Substantial efforts have been made over the past decade by both FAO's Statistics Division and the Women and Population Division to include gender issues in agricultural censuses. However, this source does not normally provide data on household labour force participation and the role of household members in the agricultural holding. Therefore, it is worth while exploring alternative opportunities for further development of relevant data systems that are both cost-efficient and organizationally sustainable. Four possible strategies of action are considered:
1) In addition to the data provided by agricultural surveys, available subnational-level data from population censuses and labour surveys could be retabulated. For example, the categories of seasonal and part-time work that cannot be included in the agricultural census are most likely to capture sex differentials and emerging trends by area of agricultural activity, i.e. women's supposed overemployment in areas of low productivity, compared with the category "permanent workers". The same categories could also be beneficial in uncovering patterns of (supposed male) time-related underemployment. According to the latest ILO standards, part-time activities can be used as a measure of underemployment when they refer to "involuntarily working less" (ILO, 1999). In order to carry out the recommended improvements in the use of available data, FAO is ready (subject to availability of funds) to initiate interagency collaboration with ILO and other UN partners, for the development of a specialized database on gender statistics relating to agricultural and rural development.
2) Another strategy of action, accepted by the 16th Session of the African Commission on Agricultural Statistics (AFCAS) in 1999, is to develop common coding systems and data merging between population and agricultural censuses in countries where the unit of enumeration in the agricultural census is the household. This can be done through the use of interfile linkage techniques, recognized as a way of maximizing the relations between available datasets for multiple purposes, while minimizing the general cost of data collection. Such a strategy is economically feasible. It also provides a wider and deeper analytical perspective, since it enables the use of statistical modelling for the analysis of complex interrelations between variables of a different nature (e.g. demographic, social and environmental forces of diversification in rural labour markets).
3) The gender efficacy of agricultural censuses can be increased at almost no additional cost, enabling up to two agricultural operators per household to be enumerated. This allows for a more accurate documentation of women's participation in the management of agricultural production units. Even in countries where the holding (and not the household) is the unit of enumeration, common coding strategies allowing for additional cross-tabulations are possible because many agricultural censuses are sample-based and their sampling frames are mainly drawn from population and household censuses. The 16th Session of AFCAS reported that several countries, such as Côte d'Ivoire, are already using such strategies to achieve cost-efficient and improved availability of more structured and comprehensive data systems. AFCAS recommended that FAO include these undertakings in its pilot projects in order to study them as test cases.
4) Diagnosis and evaluation, which are used increasingly in surveys, could be used more often and for a variety of purposes. For example, evaluation studies and quasi-experimental comparisons for the evaluation of gender-specific policies in rural development are necessary but not widely used. Yet these methods are essential in areas such as the evaluation of productivity and technical efficiency by sex in various socio-cultural, family and production contexts (World Bank, 1999c).
In most cases, qualitative indicators as well as quantitative measures are needed in order to achieve meaningful and valid analytical models. For example, the stage reached in their life-cycle may be a more important source of variance in women's access to resources controlled at the household or extended family level (such as land and cattle) than men's. But since the significance of this life-cycle stage itself may vary among different cultures, ethnic and socio-economic groups or birth cohort, age becomes a necessary quantitative measure to pair with qualitative indicators so as to obtain more robust and comparable patterns of resource allocation at the household or family level.
Gender relations and inequalities are strongly affected by two main factors: legislative change and technological innovation. Emerging new forms of inequality, especially in agriculture, cannot be identified without a strong emphasis on qualitative indicators and improved methods of collecting and combining analytically qualitative and quantitative information for exploratory, evaluation and validation purposes. In order to ascertain accurately the real situation of rural women and their labour contribution, development agents should draw qualitative information from sources such as socio-anthropological studies, values and attitude surveys, market analysis and feasibility studies as well as special studies on key issues such as land and credit access, institutions and participation in rural organizations.
Gender statistics are frequently not produced because of a lack of interest and demand from potential users. In order to formulate statistical needs accurately, policy-makers, public and private development practitioners, NGOs and other involved users of gender-related data need to consider the following points:
There should be a permanent dialogue between users and producers of statistics to ensure the relevance and validity of data. The synergy of knowledge and expertise drawn from various categories of users (such as planners, decision-makers and gender specialists) and statisticians will help identify the fundamental aspects of gender issues (causes, consequences, interdependencies) and facilitate monitoring and evaluation of these issues over time. This dialogue is particularly important in the determination of the categories of data to be collected and the selection of indicators.
Relevant data collection will also depend on the ability of statisticians to extend their horizons to social and gender issues. It is also important for users (such as gender experts and decision-makers) to upgrade their skills in order to understand, interpret and use statistical data correctly. Collaboration can be furthered through workshops, technical committees, expert consultations and the integration of statisticians in policy-making processes.
The performance of governments in creating an enabling environment for sustainable agriculture and rural development will depend more on the availability and quality of statistics regarding rural producers. Data disaggregated by sex and age will be necessary not only to address issues of gender equality but also, and increasingly, to address issues of labour force dynamics and development, productivity and poverty.
Since governments are entirely responsible for their own data collection, decisions on the inclusion of a gender dimension in their statistical operations are in their hands. They need to review existing sources to identify the appropriate measures to be taken in determining the policy and technical framework for the compilation, analysis, presentation and diffusion of data. They must also strengthen the user-producer interface to create the synergy that is indispensable for the production of good-quality and reliable statistics on gender in agriculture.