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.