exception, the total number of behaviours recorded in each culture was at least 1,000. The end result was a comparison of behaviours of males and females of different age sets in six cultures. General terms, such as "nurturant" or "aggressive," were used to describe the behaviours, which were characterized as the outcomes of child-rearing practices and sex-role development in the different cultures.

While this method of observation and reporting potentially provides more information than simple spot-checking, the value of the data depends heavily on the scope of the activities record. For these studies, the behavioural categories, "acts sociably," "insults," "offers help," "reprimands," "seeks dominance," "seeks help," "suggests responsibly," "offers support," "seeks attention," "assaults sociably," "touches," "assaults," in which investigators were trained in advance to record, may have been useful for cross-cultural comparison of child development, but hold little meaning for analysis of other behaviours.

By contrast, studies of mothers' activities and child care carried out by the same methods in ongoing projects on cross-cultural child development do utilize categories of interest for other social scientists, including nutritional anthropologists.

Bloch (1979) collected five-minute observations on samples of Senegalese mothers involved and not involved in seasonal work in the cash economy. She stratified her sample of 50 mothers in terms of the age and sex of a mother's youngest child, using age categories of 0-6 months, 712 months, 13-18 months, 19-24 months, 2 years, 3 years, and 4-6 years. She compared differences in proportions of time allotted to personal activity, primary child care, domestic work (not primary child care), and economic activity. While she did not analyse separately the food-related activities of the mother in relation to the youngest child, number of children, or the economic organization and livelihood of the mother, her activity records of spot observations with five-minute narratives potentially provide such data.

Spot observations on the interactions between children of different ages and their environment can also potentially contribute to questions of nutritional concern, particularly when the records include information on factors such as proximity of the child to mother or mother surrogate a measure of child care. This measure of mother-child interaction can then be further tested to see how it relates to nutritional status.

In assessing the value of spot observations over more impressionistic methods of observation recording, one should note that the "spots" produce immediate quantitative "snapshot" data, remove the problem of observer bias or interference with the individuals observed (but not bias by observer), and provide reliable figures on activity profiles by different classes of individuals throughout the year. The "snapshot" will not give one a clear idea of: the duration of any particular activity; the sequencing or co-ordination of units of activity; or the time/quality of social interactions.

Nevertheless, some of these deficiencies can be made up by either combining spots with narratives or by making other kinds of longer observations. Spot observations can probably provide a reliable indication of factors related to the tasks of the elderly and of the young? e.g. the ages at which children of either sex begin to work, the patterns by which they become responsible for tasks and master them, and the hours and kinds of contributions they make to the household economy. If sufficient data are collected in different seasons and among different kinds of households, the method can also supply an accurate account of differences in activity between households with varying social structures and occupations and significant seasonal differences in work and leisure-time use. The use of spot observations to sort out differences between household work patterns, rather than "average" activity profiles, might provide the raw data from which to work towards a new measure of "household function," which can then be correlated with other factors, such as nutritional status.

None of these analyses deals specifically with the subject of time allotted to food versus other activities. Indeed, Erasmus (1955) highlights one of the problems with classifying the data in his example of eating: Is eating a "work" or "leisure" activity? For women, who fit eating in between and around food preparation and feeding the rest of the family, eating is a "work" activity, but for men, who spend more time eating, it could certainly be counted as a leisure activity.

Spot observation, nevertheless, seems to offer an ideal way to characterize the activity patterns of a culture, in particular food acquisition, food preparation, and the scheduling of meals and snacks. Significant differences in activity profiles could then be used to select households for longer observations to refine and extend the initial spot observations. Any intake data, of course, would have to be collected separately from the initial stage of spot observation study. Spot observation is also a potentially useful way to characterize labour allocations in different households. Such information can shed light on eating patterns and schedules and the relationship between women's work schedules and (a) meal preparation and (b) participation in development projects.

Spot checking could also be used to show the proportion of time allotted to different feeding behaviours and resultant food intakes. Results of spot checks on mothers' activities and patterns of child care for different ages and sexes could also be used to develop measures of nurturant behaviour. Also, spot-checking could become an efficient method of recording how family labour-time-allocation patterns relate to feeding patterns and nutritional well-being, particularly where there are significant differences in labour use among families. Where education and labour demands impinge on children's time, spot-checking can reveal conflicts in household and school schedules that interfere with children's food intake. For example, one could establish the meal-time profiles of households, and check these against children's work and school schedules.

Although random spot observations have many advantages, one should also consider some disadvantages. Spots are time-consuming to collect and may require large amounts of time for travelling from one household to the next, as many observations are needed to make the results statistically reliable. Each of the studies cited above covered a small sample of households and required substantial time commitments. Another disadvantage is that activity data, averaged across households and seasons the form in which most "spot" studies have been presented - do not allow one to analyse how differences in household organization or individual capacity affect work efficiency. Even when combined with narratives, there is insufficient opportunity to observe social interactions, work rates, co-ordination, and other qualitative differences in time use and performance. Such data must be collected separately by observation and interviews.

Since very small numbers of subjects and a larger number of spot observations are used, this method also entails special problems of sample selection. What constitutes a representative sample must be determined carefully and depends on the variance of the sample. Also, variations in the numbers of children in the household and specifics of household organization make attempts at statistical comparisons of behaviours and activity schedules of age/sex categories within and between cultures difficult, and the interpretations of such findings must be carefully qualified. Moreover, since the many spot observations or spots with narratives are potentially disruptive to the normal flow of activities, given observer presence, one can question whether the data they produce have greater scope and utility than data provided by day-long observations or some other form of short day observations. The latter two methods can provide information on sequencing and duration of acts, and, in food-related studies, can also supply data simultaneously on eating behaviours and intake by the focal personae of the study.

Day-long observations

Economists, anthropologists, and others have used day-long observations to study people's activities, in order to obtain accurate records of sequencing and duration of activities, particularly in populations whose sense of time does not lend itself to accurate recall of this kind. With this method, the observer may stay with the household during all waking hours. However, this much observation is exhausting for both the observer and the observed, and usually a compromise of 12 to 13 hours, covering most of the working day, is made. On the following day, household members relate how they spent the additional hours, prior and subsequent to the observer's presence, to complete the record. During the observation period the observer records all activities of the member(s) on whom the study is focused: e.g. the focal woman, if understanding her activities and their impact on food provisioning is the purpose of the observation, or children of particular ages, if the purpose of the study is to see how they acquire food through regular work, foraging, and stealth. Depending on the spatial and social layout of the residential units, proximity to the market-place, and so on, it is sometimes feasible to leave the focal household member performing a repetitive task and observe other members of the household who may be in the vicinity. Otherwise, one must rely on respondent reports of their activities in the interest of getting complete data on the focus of the study.

To record observations one of several procedures can be used. The most detailed and accurate device is a time record sheet, indicating starting and stopping times of all activities, interactions, and events. These records can be made using a timepiece, usually a watch. At the end of the day, for each focal individual, time spent on a standard list of activities can be summed for comparisons between households of different types, categorized by occupations, and ages of children. Alternatively, one can set up blocks of observation time during the day, record what activities take place during particular hours, and then compare activities across households and categories of individuals. For example, "morning" or "late-afternoon" activities can be compared for adult men and adult women.

Examples of studies using day-long observations are Peet's work on the economic value of Nepalese children (Nag, White, and Peet, 1978) and Evenson's research on children's household contributions in the Philippines (Evenson, Popkin, and Quizon, 1978). Peet collected work-input data for all children and adults in 45 to 50 households, once a month over a period of seven to ten months, depending on the household, as part of an economic survey of 674 households in a Nepalese village. During a month-and-a-half-long period when he could employ schoolchildren as field assistants, the data collections were extended to 106 households. Data were collected mainly by observation, in some cases supplemented by interview (Nag, White, and Peet, 1978). Results were reported in terms of 11 standard activity categories for each of eight age groups for each sex (table 6a). The time children spent on productive work and schooling were then compared, by age and sex, with results of a Javanese survey by Nag, White, and Peet based on data collected by the 24-hour-recall method (1978) (table 6b).

Table 6a. Average time input (in hours) per person per day in different work activities among females of various age groups in a Nepalese village

Age group and sample size

  6-8 9-11 12-14 15-19 20-29 30-39 40-49 50+
Activity (29) (30) (25) (33) (28) (50) (34) (26)
Child care 1.7 0.9 0.6 0.2 0.7 2.9 1.3 0.4
Household food preparation 0.2 0.9 1.5 2.0 1.8 2.8 3.1 3.2
Firewood collection 0.1 0.1 0.2 0.2 0.2 0.1 0.1 0.0
Other household maintenance work 0.4 0.7 1.0 0.8 1.0 1.1 1.1 0.8
Animal care 2.2 4.7 3.5 3.4 2.5 1.8 1.8 1.6
Wage labour (agricultural) 0.0 0.1 1.2 2.1 1.8 2.0 1.9 0.7
Wage labour (non-agricultural) 0.0 0.0 0.0 0.0 0.1 0.1 0.0 0.0
Handicrafts 0.0 0.0 0.0 0.2 0.1 0.1 0.2 0.1
Reciprocal labour exchange community labour 0.0 0.0 0.0 0.1 0.2 0.0 0.0 0.0
Agricultural work (own land) 0.3 1.0 1.5 2.3 3.3 3.0 3.0 3.0
Production of articles for sale a 0.0 0.0 0.4 0.0 0.4 0.2 0.2 0.9
Total household maintenance (1-4) 2.4 2.6 3.3 3.2 3.7 6.9 5.6 4.4
Total directly productive (5-11) 2.5 5.8 6.6 8.1 8.4 7.2 7.1 6.3
Total all work (1-11) 4.9 8.4 9.9 11.3 12.1 14.1 12.7 10.7

a. Includes collection of forest products, blacksmithing, and tailoring; excludes portions of handicrafts, agricultural products, and animal products which are sold.

Table 6b. Work inputs of children in various age-sex groups expressed as percentages of the average work inputs of males aged 15 years and over

 

Age group

 

6-8

9-11

12-14

15-19

Javanese village  
All work  
Boys

43

38

56

91

Girls

39

62

99

120

Directly productive work  
Boys

23

24

42

71

Girls

15

39

59

86

Nepalese village  
All work  
Boys

37

65

73

93

Girls

47

82

96

110

Directly productive work  
Boys

37

61

69

93

Girls

30

71

81

98

Using a more systematic design, Evenson's survey team in the Philippines also studied the economic "values" of children, first with 24-hour-respondent recall, then with observations. During the latter phase of their study, which involved approximately 600 observation days, they covered 99 households at three times during the year to obtain an accurate record of the activities of all men, women, and children. To ensure that the presence of observers would not bias activities, they spent two days in each household for each day of data collection: one day to accustom members to the observer's presence and the second to record. Evenson had a large research team and, therefore, was able to cover a large sample. With a smaller team this type of method would have to be restricted to small samples.

A major limitation of the method is that it is exhausting for observer/recorders. In practice, one should schedule one day to rest and tally time observations for every two days observing per recorder. Despite this drawback, the results of such labour-time investments are rewarding. Quizon-King (1978), reporting on the data obtained by both recall and observation, shows that much more accurate records of labour-time allocation by household members can be collected by observation. Also, Quizon-King was able to stratify her sample according to father's occupation and numbers of children and compare labour-time allocation by different family members across households to discover significant differences in time budgets.

Nutritional Anthropology Studies Using Day-long Observations

While time-consuming and, therefore, necessarily limited in household coverage, the data produced in day-long observations are rich, diverse, and useful for many purposes. The method is the best for getting data on what children eat and how often and for what periods of time, while simultaneously collecting other food and activity data from households. For example, Messer (1981) was interested in documenting how different occupational characteristics of households affected women's and children's work and the focal woman's management of food for household members in a rapidly modernizing Mexican community. Households were, therefore, selected on the basis of the occupations of adults and the numbers and ages of children. While the original research plan had been to select households by occupation from a random survey sample, in fact, households were selected on the basis of occupation, child characteristics, and willingness to co-operate. Messer and a trained North American research assistant observed 20 households, four days each, two days in the wet (non-school) season, two days in the dry (school) season. Their observation periods were 12-13 hours long, from 7 a.m. to 7 p.m., and they concentrated on the focal woman in the household, but also recorded the activities of men and children around her (Appendix 1).

While the woman was performing repetitive tasks such as laundry, other members of the family were watched in more complete detail than was possible while focusing on the mother. Household arrangement of activities in a central room facing out onto a central courtyard made it possible to follow in most cases women's work, men's work, children's games, and, to a great extent, children's foraging, either in adjacent lots or in households of adjacent members of the "extended" family. Ordinarily the observer remained in the background.

Messer found that households that had carefully controlled their activities, such as child punishment, during participant observation, were capable of ignoring the observer, as she recorded unobtrusively throughout the day. Adults followed their natural routine, which occasionally included fits of anger and brisk thrashing of children. Though children initially explored the potential diversionary value of the observer, they were also able to ignore her presence, as she stoically ignored them. They continued chores or play; she continued her observation and recording. The only time the observer interrupted the normal routine of the household was at mealtimes when she weighed all food portions before consumption, and tallied all wastage per person afterwards. In addition, all food and other expenditures were tabulated for each observation day, either as they were purchased, or by brief questioning afterwards.

Day-long observations not only allow for the possibility of collecting complete information on food intake, they also may reveal bottlenecks in women's labour time, that is, when competing tasks prevent her from doing some of them well, if at all. Using this method, aspects of sanitation, illness, and healing that affect eating and the nutritional status of household members can be recorded. These variables would be missed by other methods. Day-long observations can also avoid some of the criticisms of less extensive participant observation cited by Johnson (1975), among others, that most anthropological observations are "impressionistic" - i.e. they do not provide a representative sample of the population and they may not get enough observations of activities at different times of the day or seasons of the year to fully represent activity patterns. By carefully selecting households on the basis of some systematic principles and setting up visits to cover the cycle of natural and cultural seasons, day-long observations can eliminate most impressionistic bias. The major source of bias remaining is that, since one observer cannot record everything, she obviously makes a number of choices about what to record. Day-long observations also avoid error created by informant impressions of the amount of time spent working by adults or playing by children, biases that may be introduced by the less time-consuming method of collecting 24-hour-recall data from respondents.

Recall Methods

Whereas observation methods must be used in cultures where people are not accustomed to calculating daily activities on an hourly basis, 24-hour recall may be a more efficient method to collect data where members are used to thinking within an hourly time-frame. With the serial recall method, a trained interviewer visits each household selected at random on a regular basis (every sixth day, every fifth day) so as to avoid always arriving on the same day of a cultural time-cycle. She asks a key household member or a group of household members to relate, sequentially. the time spent on all activities by each household individual for the previous twentyfour-hour period. Some interviewers prefer group interviews, since household members check and correct one another's activity claims and, therefore, provide more complete and reliable information. Ideally, the respondent can recall with some accuracy the starting and stopping times of the activities by citing coincidental events, such as the beginning and end of the school day, radio programmes, fixed mealtimes, and calls to prayer. The interviewer can then calculate both the hour and duration of activities of interest.

Alternatively, the interviewer can ask for "activity-specific recall," in which case the respondent is asked to produce a time-activity list indicating how much time is spent on each of a number of activities, classified by the respondent or suggested by the interviewer in advance. This method forces the respondent to calculate time budgets for herself or himself, and then for other members of the family. Comparing individual time budgets allows for some crosschecking for accuracy, based on the interviewer's composite picture of household functioning. In serial recall, as in day-long observa tion, recording may be by the diary method, in which the interviewer records times of starting and stopping of all activities, supplied by the interviewee, or by marking a time-sheet, already divided into blocks of time to indicate hours of the day. Cain (1977), in his study of fertility patterns in Bangladesh, used the latter method. The sample consisted of 120 households selected at random by cluster sampling, visited every 15 days "after work hours." Data sheets were made up in advance, by household, supplying an individual time column for all household members. He estimated one hour to complete each household budget form. The interviewer simply drew horizontal lines to mark starting and stopping of each activity, all household information having been pre-recorded (fig. 2). In this manner, trained research assistants could cover two households in the evening, after 5 p.m. Households were asked to list all activities, by type and duration, for the 24-hour period prior to the interview (5 p.m. to 5 p.m.). The sheets were checked for inconsistencies by comparing household members' columns, then brought to a centre in Dacca for coding. Activity times were calculated, summed, and coded for each household member and recorded on household worksheets. The data were then transferred to individual worksheets on which additional rounds of activity data were entered. The processing of each individual time-budget form took 3.5 person-hours to complete: one hour to collect the data, half an hour to field-edit the budget forms, one hour to code times and activities, and one hour to transcribe data to worksheets and to subtotal time segments.

Time budget  Date Checked ____________
Sample point ID Length of interview Coded (1) ___________
Strata  H/H type  Interviewer Coded (2) ___________
Today   Parents   Children  
5 a. m.          
7 a. m.          
9 a.m.          
10 a. m.          
11 a.m.          
2 p.m.          
< < 5 p m          

Fig. 2. Diary method for serial recall (after Cain, 1977).

Ong (1982) also used a random sampling design and serial recall in her time-allocation survey of 40 households in a Malaysian village. She was interested in discovering varying patterns of work and leisure allocation by sex and groups in an economy dependent on income from both farms and wage work, with young women forming an important group of wageearners as factory-workers. She found that a trained research assistant could cover approximately four households per day. Households were visited once every seven to eight days to collect time-allocation data.

Respondents in Ong's study (1982) were very time-conscious, and she had no trouble collecting 24-hour-recall data in this Moslem society. She also tried to collect income and expenditure data, although she planned an economic-nutritional component to follow the timeallocation data collection, given the limits of respondent patience for interviews. Data were tallied and coded after each recall, and computer analysis was used to find total averages for each individual for each activity over a six-month period. Time-allocation patterns were later interpreted in relation to class and fertility patterns of households.

As a third example, Chapin (1974) used serial recall in a survey of recreation activities of households in the Washington, D.C. area. This method was chosen because the researchers wanted to cover a large number of households with limited interview-time input. Their purpose was to record a detailed plan of people's preferred ways for allocating time to leisure activities. Households were selected by a random design principle within the study region. They found the poorer segments of the population insufficiently skilled to keep diaries, so they used a detailed set of questions, and the respondents, primarily women, were able to produce elaborate timesheets within an hour's questioning. Interviewers were able to get a precise plan of the typical activities of each adult and child for weekdays and weekends and, with qualitative questioning, were also able to correct for irregularities in any particular day's events.

Recall, if it can be certified by observation checks not to overestimate or underestimate time allocations, is an excellent way to record what happens in the course of a day in households of different socio-economic status within a community. One can stratify household samples for the purpose of seeing how households and individuals differ, by class, in terms of how they spend their time, and the relationship of these different activity patterns to other aspects of behaviour, e.g. fertility and nutrition. Recall also makes efficient use of the interviewer's time. A trained interviewer can cover approximately four households per day (Ong, 1982) or two households per night (Cain, 1977), with minimal intrusion. A respondent may also verbally describe activities she did not have time to write down in a diary entry without interrupting her own activities.

Disadvantages are that adults may tend to underestimate, for example, the productive work of children (Quizon-King, 1978). Respondents may also fail to relate time spent doing several things at once; for example, mothers, without some prompting by an interviewer familiar with the culture, may recall very inaccurately time devoted to child care that was supplementary to some other primary activity, such as cooking. Having more than one person present at the interview, particularly older children or other adults also responsible for child care, may also help to overcome some of the limitations in recall that result from honest error or the desire to present one's time allocations in the best possible light. The latter may lead to overreporting of productive activities and underreporting of non-productive ones.

An additional disadvantage of the recall method is that it is often not feasible to collect other kinds of information concurrently. Eliciting a full record of all family members' activities and complete production/consumption data, plus full dietary recall statements, is painstaking, time-consuming, and beyond the patience of most respondents. Such information on food expenditures in money and time and food consumption data, however, may be collected separately and added to the activity data. For example, food interviews on the structure and content of meals and snacks, and the social division of labour in food acquisition and preparation, in combination with activity data, based on recall, could help one analyse significant differences in eating patterns. However, the information would have to be collected in two separate stages, and neither stage would be likely to give detailed information on snacking behaviour, which is most accurately collected by observation.

The Diary Method

As a final alternative for collecting 24-hour data, literate, time-conscious respondents can be asked to keep diaries, recording starting and stopping times for all activities. The method has been used by economists and urban planners to understand (a) labour and leisure time-use by European and American housewives, (b) urban activity patterns, including time spent commuting to work and time related to recreation activities, and (c) marketing (Szalai, 1972). Data from self-reporting have also been used by anthropologists to understand differing patterns of household-labour allocation in various food-producing societies. The method works best where respondents clearly understand the task involved and do not find too burdensome either the momentary interruption to record or end-of-the-day recall recording. If these tasks are too disruptive they will interfere with the respondents' daily performance and also jeopardize the study.

Diary-keeping should be restricted to studies where detailed data on one respondent are required, as where a mother's activities are followed, and other household members 'work, educational, and recreational schedules are incidental. One could not, for example, gather detailed data on children's activities by this method, although such information could be gleaned, on occasion, by the recall method. For example, a mother might remember what children were playing, where they ran for snacks, and what they ate, but not have time to write all this down during the day. Alternatively, diary-keeping can be combined with a daily recall strategy in which a respondent acts as the interviewer, asking herself to remember the sequence of activities at the end of a day. She would then present this information to the researcher in lieu of a separate interview.

To use the diary method. the researcher should first gain experience in the culture through observation and interviews aimed at defining the kind and duration of the activities of research interest. This preliminary work forms the basis for constructing an appropriate list of activities for the analysis and for sensitizing diary-keepers to relevant categories for their records. This preliminary list can be submitted to respondents for their comments and suggestions on how to improve and systematize the records and analysis.

Most sample notebooks use time sheets, ruled or sectioned off with the hours of the day, so that one can simply draw a line at the starting and stopping times of activities, and fill in the appropriate activity label. Alternatively, one can record times of stopping and starting activities, which can then easily be summed.

The main advantage of diary recording is that it potentially allows recording of data on a greater number of days for each household, since the initial hour or more of recording the primary data by the observer is avoided. However, this data must then be processed, so time constraints for the researchers should not be underestimated.

Experimental Designs

In addition to the time-data collection methods just described, psychologists have proposed several '`time-frame" methods for observing and measuring infants' and children's behaviours. These methods can be used to show how nutrient intake affects the cognitive, emotional, and physical development of infants and children. Such methods can also reveal if there is a discernible relationship between nutrient intake and behaviour, such as a child's responsiveness to persons and objects in the environment.

Instructive examples of the use of time-frame methodology are a series of studies by Chavez and Martinez (1979) on the effects of mother/child nutritional supplementation on child behavioural development. The investigators use both qualitative and quantitative reporting to classify actions within the mother-child-environment system. Observations were recorded in terms of 37 descriptive states of the child in relation to its environment, of the mother's actions toward the child, of the child's actions toward the mother, and of the father's, siblings', and other adults' relationships toward the focal child.

Classifications and records for each focal child were made after observation for 72 continuous hours in the household. These observations also enabled the observer to record all feedings. The first day was spent observing lactation and accustoming the child and other household members to the observer's presence. On the following days, the observer sat herself in a corner for periods of 11/2 hours in the morning and 11/2 hours in the afternoon, and pretended to read, while she observed the mother-child interactions. Every 30 seconds she would lift her head and note if the mother was talking to, smiling at, feeding, or otherwise interacting with the child. She then recorded behaviour frequencies as a proportion of total behaviours observed. Following the child's relationship with other members of the family was more difficult, since their presence in the household was less regular than that of the mother. To record these interactions, the observer recorded activities (interactions) at 10-minute intervals each hour, over a day of 12 hours. Thus, she captured activities at 8.30-8.40, 9.30-9.40, and so on.

To study play activities experimentally, the investigators constructed three-by-three-metre squares, with 30 cm fencing, and placed the child in the centre with the mother at one end and a pile of toys at the other. They then timed numbers of interactions of children with mother and toys to measure activity levels, maternal dependency, and also, to a point, aggression.

Results were analysed using Chi-square statistics, comparing two groups: well nourished and poorly nourished. Infants were followed at ages of 2, 8, 16, 24, 36, 52, 72, and 96 weeks. Graphing per cent times (y-axis) against weeks of age (x-axis) they compared time in the cradle; time sleeping; time playing (by proximity of the mother); time carried (by both arms or on the shoulder); time the mother spent smiling, talking, or playing with the child; time crying; time babbling or talking; time child spent smiling at the mother; rapidness of response (by infant and mother) to infant's dirtying; and accident prevention. The categories were primarily of behavioural interest, although data on sanitary conditions are also of interest for interpretations of health. Interactions of father and other family members were also graphed, and showed significantly greater interaction between a well-nourished child and its environment and significantly greater stimulus levels provided to the well-nourished compared to the malnourished child. Generalactivity-level behaviours were also calculated; in general, the well-nourished child was more active and spent more time exploring his or her environment. Finally, in the experimental set-up, observers graphed the positions of the child on a quadrant, showing initial position of infants and routes to mother and/or toys, and calculated "intensity" of a standard set of behaviours: cries, is upset, looks at toys, plays with toys, and talks. Comparing individual cases, these results again show the well-nourished child to be more active. One might question, however, whether participation in the supplementation programme, apart from the nutritional effect, did not somehow affect the activity levels of the better-nourished group, both children and environmental adults. (All data are taken from Chavez and Martinez, 1979).

This series of studies used ingenious methods of timed observation rather than time allocation to provide answers to questions about nutrient intake, child behavioural development, and household functions. The method is exhausting and time-consuming for the observer, and only a trained observer who has the confidence of the households observed would be able to carry out such a study. Interpretations must necessarily be based on observations of very small comparative groups (groups of 20 in the Chavez and Martinez study). Nevertheless, this combined sequencing of continuous hourly sweeps of activity-observation over the course of a day and intensive observation of mother and child pairs during morning and afternoon periods provides a method to measure activity and stimulus levels, which can then be correlated with levels of nutrient intake. If the investigators have the kinds of facilities that will allow them to follow experimental and control groups or a stratified population sample over a period of two years. and a well-trained observer staff, then such studies, which provide critical information on child behavioural development, are feasible.

For investigators without such resources, calculation of child activity levels, quality of activity, interaction levels with the household environment and qualities of interactions can best be extracted from day-long observation data, recorded either continuously or at short intervals. For studies focusing on children's behavioural developments, intensive observations with spots or spots with narratives have been employed by psychologists (e.g. Bloch, 1979).

Summary

Some of the preliminary questions that must be answered in designing either a time-allocation study or a time-frame study are: By what method should data be recorded? Should single individuals be followed or should social groups be scanned? Is it wiser to focus on certain target activities or social interactions or to rely on a random sample of behaviour to provide an accurate picture of all activities, including those one may be most interested in? Many of the studies described here have tried more than one method before settling on the one that will yield the best results, given their research interests and constraints (Quizon-King, 1978).

Sampling also presents difficulties. Among them is the claim by some researchers that the results of following single individuals, as opposed to periodic observations of their activities in the context of censusing the activity of a whole group, yield equivalent results; others dispute this. Altmann (1974) provides a thorough discussion of these issues in the context of primate behavioural studies.

In almost all of the time-allocation studies described here, with the exception of the studies by Erasmus (1955) and Johnson (1975), research assistants were relied upon, and it is appropriate to add a note on that subject here. To cover a large number of households by either observation or interview, the principal investigator will need research assistants. Appropriate personnel from the household's point of view are those whom households do not consider to be interfering and to whom they will allow ready access. Research assistants must also be able to provide reliable records.

Social scientists working in Islamic societies, particularly in South-East Asia, have found that the best research assistants are literate, with the equivalent of at least a secondary-school level of education, and come from the observed household's own section of the town or village. Closely related persons are in the best position to gain access to the detailed, intimate household time and monetary budget information, and are also in the best position to catch errors in reporting.

In contrast, studies in Mexico have used well-educated personnel from outside the community being observed. Excellent technical assistants have been of Mexican origin, with a secondary-school education plus additional training, or have been North Americans. Households were reluctant to allow local persons access to the intimate details of family life, eating patterns, and expenditures.

There is, then, no universal rule as to who will be accepted to record the highly confidential data on household functioning. In each case, the investigator will have to evaluate assistants on the basis of her or his own professional perception and knowledge of the culture. The acceptability of local versus non-local assistants and the acceptability of male or female personnel for certain kinds of research will have to be assessed.

Processing and interpreting data

No matter which recording method or combination of methods is chosen, all entail common problems of processing and interpreting data. Investigators must be sure to collect sufficient information on the various kinds of households and individuals of nutritional interest. They must also be sure to collect data in time and activity categories that will ensure accurate presentation of significant variations within communities or between different societies. Four topics important for analysis and interpretation of time-use data are (a) social units, (b) activity categories, (c) cultural units of time, and (d) cultural time sense.

Social Units

General procedures for selecting communities and sampling within them have been addressed in the chapter by P. Pelto in this volume. For studies of time allocation, the community usually has some social or economic characteristics of particular interest to the investigator and is not selected at random within the region or nation. Within the population, households may be selected by some random principle, such as random numbers table; all households within a certain distance from the researcher; or every fourth household. Household selection may also be done on the basis of certain characteristics, such occupation, social status, or the presence of one or more children of particular age. In addition, given the intrusiveness and difficulties presented by time-allocation studies, households are selected in terms of their co-operativeness.

In selecting characteristics and categories for meaningful comparisons between households, it is important to consider not only cross-cultural comparability but also categories that are culturally significant in the society under study.

For example, to compare activity patterns meaningfully among women in African lineage systems, stages in the woman's life-cycle (e.g. having no children, having one or more children under six, having one or more married sons, being a widow) may be more significant than chronological age (Hemmings-Gapihan, 1981). Household samples may also be stratified along other relevant dimensions such as adult occupations that affect their presence/absence in the household, the presence of other adults or older children available to help with child care, and numbers and ages of children. For example, Quizon-King (1978) divided her sample into three family types by size, numbers of children, and father's occupations. In societies with sex/age grades, or in which specific ages are deemed appropriate to take on work, special attention should be paid to selecting these cultural categories, particularly in the observation of children's activities. For studies of women's activities in relation to child care, the age of the youngest child and the availability of surrogate mothers may form significant dimensions. For evaluating children's contributions to household work and income, numbers of siblings per household or per child may be relevant (Nag, White, and Peet., 1978).

Activity Categories

An additional concern is to establish categories of analysis that are both culturally complete and cross-culturally analysable. Categories should be chosen only after collecting several rounds of initial data in the field; even then, the categories may have to be amended. The list should include a careful breakdown of cash and home chores highly specific to the occupational life of the community, for purposes of within-culture comparisons. One can then try to combine these into a few cross-culturally valid general categories such as work outside the house, work inside the house, and leisure (Erasmus, 1955), or food acquisition, manufacture, and preparation, childrearing, eating, hygiene, and visiting/idle (Johnson, 1975). Several examples of activity categories and data tabulations are provided in tables 4-6. Special considerations, as mentioned earlier, must apply to the choice of categories to record children's activities and social interactions (e.g. resting, stimulus-seeking, stimulus-receiving) or aspects of child care (e.g. distance of child from mother or surrogate mother).

If computer programmes are to be used to aid in the analysis, precoding daily activities and entering precoded calculations as they are recorded is advisable. Whether data are collected by observation, recall, or diary records, they should be processed into culturally appropriate and cross-cultural (general) categories simultaneously with or immediately following collection, so that any problems in analysis or interpretation can be checked. Time must be written into the research schedule for this activity.

When daily activities are being tabulated, it is necessary to be consistent in counting time allocated to multiple tasks performed simultaneously. For example, in counting time allocated to child care, a useful distinction can be made between "primary'' child care (no other activities involved), and "secondary" child care, carried out simultaneously with other activities such as food preparation. Also, care of infants may be distinguished from care of older children. For studies of women's or children's work, one might want to collect quasi-experimental data to compare the rate of task performance with and without the simultaneous burden of child care. For a study of the "cost of children," one might want detailed information on how children of different age categories are cared for. Time allocated to household production can be construed to include time spent getting to and from locations of food acquisition, or, alternatively, travel time may be accounted for separately.

Finally, decisions must also be made on how to classify ambiguous activities in cross-cultural comparisons of time spent working. Erasmus's example (1955) of eating as female "work," in contrast to male leisure activity in Mayo society, and Minge-Klevana's example of children tending cows as "leisure," cited above, illustrate problems of classification. Moreover, in many societies it is difficult to disentangle "socializing" from time spent "working." (A good discussion of the problem of counting household "work" is contained in Minge-Klevana, 1980).

Units of Time

Time units used by the people under study to arrange their activities are a third factor to keep in mind when processing and interpreting time-activity data. Daily activities may be patterned according to the prevailing daylight hours and relative heat or cold. For example, in cultures without electricity, most productive work will probably take place before nightfall and after sunrise. However, in cultures with sources of artificial lighting, particularly those practicing some sort of cottage industry in the home, a significant amount of production may take place in the evening, particularly by women who work mainly after the children are asleep. Seasonal shifts in work hours may occur depending on the prevailing heat of the day. For example, the very early and very late daylight hours are devoted to work during the hot, dry season in the Middle East. To accurately describe time use in such situations, one will probably want to calculate separately the work patterns in the different seasons; averages are less useful in evaluating how people work in relation to their food supply.

Ritual and market cycles will also have to be taken into account at the initial stage of data collection and in the analysis and interpretation of results. White, for example. analysing the work input of Javanese children in 20 households, visited each household every sixth day so as to coincide neither with the five-day market cycle nor the seven-day administrative week (Nag, White, and Peet, 197X). In cultures where people pace activities according to the market cycle, one may want to consider production and selling periods in evaluating variance in activities over different days of the market "week." For example, in Messer's field-work in Mexico, people weaving and sewing goods for delivery to a middleman might work in ten-day cycles. The two days before the pick-up are extended, frenetic workdays, and the days after the goods have been picked up are periods of relative "rest."

While economists and nutritionists seem to prefer "average" figures in constructing time budgets and activity schedules, it is important to realize that time is used differently on different days. It is, therefore, important to calculate the significance of such different days and of such differences in determining average schedules or noting variance. Records of such pacing in human energy expenditure and energy intake may provide data of interest for addressing current controversies about the great variability in work ability and caloric needs. In poor, traditional Jewish communities, collecting intake and activity data on the Sabbath would skew results upward for nutrient intake, downward for work; yet missing the Sabbath entirely would not fully account for weekly nutrient intakes or regeneration of work, as well as spiritual, energies. Similar examples can be cited for most cultures.

In assessing the relationships between patterns of consumption and work, then, records of such variations should help qualitatively, if not quantitatively, to identify the significance of such patterns. Additional factors to be considered in presenting and analysing time-activity data are: (a) seasonal work demands, such as the extraordinary time demands on women agriculturalists during peak planting and harvesting seasons which often interfere with food preparation and child care; (b) children's school schedules, which may conflict with the ordinary family eating schedules and thereby jeopardize adequate food intake and health status; and (e) religious holidays, often whole seasons, which alter ordinary work and food-intake patterns.

Ethnographic Interview

It is advisable to carry out ethnographic field research prior to the systematic activity observations, and such ethnographic observations should continue throughout the research period. To supplement other data on activity and cultural schedules throughout the year, one will want to collect information on the following qualitative features of cultural time sense, as research abilities and research time permit: normal or preferred allocations of time; spacing and pacing of activities; and judgements of efficiency. Useful questions may be: What are the ranges of time usually allotted to specific tasks? What characteristics of individuals or technology improve or decrease efficiency of performance? Do people have notions of how long it should take to perform typical tasks, i.e. do they have this notion of "time sense"? Do they have standards that they apply informally to people in their culture about performance in time? Do they have ideas of rushed versus leisurely pacing, and do they relate their own endurance in the performance of certain tasks to nutritional or health status? Do they associate food intake directly with work, work performance, and degree of weariness in performing tasks and health status? Such information from the cultural or folk perspective may also contribute to evaluation and interpretation of intra-household distribution of food, work performance, and the relationship between nutrient intake and infection from the scientific perspective.

Procedures for collecting and analysing time-activity data efficiently and effectively

If one is interested in studies of women's time allocation, children's work and play patterns, and the organization of household activities in relation to food, health, and nutrition, it is probably best to collect data in sequence. By interview (activity-recall) methods, one can collect data on activities, income, and food purchases in standardized sequences (see descriptions of Nag, White, and Peet, 1978), so as not to overburden the respondents. Later, dietary (consumption) data and nutrition-health data can be collected. Or, one might observe activities (long-day observations) and later collect food and income data by interview. Alternatively, one can try to collect data on a range of questions, including activity profiles, food intake, and parent-child interaction, through long-day observations. In the following section, I describe the procedures and perils of collecting and analysing both activity and nutritional data simultaneously, for focal females and children, if not all household members.

In the study in Mexico by Messer (1981), activity patterns and food intake in a Mexican community were studied. Messer was interested in general in how the organization of household labour, including children's labour, contributed to income, general home care, food consumption by different household members, and health. She was also concerned with learning how children's activities and health status might interfere with the focal female's or general household's well-being.

Messer approached the work by observing one woman's activities in each selected household over full-day periods, and measuring nutrient intakes of all persons eating in the household. As noted above, households were selected according to occupations, and numbers and ages of children.

To acquire household data, an observer was in the household from 7 a.m. to 7 p.m., the usual waking hours of women and children. Activities which went on before the arrival of the observer and after her departure were recorded during interviews after the observation day. Activity data were collected by continuous reporting: using a watch, the observer wrote down the inception and cessation of all activities and interactions, describing the actors and what they were doing. She then listed activities, by times, summing durations, according to standard categories: food preparation, household work, cash work, leisure, visiting, shopping, personal hygiene, and child care. Tasks such as laundry, animal husbandry, and sweeping were recorded under the general category 'household," and tortilla and meal preparation under the general category "food preparation," in the original field diary record, and were then converted to the main category in the initial time computations. Primary child care (without opportunity cost to other activities) or secondary child care (with other activities proceeding simultaneously) were counted in summing times allocated to child care. Observers found they could usually record and sum time records for the focal women simultaneously. Activity profiles of focal women, in hourly schedules, as well as proportions of time allocated to each activity, were analysed from these data. The data were presented by person, and compared across households, to show essential differences in women's work patterns and food-allocation patterns across households.

This method of data collection had the advantage that it enabled the observer to collect data on questions of children's nutrient intake and activity patterns, alongside that of the focal female. From comparative qualitative and quantitative records of tasks performed, the method allowed for efficient characterizations of task performance, and sanitary and health factors potentially affecting the focal female's time and children's nutritional well-being. Children's foraging patterns were followed. It was noted whether or not the focal woman in the household consistently supplied young chidren with food, whether they "foraged" in surrounding brush or their own cupboards, or "ate around" at the different hearths of extended family compounds. Since most residence is patrilocal in this community, grandparents, aunts, even sisters and brothers supplied food on demand to youngsters, who also gathered fruits in the bush. Meal patterns were analysed. It was noted that the poorest families regularly prepared only one hot meal per day, in contrast to other families, which prepared two. In many cases schoolchildren also ate only one hot meal per day and substituted snacks (tacos, sandwiches, or sweets) during their school recess period.

The pace of activities was analysed. A standard activity like tortilla manufacture was seen to vary not by the human physical energy availability of the focal female, but rather by the fuel available, the quality of the masa (dough), and the numbers of other activities (food preparation, child care) that were carried on simultaneously. Also, the organization of the task could vary, depending on the numbers and ages of other female children in the household.

It was also possible to observe any extra-household factors that affected health and hygiene. For example, in one household, a young mother with four sons under five years of age struggled in vain against illness. Although she herself washed all dishes and bottles in soapy hot water, and boiled all water and milk for her household, the family lived in a multiple family compound, where her four sons played constantly with the other children. Although they ate clean, hygienic food, drank from sterilized bottles, washed their hands with soap before eating with her, and also after defecating, they were constantly exposed to dirt and faeces deposited in the yard by the other family, the less sanitary food they shared in their snacks, and their continuous mutual infections. The focal mother was forced to clean a continual layer of faeces from her patio as well as clean up after her constantly sick children.

While the time the mother spent cleaning up after the children may not have been large, she commented, and it was possible to observe, how infections kept her children unhealthy and small, and wore her down. Furthermore, she had had some nutrition/ health/sanitation education, yet her children were recurrently ill. Such observations of unhygienic residential layouts beyond the control of the individual household suggest why exposure of mothers to nutrition and health education does not necessarily result in superior nutrition and health in their children.

All-day observations also enabled the observer to record the complete feeding habits of children. While children were observed to be anorexic during bouts of illness, such as bronchitis and fevers, healthy youngsters were observed to eat almost continually, munching on bread, fruit, and tacos, and having sweet beverages hourly, particularly on days following illness. Additionally, children under five generally fed constantly on bread, fruit, and beverages in and around their own houses, and also had the habit of snacking in neighbouring (particularly relatives') households. Observation explained in part why children seemed to eat very little at meals; they ate small quantities constantly throughout the day. These data were presented by time and type of food eaten per child per observation day.

In addition to these advantages, continuous recording by the observer creates a data set that can be referred to at a later time to answer questions about parent-child interactions and children's activity levels. Time and descriptive data can be analysed to indicate (a) time "bottlenecks" when cash work competes with time for food preparation and inhibits the focal female from providing better food and health care; (b) the more general implications of women's participation in the cash economy for the time available for household and child maintenance, and the potential effects of labour-saving technologies; (c) related changes in dietary staples due to food preferences but also to the time required to prepare them; (d) patterns of adult-child interaction and how they affect children's food intake, emotional style, cognitive development, and health; and (e) the potential acceptability of nutrition and health programmes, given food habits and women's schedules.

The disadvantages of the combined method for collecting nutritional/activity data, however, are also great. While the quantity and quality of observations are enormous, so is the time required for processing the data for within-culture analysis or between-culture comparisons. The original record must be sifted through for each separate aspect of the study, since it is not feasible to precode every possible aspect of nutritional and behavioural relevance. Samples must necessarily be small, unless one has a large research team. Even so, numbers of days of observation will be limited, particularly if one follows the Philippine procedure (Evenson et al., 1978) of spending one day just accustoming the household to one's presence for every day of recording. Where there are weekly cycles or seasonal variations in activity schedules, one must carefully schedule one's research time to capture their significance for time use and diet without skewing one's interpretation of what usual time allocations are and how they influence nutrition. Finally, this method is very intrusive; it may not be acceptable in all environments and must be very carefully implemented if combined with large, multidisciplinary studies that are constantly in households collecting other kinds of health, activity, and economic data. Otherwise, one will end up altering the very phenomenon - activities in time - one is trying to study.

Even with these problems, however, the quality of observational reports and the possibilities for suggesting which factors of time use and household organization influence nutrition are probably greater than if time data are collected by other methods..

So far, the analytical insights ascertained by this method have been qualitative rather than quantitative - for example, the suggested effect of residential patterns on hygiene, infections, and nutrient intakes of youngsters, and the toll these take on the young mother. The video-like script remains to be analysed.

If nutritional anthropologists are seriously interested in doing detailed studies of household behaviours, and especially children's behaviour, in order better to understand the linkages between nutrition and human behaviour, they would do well to review the methodologies for observation and data analysis currently being developed by some primatologists (Clutton-Brock, 1977; Altmann, 1974). These researchers simultaneously observe and code their data in standard behavioural units and are advanced in developing interactive data analysis programmes for expediting interpretation.

Conclusions

There are a number of potential applications of time-allocation study methods for nutritional anthropological questions. Few would question the assertion that the organization of household activities, particularly by the focal female, is an important determinant of the nutritional status of household members. The Protein-Calorie Advisory Group (PAG) of the United Nations, for example, has suggested that more studies of women's activity patterns and schedules are needed to evaluate adequately the factors leading to undernutrition, as well as the hindrances to greater economic participation by women (1977). More data on focal women - their activity schedules and patterns of interaction with children - would give a finer index of women's work and its potential impact on household diet than simple correlations of their economic (occupational) role with the nutritional status of children. Studies of the expenditure in rime, not just in cash, can provide another dimension to understanding food choices and householdresource allocations.

Detailed studies of household organization and of how some households manage to allocate time to numbers of cash-producing and socializing activities, yet still eat well while others languish, suggest that data are needed on the structuring of activities in time, rather than simply the time use of individuals as representatives of social categories. Day-long observations in nutritionally successful versus less successful households, facilitated by interviews on how tasks are organized, would probably yield the most complete results. Purposeful selection of households for observation could proceed after initial sorting by nutritional status of children, occupational status of adults, or particular political positions of focal females. While it is interesting to know the average time allocations of focal women, it is also useful to know the range and variations of activity patterns, e.g. how the community leaders pass their days in contrast to less outstanding members of the community. It is also important to know if those households with fewer malnourished are operating differently from others, as shown by time activity profiles and the social organization of production and consumption.

There remains the question of how many persons and how many observations one needs to make a study reliable. These will depend on the variance of the study population. Also, is the outcome jeopardized if households are chosen opportunistically rather than randomly, and visits scheduled for convenience, rather than by some random visit principle? Again, initial observations in the culture under investigation provide the major clue. Are activity schedules and activities significantly different across households? Are tasks performed at constant task rates or does work density or efficiency differ? Is it reasonable to rely on a random sample if one wants to be sure to cover a range of different daily schedules on limited resources? Perhaps the best solution to these problems is a phased study, which collects community data by spot observation or interview methods (the easiest to precode, code, and analyse). Preliminary findings can then be used to select households for the study of more specific questions on organization of work, nutrient intake, and activity scheduling by intensive observation, interview, or diary methods.

If possible, it is preferable not to do all recording tasks simultaneously, or at least not on the same days. While Messer managed to collect by observation and food-weighing day-long activity data and food intake, the study was designed as much to develop methods to analyse nutrient intake in relation to "function" as to give comprehensive data on different food intakes among households in the community. It is probably better to collect food and activity data in sequence: to use activity recording days to cite what is eaten, then to record what is eaten, quantify the items, and analyse nutrient content on subsequent, regularly scheduled occasions (e.g. 24-hour activity recall, full or 12/13-hour-day direct observation, 24-hour food recall or direct observation). It may also be more important to collect cycles of food intake/expenditure by week, rather than by day; daily or even two- or three-day "averages" may be inaccurate. All these factors suggest that, it possible, nutrient intake data should be recorded separately from activity data, although all activity data will contain information on food.

Finally, some cautionary notes are in order. There are diverse ways to collect activity data in time, and these serve different purposes. Be sure of your major objectives (e.g. to understand how a focal female's time allocation affects food provisioning in individual households). Be sure as well of your "hidden agenda," for example, to understand how the significance of subsistence activities and the time allocated to them change as a result of the different occupational organizations of households. Choose a method that will allow you to collect data to answer these questions, given your time and resource constraints. Be familiar with the other activity studies, whether or not recorded in terms of time profiles, of the region in which you are working, or of the categories used to describe children's behaviours, if you are interested in comparing child development in different households and relating this data to the nutrient intake of children. This kind of review will help you to record data in cross-culturally comparable categories for your geographical or intellectual area of study.

If you are collecting data by diary, recall, or spot observations, make sure you check for accuracy periodically by observation. It may be that one of these methods is the most efficient for collecting data on activity frequencies; this will enable you later to sketch in some of the background data missed by oral or written self-reports or random visits. For example, such methods may underreport the time that children spend playing or working, if the numbers of reports are limited.

If you are principally interested in children's nutrient intake, a child-following method may be the only appropriate one, but without substantial labour resources this will necessarily limit your sample. It is probably worth while, in such cases, to consider methods developed in primate studies to report dietary results (Hladick, 1977, pp. 328, 340).

In conclusion, anthropologists and related social scientists have been studying activities, and their significance in time, as segments of more general ethnographic studies, but they have not, for the most part, focused on the questions of vital nutritional interest. This chapter has identified the essential functional (behavioural) measures of nutrition, reported in terms of activities, which might productively be studied "in time." It has also discussed at length what the methods for studying time allocation and activities are, with their advantages and disadvantages for shedding light on particular sets of social, economic, psychological, and cultural questions. The processes of data collection are time-consuming, as are the periods of data analysis, but it is necessary to allocate some of our anthropological resources to the task. I have not come to any firm conclusions on the circumstances under which it is more productive to pursue systematic qualitative data collection, rather than complex quantitative analytical frames. The values of different methods will be determined by the questions. My only cautionary note is: in the beginning and the end, one should never sacrifice potential qualitative categories for understanding the cultural processes by which children and other human beings get nourished in the interest of designing a quantitative instrument.

References

Acharya, M., and L. Bennett. 1983. Women and the Subsistence Sector. Economic Participation and Household Decisionmaking in Nepal. World Bank, Staff Working Paper, 526. World Bank, Washington, D.C.

Altmann, J. 1974. Observational Study of Behavior: Sampling Methods. Behav., 49: 227-265.

Asia Society. 1978. Time Use Data: Policy Uses and Methods of Collection. Report of a work shop by the Asia Society in New York. Asia Society, New York.

Bloch, M. 1979. Maternal Employment and Young Senegalese Children's Behavior. Paper presented at the symposium Studies of Children 11: Worldwide, annual meeting of the American Anthropological Association, Cincinnati, Ohio.

Cain, M. 1977. Household Time Budgets. Bangladesh Institute of Development Studies VFS Methodology Report, 1. Mimeo.

Calloway, D., et al. 1979. Precis: Collaborative Research Support Program on Intake and Function. Mimeo. University of California, Berkeley, Calif.

Chapin, F. S.. Jr. 1974. Human Activity Patterns in the City: Things People Do in Time and Space. Wiley Urban Research Services, New York.

Chavez, A., and C. Martinez. 1979. Nutrici desarrollo infantil. Interamericana, Mexico City.

Clutton-Brock, T. H., ed. 1977. Primate Ecology: Studies of Feeding and Ranging Behaviors of Lemurs, Monkey, and Apes. Academic Press, New York.

Erasmus, C. 1955. Work Patterns in a Mayo Village. Amer. Anthropol., 57: 322-333.

Evenson, R. F. 1981. Food Policy and the New Home Economics. Food Poll. 3: 180-203.

Evenson, R., B. Popkin, and B. K. Quizon. 1978. Nutrition, Work, and Demographic Behavior in Rural Philippine Households: A Synopsis of Several Laguna Household Studies. Yale Economic Growth Center, New Haven, Conn.

Fortmann, L. 1981. The Plight of the Invisible Farmer: The Effect of National Agricultural Policy on Women in Africa. In: R. Dauber and M. L. Cain, eds., Women and Technological Change in Developing Countries, pp. 204-214. AAAS Selected Symposium 53. Westview Press, Boulder, Colo.

Franklin, D., and 1. V. de Vald 1979. Desnutricinfantil y su relacion el tiempo y las habilidades de la madre. Separata de cuadernos de econom 16: 343-358.

Gaulin, S., and M. Konner. 1977. On the Natural Diet of Primates including Humans. In: R. J. Wortman and J. J. Wortman, eds., Nutrition and the Brain, vol. 1. Raven Press, New York.

Hayden, B. 1981. Subsistence and Ecological Adaptations of Hunter-gatherers. In: R. Harding and G. Teleki, eds., Omnivorous Primates: Gathering and Hunting in Human Evolution. Columbia University Press, New York.

Hemmings-Gapihan, G. 1981. Baseline Study for Socioeconomic Evolution of Tangaye Solar Site. In: R. Dauber and M. L. Cain, eds., Women and Technological Change in Development Countries. AAAS Selected Symposium 53. Westview Press, Boulder, Colo.

Herrera, M., et al. 1980. Effects of Nutritional Supplementation and Early Education on Physical and Cognitive Development. ln: Life-span Developmental Psychology, pp. 149-184. Academic Press, New York.

Hladick, C. M. 1977. A Comparative Study of the Feeding Strategies of Two Sympatic Species of Leaf Monkeys: Presbytis senex and Presbytis entellus. In: T. Brock, ea.. Primate Ecology. pp. 324-353. Academic Press, New York.

Jochim, M. 1981.. Strategies for Survival. Academic Press, New York.

Johnson, A. 1975. Time Allocation in a Machiguenga Community. Ethnol., 14: 301-310.

Kumar, S. K. 1977. Role of the Household Economy in Determining Child Nutrition at Low Income Levels: A Case Study in Kerala. Occasional Paper, 95. Department of Agricultural Economics, Cornell University, Ithaca, N. Y.

Lee, R. 1969. !Kung Bushman Subsistence: An Input-Output Analysis. In: A Vayda, ea., Environment and Cultural Behavior. The Natural History Press, Garden City, N. J.

Mead, M. 1955. Cultural Patterns and Technical Change. Mentor, New York.

Messer, E. 1981. A Ten Year Perspective on Mitla, Oaxaca, Mexico. Report to the Social Science Research Council.

Minge-Klevana, W. 1971. Household Economy during the Peasant to Worker Transition in the Swiss Alps. Ethnol., 17: 183-196.

______.1980. Does Labour Time Decrease with Industrialization? A Survey of Time Allocation Studies. Curr. Anthropol., 21:279-296.

Moore, W. 1963. Man, Time, and Society. John Wiley & Sons, New York.

Nag M. J., B. White, and R. C. Peet. 1978. An Anthropological Approach to the Study of the Economic Value of Children in Java and Nepal. Curr. Anthropol., 19: 234-306.

Ong, A. 1982. Women and Industry: Malay Peasants in Coastal Selangor, 1975-80. Ph.D. thesis. Columbia University, N.Y.

Popkin, B. 1980. Time Allocation of the Mother and Child Nutrition. Ecol. Food Nutri., 9: 1 - 14.

Popkin, B., and F. Solon. 1976. Income, Time and the Working Mother and Nutrition. J. Trop. Ped. Environ. Child Health, 6: 156-166.

Protein-Calorie Advisory Group (PAG) of the United Nations. 1977. Women in Food Production, Food Handling, and Nutrition with Special Reference to Africa. United Nations, New York.

Quizon-King, E. 1978. Time Allocation and Home Production in Rural Households. Philippine Econ.. J. 36 (1-2): 185-202.

Rappaport, R. 1967. Pig for the Ancestors. Yale University Press, New Haven, Conn.

Richards. A. 1939. Land, Labour, and Diet in Northern Rhodesia. Oxford University Press, London.

Rogoff. B. 1978. Spot Observation: An Introduction and Examination. The Quarterly Newsletter of the Institute for Comparative Human Development (Rockefeller University), 2: 21-26.

Scrimshaw, N. S. 1982. Programs of Supplemental Feeding and Weaning Food Development. In: N. Scrimshaw and M. Wallerstein, eds.. Nutrition Policy Implementation. Issues and Experience. pp. 101111. Plenum Press, New York.

Sharman, A. 1970. Social and Economic Aspects of Nutrition in Padhold Dukedi District, Uganda. Ph.D. thesis. University of London, London.

______.1980. Dietary Choice and Resource Allocation by Household Members. Paper presented at the 79th annual meeting of the American Anthropological Association, Washington, D.C.

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

Thompson, E. P. 1967. Time, Work Discipline and Industrial Capitalism. Past and Present. 38: 56-97.

Tripp, R. 1981. Farmers and Traders: Some Economic Determinants of Nutritional Status in Northern Ghana. J. Trop. Ped. Environ. Child Health, 27: 15-22.

Wallman, S. 1969. Take Out Hunger. Two Case Studies of Rural Development in Basutoland. The Humanities Press, New York.

Whiting, B. and J. W. M. Whiting. 1975. Children of Six Cultures. Harvard University Press, Cambridge, Mass.

Winterhalder. B., and E. A. Smith, eds. 1981. Hunter-gatherer Foraging Strategies. University of Chicago Press, Chicago, 111.

Appendix 1. Messer, mitla field notes, 4 September 1980

7 Ma making tortillas (L[5 yrs], baby [1.5 yrs] watching her. Ch [4 yrs] still asleep in house).
7.20 Ma sends L to get the tortilla basket - if she won't fall down.
Ch gets up. Ma asks if his cough is gone. He says Yes.
Ma breaks out a new bottle of Nescaf$60 pesos, note in Conasupo $57).
7.25 L says that she wants to wash her face. Ma and Pa say yes. There's water.
7.26 Ma puts up pot for almuerzo [first main meal].
Ch says he wants to drink [coffee].
Ma puts 2 tablespoons lard in pot, for cooking eggs.
7.30 Ma sends Ch with Pa to get bread from inside [bread stored in basket by altarplace].
Ma calls Pa to eat.
Pa comes. Plays with baby.
Baby: drinks a bit of Fanta [orange pop], 15 g bread, 20 g egg with lard.
Pa: 1 (200 ml) bowl coffee, 43 g bread, 90 g, 84 g, 83 g tortillas, 1 piece egg, 1 chilito, 10 g pasta with chilito.
Ch: 1/4 cup coffee, 41 g bread (he gives a piece to baby; tells her to blow on it [since it is soaked in hot coffee]. 13 g egg, 24 g tortilla.
7.35 Ma tells L to bring her petate [palm mat]. She looks for it. Ma tells her where to find it.
L imitates her: "L, L, L." She comes back with two pieces of bread.
7.37 Pa sends L for can-opener, but Ma goes after it since she knows it is high up.
Ma: 1 bowl coffee, 48 g bread. 74 g, 39 g, 40 g tortillas, 26 g pasta (leftover), 24 g egg, 3 chilies.

Calculations of time of mother

7-7.37 Food preparation, tortillas

Child care

(Times are calculated as day proceeds, just summing minutes from starting to end of activities.)

Time allocations (summing minutes over the day allocated to food preparation, child care, household maintenance, rest, cash work, visiting, shopping)

FP CC
49 (12)a
12 (25)
63 (23)
23 (20)
15 20
(32) (15)
  20
197 45
(229) (140)

Then calculate these totals as percentages of total time observed. Compare across days.

a. Parentheses indicate mother was engaged in more than one activity at the same time.