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.