-: ....'.....--............'-.-- ^ --L-- .. .. ii---. -- l
.. ,t Tfhel rwersty-,-i on
GENDER ISSUES IN FARMING SYSTEMS
RESEARCH AND EXTENSION
-- An Evaluation of
Methodologies used in Time Allocation Research
Department of Forestry
University of California
Berkeley, California 94720
Time allocation studies are frequently used to describe gender
and age based labor patterns. In farming systems research (FSR),
labor patterns have been analyzed to support a wide range of
findings, including the determination of peak labor periods
(Maxwell 1984, Price and Barker 1978), income opportunities for
female farmers (Burfisher and Horenstein 1985), the contribution
of children to farm production (Navera 1978), crop labor
investments (Barlett 1980), seasonal fluctuations in agricultural
and nonagricultural activities, (Norman eg ai. 1981) and
interhousehold differences in the family cycle (Cadeliha 1985).
The strengths, weaknesses and variations of time allocation
methodologies are rarely discussed. This paper presents an
evaluation of commonly used methods of data collection in time
allocation studies. The first section of the paper is a review
of general characteristics of time allocation studies found in
the literature. Issues requiring closer scrutiny are identified.
The second section describes four methodologies that were used in
a 1984-85 study of shifting cultivators in Negros Oriental, the
Philippines. Each of the methodologies is evaluated according to
how well it contributed to the cost effective understanding of
upland farming patterns and the organization of household labor.
Some implications for the use of time allocation data in farming
systems research projects are discussed.
Why Time Allocation Studies?
Time is a resource. Individuals must make decisions about how to
manage their time. These decisions reflect the constraints and
opportunities surrounding human goals. Farmers lead an economic
existence in which they must allocate their time among
agricultural activities. Poor farmers generally have better
access to labor than to land or capital. Information about time
allocation can illuminate facets of farm household behavior that
Paper to be presented at the Gender Roles in Farming Systems
Research and Extension Conference, February 26-March 1, 19866
FSSP, University of Florida, Gainesville.
are not obvious from other types of data. Most small farmers in
developing countries operate at the household level as a unit of
both production and consumption. As a result, the activities
of all household members are tied to the farm enterprise.
Nevertheless, the role of women, children and the aged is often
overlooked or discounted because an analysis focuses on income
generation or male "heads of households" ( see, for example,
Hayami et al. 1978; Maxwell 1984).
Time studies can show the type and quantity of labor that
different members of the household contribute. Most subsistence
level farm households engage in more than one means of support.
Their livelihood might be based on crops, fishing, livestock,
off- or non-farm wage labor, cottage industry, and forest
products. Time use data can show how a household distributes its
labor resources among such activities, and how subsistence
activities compete with time needed for childcare, cooking,
cleaning, sleeping, leisure or education (Mueller 1979).
GENERAL CHARACTERISTICS OF THE METHODOLOGY
We will limit our discussion of the many methods of time
allocation research to those relevant to gender and farming
systems. The methods can be described according to (1) the
information objective of the research, (2) the way in which that
information will be collected, and (3) the way in which the data
are interpreted. There are various dimensions to each of these
The information objective is determined by the purpose of the
research. This purpose will affect the structure of the
information, i.e. whether the data focus on certain tasks,
certain people, certain locations, or all of these factors at
once. Note that a research design might include all of these
variables, but is likely to focus only those variables of
principal interest. For example, a corn production study might
use crop activities as an "organizing variable;" data on location
or people would be collected only as they relate to crop
activities. Such a focus minimizes the amount of nonrelevant
information, but maximizes the probability that some
unanticipated, but significant factors will be overlooked. The
comprehensive method allows for the unanticipated by requiring
the researcher to examine all aspects of farm time allocation.
Costs are directly proportional to the amount of information
collected; therefore, the comprehensive approach might be best
applied as an initial baseline study to aid in the identification
of variables demanding more intensive study.
Research designs for time use studies need to consider which
levels of disaggregation are appropriate to the population and
the research objectives at hand. Information can be
disaggregated by social units (community, household, individual),
demographic criteria (sex, age, position in household),
socioeconomic criteria (land tenure, income, assets), biophysical
characteristics (climate, soil, vegetation) and, in the case of
FSR, farming system criteria (market crops, subsistence crops,
livestock). The temporal units and boundaries of the research
need to be chosen in a way that best reflects the many rhythms of
a farmer's experience: the fluctuations of the year, season, day,
labor supply and demand, a given crop schedule, the family
development cycle, or local school calendar.
Time allocation information has been collected with the use of
(1) respondent recall (Mueller 1979, Moji 1985), (2) direct
observation by the investigator (Johnson 1975), and (3) farm
household record keeping (Price and Barker 1978). Recall and
record keeping are convenient for large samples; relatively
little time is required per case. Direct observation tends to be
more accurate, but also more time consuming, both in collection
and analysis. (See Anker 1980, The Asia Society 1978, and
Birdsall 1980, cited in Acharya and Bennett 1982: 64, for further
discussion of the differences between recall and observation).
Each method of information gathering is subject to the biases
introduced by either the enumerator or the farmer. Recalled
information passes through the filter of the farmer's memory as
well as the filter of the enumerator's ears and interests; it is
therefore subject to the potentially greatest bias. Direct
observation data have the potential of less "filter" bias, but
are influenced by very presence of the investigator. The extent
of this latter bias depends on the nature of the relationship of
the enumerator with the community, as well as the manner in which
local households treat guests. Long periods of acquaintance are
likely to minimize inadvertent influence on respondent behavior.
This is one reason for the extended duration of many
anthropological studies (Lewis 1959: 5; Rappaport 1968,
Malinowski 1922 ).
It is difficult to follow all members of a household to the
locations of their activities. Researchers typically stay in the
home, observe comings and goings, and ask questions about
activities that they were not able to witness firsthand. In time
allocation studies for FSR, an alternative approach might be
adopted, since many agricultural tasks may take place out of view
from the home. One option might be to have more than one
enumerator per household so that enumerators could accompany
household members to their plots.
Sampling strategies can help to minimize bias in time allocation
research. By using instantaneous observations, Erasmus (1955),
avoided collecting data that reflect the enumerators" presence.
Johnson (1975) and Acharya and Bennett (1982) randomized
observations in order to make statistical estimates of the
distribution of time among activities. The longer the period of
an activity, the greater the probability that it was observed in
random spot checks. Time allocation with this technique is
determined by the frequency of observing a given activity. Costs
associated with lengthy periods of observation are minimized.
Whether sampling is random or systematics instantaneous or
continuous, the investigator still needs to determine how often
he will take samples--hourly, daily, weekly, or seasonally.
Trade-offs between costs and detail in information plague every
aspect of time allocation research.
The use of random observations lends itself to a representative
description of farmers' time allocation. Most research is
designed to describe "representative" or "average" conditions.
Outlying cases are often removed from analyses in order to
achieve a better fit between observations and a calculated mean
or regression line. Yet, infrequent or irregular activities may
be significant to an understanding of farm behavior. How does a
household recoup losses after a once-in-a-lifetime typhoon has
destroyed a entire standing crop of corn? How do religious
holidays affect the balance of leisure and agricultural
production? Do the activities surrounding childbirth disrupt a
household's agricultural work schedule?
Sample size is another critical element of the research design.
Case studies are not necessarily representative of the larger
population, but may offer in-depth information that is missed in
large surveys (Maxwell 1984). Both surveys and case studies have
been used successfully although often with different purposes
(see Lewis 1959 and Erasmus 1955 for examples of case studies;
Acharya and Bennett 1982, Hayami et al. 1978, and Hart 1980 for
examples of surveys). Most reports on farm time allocation
mention the total number of individuals or observations as
evidence of the study's statistical weight (Mueller 1979, Johnson
1975). While these figures give the reader a better picture of
the sample, the number of households seems to be the critical
variable in populations where farms are household enterprises.
Interpretations of Time Use Data
Time allocation data are assumed to offer useful information to
the FSR practitioner. The data are sometimes used as an
indication of labor or energy "expenditure." Input-output labor
budgets have been calculated to determine the productive
efficiency of different means of subsistence (Sahlins 1972,
Rappaport 1968). In other cases, wages are have been imputed to
estimate a farm's economic efficiency (Price and Barker 1978).
Care should be taken not to equate time and labor. Time is only
a convenient indicator of labor inputs; it indicates the
allocation and duration of labor, not the intensity of effort or
the economic value of that effort.
The amount of time spent in any one activity may not be an
adequate indication of the importance of that activity to the
farm economy.. For example, market transactions take relatively
little time, yet the household might depend on the cash income
for tools, medicine, or clothes. Furthers activities often occur
in sequences; the events of one day may affect the activities of
the following week. The farmer who was not able to sleep one
night might sleep twelve hours the following day. A week of
rain and inactivity may be offset by long hours spent weeding on
the first sunny day. The methods commonly used to study time
allocation do not easily account for such sequences.
Time use studies have typically treated an individual's, daily
routine as a series of single tasks carried out by a single
person (Johnson 1975). Joint activities such as child care and
weeding, or fuelwood gathering and walking home from the market
are not treated adequately. At what point does an activity
become a joint use of time if the individual alternates
repeatedly between tasks? Activities involving more than person,
such as cooperative harvesting groups, are also not described
well in most time allocation studies. Even when the occurrence
of joint activities is recognized, it is still analyzed as if the
tasks were performed separately (Acharya and Bennett 1982).
Farm behavior is not discretely organized into pre-coded
categories. Categorization is a device used by the investigator
to reduce the details of everyday farm life into manageable
units. The arbitrary definition of these categories and the
sometimes biased process of classifying activities can lead to an
equally arbitrary or biased interpretation of the data. See
Johnson (1975: 305-307) for an example of how the relative amount
of time spent by men and women in "productive" labor varies with
the definition of production.
To compensate for these shortcomings, various authors have
stressed the need to consider time allocation in the context of
other information (Chibnik 1980, Stone and Campbell 1984, Acharya
and Bennett 1982). Time use data can be used as the quantitative
counterpart of a broader examination of household decision
making. Ethnographic descriptions derived from participant
observation and in-depth interviews can serve as a complement to
and crosscheck on data limitations, regardless of the scope of
the query. Detailed time allocation studies are expensive; the
same conclusions can sometimes be drawn from other, cheaper
sources of information. Using P/O to note which activities tend
to be done by men or women may be sufficient for most purposes
(see Wiley 1985: 184) To minimize the expense associated with
the collection of time use information, and to avoid the
accumulation of detailed data that "never gets processed anyway,"
it is suggested that researchers streamline time allocation
studies according to the purpose at hand. The systems
perspective may be maintained by backing up time data with
information from cheaper sources. With the exception of baseline
studies, the more focused the study, and the fewer the variables,
the more useful information the researcher is likely to derive
from the data.
This first section of the paper has presented a conceptual
framework for the analysis of time allocation methodologies. The
framework is derived from the wealth of experience that other
researchers have had, and therefore includes ideas that are
probably familiar to those who have worked with time allocation
before. The framework is most useful to those designing new
studies or analyzing previous studies. A series of
methodological questions have been posed: What is the
information structure of the time study? What are the variables
of greatest interest? What is the information source--recall,
direct observation or household records? What is the sampling
strategy? What are the biases, costs and benefits of the method?
How are the data to be interpreted? Is time a surrogate measure
of another variable of interest? How does the method account for
spurious events, joint activities, sequences, or different scales
of time? Are there other sources of information to serve as
cross-checks? The second half of the paper gives the reader an
opportunity to apply the conceptual framework to an FSR project
in which four time allocation methodologies were used.
TIME ALLOCATION AT LAKE BALINSASAYAO
This section of the paper discusses the four approaches that were
used in the collection of time use data in Balinsasayao. Data on
the household distribution of time, broken down by task and
gender, are presented. The four approaches are compared
according to how well they contributed to the cost-effective
understanding of farm behavior.
Site and Project Background
The Balinsasayao Rainforest is at 9b 21' N latitude, 1236 10'E
longitude in the province of Negros Oriental. Lowland farmers
began migrating to the forested mountains of Negros in
significant numbers in the 1950's. As a result, several
scattered settlements exist in the Balinsasayao Rainforest. Land
ownership is not possible since the area has been designated as a
critical watershed by the national government.
The Balinsasayao community and project activities are
concentrated around a lake approximately 10 hectares in size.
Shifting agriculture, or kaingin, is practiced. Corn, root .crops
and chayote are the predominant crops. Average landholdings are
1.56 hectares. Most households control three to five different
plots. Plots are managed differently according to the distance
of the field from the household and the plot's productive
capacity. Fallows range between one and 26 years.
The nearest markets are in the lowland coastal towns, San Jose
and Dumaguete, about 12 kilometers from Balinsasayao. Farmers
walk this distance since there is not any public transportation.
The average household sells produce in the market once a week.
Chayote, squash, root crops and abaca (Musa textilis, Manila
hemp) are commonly sold in the lowlands. All of these items have
relatively low market values; the amount of produce that can be
sold in any one trip is limited by the number of pounds the
farmer can carry on his or her back. The site and project are
described in more detail in Appendix A.
Time Allocation Methods
Four time allocation studies were administered in the
Balinsasayao community between April 1984 and July 1985. Each
of the studies was designed to answer specific questions about
the organization of household activities. The first method,
participant observation (P/O), was used to determine norms
regarding the division of labor within households. In-depth,
open-ended interviews and residence in the community provided
data over a 16 month period. P/O was also used as a crosscheck
for subsequent studies.
The second method was a comprehensive case study of four
households. Households were chosen to represent the different
stages of the family cycle. The investigator lived with three of
the households for five consecutive days, and one household for
six days, resulting in 21 days of observation. The time of day,
duration, person, task, and location of task were observed. When
the investigator could not observe all of the household members
at once, he or she had to rely on the farmers' descriptions of
their activities. The objective of the case studies was to
collect a comprehensive set of data as a baseline for subsequent
The case data showed that men and women's time use differed by
field type. At Balinsasayao, fields may be classified according
to the intensity of their cultivation. Intensely cultivated
fields include crops such as corn, beans, and peppers. These
fields tend to be close to the home, have soils with a high
productive capacity, have high levels of solar insolation, have
low risk of loss due to pests or theft, Nonintensely cultivated
fields include root crops, chayote, squash, abaca and other low
input crops. The case profiles showed that womenpent more of
their time on high intensity field than on low intensity fields.
Men spent more of their time on the low intensity fields. (See
Differential allocation of time by field type was examined in
more detail in the third and fourth studies. These studies were
used to test hypotheses about gender and age-based patterns in
agricultural activities, particularly differences between use of
intensely and nonintensely cultivated fields. The third study
was a 24 hour recall of household time use. Respondents were
asked to recall the time of day and the duration of activities
of each member of the household. The fourth study was a one week
recall of field activities. In-depth information about the field
location, activity (planting, weeding, harvesting etc.),
participating household members, crops planted or harvested, and
the date of the field visit were noted. This was the only
method that did not record the duration of the activity.
The data for the third and fourth studies were gathered during a
seven month survey of ten households. The sample represented 71
percent of all households within one kilometer of the lake.
Households were visited approximately every four weeks. A total
of 53 days was recorded for all households. Female heads of
households were interviewed because they were the ones usually at
home. In two instances an adolescent and in 16 cases male heads
of households were interviewed. Since households were unaware of
the interview schedule, there were several occasions where no one
was home. One household shifted its residence to a distant field
shortly after the survey began, and was therefore only
Comparison of Methods: Quality of Data
The survey and field data gave the best representation of the
community since they were based on the largest sample size. Most
of the respondents had no difficulty remembering the previous
day's activities; everyone had difficulty recalling one week of
field visits. Field recall seemed easier when questions were
structured by crop related tasks, rather than by field visits,
i.e. asking "what did you plant yesterday," rather than "what did
you do when you went to your corn field?" The field results were
important in the way they corroborated and expanded upon the
survey data. Respondents answered in more detail about what they
were doing, who was doing it and which crops were involved. The
field questions also served as a baseline for a more in-depth,
longitudinal study (in progress) of labor inputs on different
Survey responses were short. The previous day was described
according to what the respondent considered to be the relevant
activities of the day. This was useful information in itself.
Respondents recalled primarily agricultural tasks, watching
children, marketing, and food preparation. Nonwork activities
and tasks of short duration were conspicuously absent.
The comprehensive case data show all activities for all members
over three years of age of a limited number of households.
Information was recorded for such nonwork activities as sleeping,
eating, personal hygiene, socializing, resting. Short tasks such
as fetching water, chopping wood, feeding livestock, sweeping the
floor and business transactions were also observed. Any
misinformation in the case data is due to the investigators'
interpretations of behavior or the inability to view all acts.
The comprehensive case method provided the most complete
description of multiperson and joint activities. In the survey,
respondents tended to underrepresent the number of people and the
number of tasks per activity. Children were underrepresented as
well. There are a number of explanations for this. First,
informants might consider children as helpers "in training;"
informants would be more likely to list activities according to
the person in charge, usually an adult. This distinction is
useful to an understanding of household decision making.
Informants generally neglected to mention the participation of
nonhousehold members, e.g. visiting relatives, a neighbor who
helped cook dinners or a group of friends who helped cut timber.
Direct observation and P/O showed that many activities were
indeed joint, or involved more than one person (see Results
Information derived from participant observation helped to
compensate for methodological -shortcomings in the other
approaches, for example, the limited number of households in the
case and survey and the limited duration of all the studies.
Through P/O, inferences could be made about time use during the
entire agricultural year, even though only seven months were
formally sampled. Likewise, familiarity with the community and
the idiosyncrasies of households and individuals facilitated
extrapolation about time use for the entire Balinsasayao
community. As has been noted elsewhere (Colfer 1985), P/O
provides the fundamental background information necessary for a
systems view of human behavior.
Living with a household while trying to observe its behavior
introduces the bias of the investigator's presence. This bias
was minimal due to the close rapport that both investigators had
with the community, as well as the fact that having visitors is a
common, culturally encouraged event. One of the investigators
(the author) had lived in the community conducting field surveys
and interviews for nine months previous to the time allocation
study; she had lived with one of the case households for six of
those months. The other researcher had conducted interviews and
lived in the area during the previous four months. Acts of
hospitality, such as morning snacks and afternoon coffee, were
more frequent than they probably would have been without
visitors, and consequently were overrepresented in the case
studies. Five days was a relatively long time to observe the
activities of every individual. It was exhausting for the
investigators as well as the households. After two days, most of
the household members seemed anxious to regain their privacy.
One advantage of the survey was that it was free from any bias
that might have resulted from intense observation.
Costs and Benefits
There are certain costs associated with the collection of data
for each of the methods. These are the costs of the
investigators' times the costs of the farmers' time, and the
costs of analyzing the data. These costs must be weighed against
the benefits of information according to its depth, accuracy,
precision and quantity. On a per day basis, the comprehensive
case approach was the most expensive method. The time required
to collect and analyze the quantity of information was not offset
by the amount of information that was relevant to understanding
household land use and farming practices. However, the nature of
participation in some activities would not have been quantified
as well through any other method. For example, the joint
activity of childcare and food preparation by women was not
recalled by respondents in the survey (see Results section).
P/O was the least expensive method on a per day basis, but the
level of information was too general to be of use in a
sophisticated analysis. Quantitative data are useful for
comparative work. The survey and field methods provided the
greatest amount of relevant information at the lowest cost. The
four methods were, however, complementary and served their
particular purposes well.
There are benefits associated with time studies that are unique
to FSR; the process itself generates an on-going link of
communication between the researchers and the farmers. That link
might be used for informal exchange of extension information,
expression of farmers' changing attitudes, or simply to show that
the researchers are interested and concerned about how the
farmers live their lives.
ANALYSIS AND RESULTS
Since the emphasis of the paper is on the methods associated with
time allocations the results presented here are only a brief
summary of some of the findings. This section will focus on a
comparison of the comprehensive case studies and the survey.
Data were analyzed on a computer with Statistical Package for the
Social Sciences (SPSSX) programs. Activities for each study were
grouped into thirteen major categories: land preparation,
weeding, planting and harvesting, livestock care, fishing,
gathering and forest use, processing/manufacture, marketing,
income generation and exchange labors domestic chores, daily
needs, leisure, and visiting. The types of activities included
in each of these categories are described in Appendix C. The
time recorded for each activity included associated travel time.
Time values were rounded off to the nearest hour in the survey,
but were left in fractions of hours for the case study in order
to avoid eliminating activities of short duration.
Tables 1 and 2 show the time distribution among activities for
the comprehensive case studies and the survey. According to the
case studies, households spend most of their time in daily needs,
marketing and domestic chores, respectively. The survey shows
that the most time is spent in weeding, domestic needs and land
preparation. There are several explanations that might account
for the discrepancies between the two sets of results: (1) The
timing of the case study and survey reflect differences in the
agricultural calendar. The surveys occurred during the dry
season when land preparations weeding and planting were common.
The cases occurred one month earlier, when the rainy season was
still in progress (see Fortmann 1985, Chambers et al. 1981 for
further discussion on the importance of seasonality). (2) Daily
needs, such as eating and sleeping, were generally not mentioned
by survey respondents, and therefore were underrepresented in the
Tables 1 and 2 also show the distribution of time by activity
according to the breakdown by gender. The data were not weighted
according to the relative frequencies of males and females (see
Appendix B); therefore, the figures must be interpreted
carefully. For example, one may correctly say "of all the time
spent in weeding, 64 percent was by males and 36 was by females,"
but not "that weeding is more of a male role than a female role."
The results of the case studies and survey both show substantial
differences between male and female participation in land
preparation, harvesting/planting, fishing, visiting, and income.
Statistical significance exceeds the ten percent confidence level
for harvesting/planting in both data sets.
The differences between the case studies and survey findings may
be partially explained by several intervening events. One of the
households rebuilt its roof during the survey; consequently,
women spent large amounts of time gathering E2g2D grass, Im grgta
Ylindrica, while the men thatched the roof. Another household
had two women who were sick for several days during the survey.
This probably skewed the daily needs category in the survey
Two of the results in Tables 1 and 2 can be explained by by
information collected through other sources. First, the case
data in Table 1 show that women are found in daily needs
activities more often than men. Experience from P/O suggests
that males and females spend nearly equal amounts of time in
sleeping, dressing, and personal hygiene, but women and children
tend to spend more time eating. Of the 60 percent of time spent
by males in daily needs, 83 percent is by males under 16, i.e.
children. Second, visiting is probably higher for men in the
surveys because informants tended to mention major trips
(dominantly male) out of the community, but not casual visits
within the community (dominantly female). Casual visits by women
within the community were, however, observed directly in the case
studies and in P/O. The discrepancies in the analysis support
Distribution of Time by Activity and Gender: Case
Activity Percent Breakdown by S2e
of Time Male Female
Livestock 1.3 55 43
Land preparation 2.8 100* O*
Weeding 1.4 64 36
Harvesting/planting 1.9 89* 11*
Fishing 1.2 96 4
Gathering/forest use 0.3 93* 7*
Processing/manufacture 2.6 98 2
Income/exchange labor 3.4 100** 0**
Marketing 14.2 94** 2**
Domestic chores 8.9 44 50
Daily needs 50.3 60 34
Leisure 2.3 51 47
Visiting 5.3 33* 66*
Distribution of Time by Activity and Gender: Survey
Activity Percent Breakdown by SBe
of Time Male Female
Livestock 1.7 31 69
Land preparation 12.7 93 7
Weeding 26.0 60* 41*
Harvesting/planting 8.3 60** 41"
Fishing 3.8 70 30
Gathering/forest use 1.5 48 52
Processing/manufacture 2.7 93* 7*
Income/exchange labor 7.6 87 13
Marketing 7.2 46 54
Domestic chores 19.2 42 58
Daily needs 5.6 43 57
Leisure 1.3 49 51
Visiting .7 69 31
Activities included in each category are listed in Appendix C.
Percent was calculated as time spent by all individuals in
each activity divided by time spent by all individuals in
Percent was calculated as amount of time spent in activity
by gender group divided by amount of time spent in activity
by all individuals. indicates t-test significant difference
of 10 percent **indicates significance at 1 percent level.
the contention that time allocation data should not be
interpreted in isolation. Multiple sources of information can
compensate for the inadequacy of time use data in explaining
activity sequences, sporadic events, or special subgroups of the
Joint and Individual Activities
The case and survey data show that certain activities tend to be
performed in combination with other people or other tasks. Joint
activities included weeding and land preparations weeding and
planting, fetching fuelwood and water, weeding and off-farm
labor, land preparation and off-farm labor, food preparation and
house watching, and child care and house watching. Other joint
activities, which were noted with the comprehensive case method,
but not in the survey were washing clothes with bathing,
childcare with food preparation, and any field activity that
required some travel with fuelwood collection.
Activities that typically involved more than one person were
weeding, harvesting, marketing, relaxing or socializing, eating
and birthing. There was no significant difference between men
and women's participation in activities involving more than one
person. Some activities were only carried out by individuals:
care of livestock, personal hygiene, and making tools or baskets.
Other activities involved single individuals at least 0B percent
of the time: land preparations planting, childcare, food
preparation, and laundry.
Location of Activities
Information on the location of activities was analyzed for the
survey data. The findings are summarized in Tables 3 and 4.
Sixty-four percent of the time spent on activities in the home
was contributed by women. Of the activities located at some
distance from the home, 71 percent of the time was performed were
done by men (See Table 5). This location differences reflect the
inside/outside dichotomy noted in other gender studies (Acharya
and Bennett 1982, Rosaldo and Lamphere 1974). Women tend to be
more involved in activities dealing with the home and family
life. Men tend to be more involved with public affairs and
events external to the home. At Balinsasayao, this division is
probably due to the childcare and domestic responsibilities that
women assume. There is very little time available for these
women to participate in the "outside world."
According to the survey data, women spend 1I percent of their
field time on nonintensively cultivated and e8 percent of their
time on the intensely cultivated fields. Men, on the other hand,
spend WG percent of their field time on the nonintense plots, and
3S percent on the intense plots. Women are spending significant
(chi-square test significance of .09) amounts of time on high
intensity fields, probably because these fields are usually close
Percent of Time Spent by Gender and Locationi
Intensely cultivated fields
Nonintensely cultivated fields
Out of community
Percent is calculated as amount of time spent by
gender classification divided by time spent by both sexes in all
Percent of Time Spent by Gender and Field Typesi
1Percent time arranged as row percent, column percent
total percent, respectively.
to the home. This probably also reflects the large amount
time required in weeding on these plots, and the tendency
women to spend most of their field time weeding (see Tables 1
Nevertheless, women do not necessarily have more control than men
over crop production on these fields. Tables 1 and 2 show that
men do most of the planting and harvesting--activities that
affect production much more than weeding. Furthermore, Table 4
shows that of all the time spent on intensely cultivated fields,
5v percent is by men and '4 percent is by women. On the
nonintensely cultivated fields, 71 percent of the time is by men
and Qq percent is by women. A clear dichotomy exists, but the
- - - - - - - - -
conclusions must be stated with care. Men appear to have more
control of the less intensely cultivated fields, but women
probably do not enjoy a parallel control over the intensely
cultivated plots for the reasons cited above.
Substitutability of Household Members
The P/O information shows what respondents considered to be the
typical division of labor, and what the researcher observed to be
the usual flexibility among these divisions. A comparison of the
case, survey and P/O data show that certain tasks accommodate
more substitutability than others. For example, fetching water,
weeding, and gathering fuelwood were jobs that anyone in the
household could and would do. Activities that were more
restricted to a particular position in the household, or
gender/age group included washing clothes (women), home repairs
(men), and fishing (children). Information about
substitutability would be relevant to determination of labor
CONCLUSION: IMPLICATIONS FOR FSR
Time allocation studies can be. useful to the analysis of farming
system behavior, yet the methodology used to gather this data is
not always made explicit (see International Labor Organization
1981: 40, Gascon 1979). It seems critical to state which methods
were used, not only to facilitate interpretation of the findings,
but also to guide researchers in the future design of other time
The Balinsasayao research shows that time allocation studies can
provide useful information about gender to the FSR and extension
practitioner. The inside/outside dichotomy between men and
women and the locus of their respective activities suggests that
the design, introduction and evaluation of new technologies must
take gender into account. Technological innovations that affect
nonintensely cultivated field will involve men more directly than
women. Likewise, the predominance of one sex in certain
activity categories, such as men in land preparation and women in
food preparation, indicates the need to introduce technologies
suited to the activity's gender classification.
Communication links are vital to the process of research and
extension that is at the core of FSR. Since men make more visits
and are involved in out-of-home activities more often than women
are, it is more likely that they have better access to
information. Since settlement at Balinsasayao is scattered,
women who stay at home must rely on visitors and family members
to bring them information.
The experience of the Balinsasayao research suggests that 24 hour
day, comprehensive case studies can be used to gather reliable
baseline information. Short duration (4-5 day) case studies
might be conducted as part of a rapid rural appraisal to provide
data for the diagnostic stage of FSR. The all-encompassing
nature of a comprehensive, full day case study insures that the
nonobvious details of household behavior will not be overlooked.
The comprehensive cases may be followed up with more focused task
or individual selective studies during the latter stages of the
Time allocation methods are well suited to the interdisciplinary
nature of FSR. The data can provide information about planting
and harvesting schedules for the agronomist, details on labor
supply and demand for the agricultural economist, patterns of
food preparation for the nutritionist, or insights about
household networks for the anthropologist and community
organizers. The analysis of time use is a way of charting farmer
behavior in all of its capacities. The flexibility of time
allocation research permits investigators to pursue questions of
specific or general nature.
The conceptual framework presented in the first part of this
paper can be used to develop methods of time allocation research
that create the closest fit between the purpose of the research
and the data generated, as well as to evaluate whether the
methodology achieves this fit with a minimum of cost and bias.
The framework helps to structure our comparison of the four
methods used at Balinsasayao. The Balinsasayao project
illustrates that a range of methods can produce a corresponding
range of results. While there were strengths and weaknesses
associated with each approach to the study of time use, the
combination of all methods provided a comprehensive understanding
of the organization of time in the farm household. Methods do
not have to be used in isolation, complementary approaches should
be used together.
Jeff Romm and Louise Fortmann provided useful comments on an
earlier draft of this paper. Ramon Somido helped to collect the
data. Thanks are due to the eleven households that generously
gave us their time and allowed us to invade the privacy of their
daily lives. The author is also grateful for the assistance
given by the University Research Center, Silliman University in
Dumaguete City, Philippines. The research was funded by a
1980 Research in Women's Roles and DemBograhic Change:
Survey Questionnaires for Households. W2meDA Men and
Communities with Background EgeRlnation. Geneva:
International Labor Organisation.
Acharya, Meena and Lynn Bennett
1982 Women and the Subsistence Sector: Economic Partici-
Ealion and Household Decisionmaking in Neeal. World
Bank Staff Working Paper No. 526. The Word Bank.
The Asia Society
1978 Time Use Data: Policy Uses and Methods of Collection.
Report of an Asian Development Seminar sponsored by
the Asia Society, September 1978, New York.
Barlett, Peggy F.
1980 Cost-Benefit Analysis: A Test of Alternative Methodo-
logies. In P.F. Barlett, ed., Agricultural Decision
Making, pp. 137-160. San Franciso: Academic Press.
1980 Measuring Time Use and Non-Market Exchange. In P.W.
McGreevey, ed., Third World Poverty. Lexington: D.C.
Heath and Company, Lexington Books.
Burfisher, Mary E. and Nadine R. Horenstein
1985. Sex Roles in the Nigerian Tiv Farm Household. West
Hartford, Connecticut: Kumarian Press.
1985 Production Patterns. Household RDevelmental vl
Stages and Participation of Household Members: The
Case of the Lake Balinsasayao Lowland Migrant Uland
Farmer. Working Paper, University Research Center.
Dumaguete City, Philippines: Silliman University.
Chambers, Robert, Richard Longhurst and Arnold Pacey, eds.
1981 Seasonal Dimensions to Rural PoverEty London:
Frances Pinter Limited.
1980 The Statistical Behavior Approach. In P.F. Barlett,
ed., Agricultural Decision Making, pp. 87-114. San
Francisco: Academic Press.
Colfer, Carol J. Pierce
1985 egIEe Factors in Farming Systems: An AntBhroelogist's
View. Paper presented at the Workshop on FSR/E and
Agroecosystem Research, 12-15 August, Honolulu:
1955 Work Patterns in A Maya Village. American Anthrogo!21
gist 57 (2): 322-333.
Fortmann, Louise. 1985. Seasonal Dimensions of Rural Social
Organization. Journal of Develogment Studies
1984 gmgaEiDg ADgsbeeh ns and Franscthone Aegroaches to FSR
and Extension. Paper presented at the Fourth Annual
Conference on Farming System Research, Kansas State
University, Manhattan Kansas, October.
1980 Patterns of Household Labor Allocation in a Javanese
Village. In H.P. Binswanger, Evenson, Florencino,
and White, eds., Bural Households Studies in Asia,
pp. 18-217. Singapore: Singapore Press.
Hayami, Yujiro, Piedad F. Moya and Luisa M. Bambo
1978 Labor Utilization in a Laguna Rice Village. ~he
EhiiiERD!e Economic Journal 17 (1&2): 244-259.
International Labor Organization
1981 Women, Technology and the Development Process.
In R. Dauber and M.L. Cain, eds., Women and Techno-
1ogical Change in DBevelging CuntrIies, pp.33-47.
AAAS Selected Symposium No. 53. Boulder: Westview
1975 Time Allocation in a Machiguenga Community. gthno=
logy 14(3): 301-310.
Lee, Richard B.
1968 What Hunters do for a Living, or How to Make Out on
Scarce Resources. In R.B. Lee and I. DeVore, eds.,
Man the HunterD pp. 30-48. Chicago: Aldine.
1951 Life in a Mexican Village: Teoz2tlan Restudied.
Urbana: University of Illinois.
1959 Five Families. New York: Basic Books Inc.
1984 The Role of Case Studies in Farming Systems Research.
.- 5Ds Di5cOSlon Papar Alo, 19- .
1985 Labor Allocation of Sundanese Peasants, West Java. In
S. Suzuki, 0. Soemarwoto and T. Igarishi, eds.,
Human Ecological Survey in Rural West Java in 1978 to
1982. Tokyo: Nissan Science Foundation.
1979 Time Use in Rural Botswana. Paper presented at a
Seminar on the Rural Income Distribution Survey
in Gabarone, Botswana, 26-28 June.
Navera, Emeline Relubit
1978 The Allocation of Household Time Associated with Chil-
dren in Rural Households in Laguna, Philippines. The
hEilinrine Economic Journal 17 (1&E): 203-223.
Norman, D.W., Newman, M.D. and Oedrago, I
1981 Farm and Village Production Systems !i the Semi-Arid
Trogics of West Africa: An InterEretive Review of
Research. Research Bulletin 1(4) Pantcheru, A. P.
Price, E.C. and R.X. Barker
1978 The Time Distribution of Crop Labor in Rice-Based
Cropping Patterns. The EhilieEine Economic Journal 7
1978 Time Allocation and Home Production in Rural Philippine
Households. The Phili=iine Economic Journal 17(1&2):
Rosaldo, M.Z. and Louise Lamphere, eds.
1974 Women EgClture and Soiety. Stanford: Stanford Univer-
1972 Stone Age Egonomics. Chicago: Chicago Press.
Smith, Joyatee and Fe Gascon
1979 The Effect of the New Rice Technology en Eamily Lsabr
Utilization _n Laqgna. IRRI Research Paper Series No.
42. Los Banos: IRRI.
Stone, Linda and J. Gabriel Campbell
1984 The Use and Misuse of Surveys in International
Development: An Experiment from Nepal. Human Organi=
1984 Tanzania: The Arusha Planning and Village Develop-
ment Project. In C. Overholt, M.B. Anderson, K.
McCloud and J.E. Austin, eds., Gender Ro1es in
Development EP21ests: A Case Book, pp. 163-184.
West Hartford, Connecticut: Kumarian Press.
Site and Project Background
Lake Balinsasayao and the surrounding community is at an
elevation of 830 meters; it is nestled among the volcanic
outcrops of the mountain range that runs through the middle of
the island. The elevation of the surrounding peaks ranges from
760 to 1790 meters. Annual rainfall, recorded for the period
June 1982 to June 1985, was between 2000 and 3000 mm. Most of
the precipitation occurs during the eight months between June and
January. Typhoons typically strike between September and
December. Temperatures range from an average of 18 C at night to
29 C during the day. Low clouds or fog is common throughout the
Soils are derived from volcanic parent material. Soil texture
and nutrient status is very site specific. Samples from over 55
fields showed that the soils are generally acidic (pH range of
4.7 to 6.1), and high in organic matter (between 2.9 and 22.7
percent). Soil texture ranges from clay to sandy loam. Soil
fertility was noticeably higher in fields that had been newly
cleared from the forest, had not been cultivated intensively or
had been abandoned for more than a year.
Land clearing and preparation occurs during the dry season,
February to May. Most of the planting occurs between February
and July; harvesting takes place between May and December.
Production is subsistence rather than market oriented. All of
the households rely primarily on family labor. Few sources of
wage labor are available. No commercial pesticides or
fertilizers are used.
Silliman University began a development project in the area in
1981 with the objective of improving the sustainable productivity
of the Balinsasayao farms. The Ford Foundation sponsored the
project. The project has worked with as many as 40 households
and as few as 7 on any of its specific activities. Efforts have
included dispersal of trees for agroforestry, introduction of
erosion control measures, organization of an "Upland Farmers'
Association" and "Mothers' Group," and cross-visits to other
communities to improve the network of communication about
innovative farming practices. On-farm trials in soil
conservation have involved the construction of bench terraces,
organic residue fences, rock walls, and contour canals. Leuceana
diversifolia and napier grass have been tested as living crop
fences for erosion control. In a one year period, 13 households
or 60 percent of the core community had adopted some form of
Sample Population Characteristics
Males and Females in Comprehensive Case Studies
VALUE OE SYMBOL EQUAL..S APPROXIMATELY .40 OCCURRENCES
1. O0 *Swxmz******X******q
.2......... *.........I.........2... ...***** **
0 4 8 12 lo 20
26 MHSSNG CASES 0
Males and Females in Survey
VALUE ONe SYMBOL O.ALS APPRDXIMATEL.Y .90 OCCURRENCES
1.00 1 Ws **Ju2g2
2.00 S*X m E**I*3
*** ....... *. *.. .. .***** .***** *** S**********
0 8 16 24 32 40
56 MISSING CASES 0
1. Livestock: feeding, pasturing, collection of roots for
pigs, cooking of roots for pigs, general care
E. Land preparation: plowing, hoeing, burning, cutting vegetation
(i.e. kai gi), visiting plots, application of pesticides
or fertilizers, related travel
3. Weeding: weeding, related travel
4. Harvesting/planting: harvesting, planting, transplanting,
5. Fishing: setting fish traps, collecting fish from nets,
fishing from boat or shore, gathering bait, maintaining
fish pond, related travel
6. Gathering and forest uses collection of fuelwood, vines,
hunting, related travel
7. Processing/manufacture: sewing, making tools, making baskets,
mats, processing tobacco, handstripping abaca, roasting or
grinding food products
8. Income: off- and nonfarm labor (includes road construction,
sawing timber), exchange labor, abaca sales, related travel
9. Market: local exchange, buying and selling in coastal
markets, related travel time
10. Domestic chores sweeping, washing dishes, child care,
cooking, preparing food for human consumption, butchering,
fetching water, chopping wood, building fire, watching
house, laundry, unpacking baskets, searching for items,
lighting kerosene lamps, construction or home repairs
11. Daily needs: eating, sleeping, education, dressing,
haircutting, personal hygiene, sickness
12. Leisure: relaxing, socializing, entertaining
13. Meeting: transporting other people across lake, meeting
visitors, community meetings, guiding visitors, interview
periods, related travel, visiting within and outside the