The justification for these surveys and their suitability lies in the fact that by definition a household is a unit consisting of either an individual living alone or a group of two or more persons living together with a common provision for food and other essentials necessary for living. Whether they are one-person or multi-person households of related and/or unrelated individuals, such households serve as the ultimate sampling units which best represent any specific population under consideration or study. The only persons who are not represented through household-based sample inquiries are the homeless, nomads, and household members who are absent from the household permanently or for a long period at the time of enumeration. If such persons do not live in another household within the country, they will not be represented in a household-based national sample survey. However, such groups normally only constitute a very small proportion of the total population within any specific age cohort. Even then, much information could be collected on those who are away from the households (such as street children) by addressing various relevant questions to the household head or a proxy. This information could, in turn, be used to design a more appropriate investigation of such persons to find out more details on all aspects of their activities, occupations, living conditions and so on.
The use of households as units of enumeration could permit the gathering of a wealth of statistical information on all or any segment of a country's population, subject to the availability of resources. The ILO methodological experiments carried out in 1992-93, and national surveys undertaken since then concerning child labour in several countries, have proved the household approach to be the most effective means for a profound assessment of the level, nature and determinants of the practice at the national level. During such surveys, information could also be collected on the activity patterns of adults not only because the additional cost involved would be marginal, but because such data are important for studying the interrelationship between the activities of children and the activities of other members of the same household and, in particular, those of their parents or guardians.
Besides providing a national picture, another advantage of a comprehensive household-based survey is that, if implemented through well-designed sampling and stratification procedures, it would permit segregation of the statistical information not only into rural/urban areas and informal/formal sectors, but also, and more importantly, into small geographical areas within any large geographical region or province. The information on small localities would be crucial for formulating and implementing policies and action programmes appropriate for combating child labour in specific geographical areas or communities where the problem may be quite serious.
It should be noted that the household-based survey of child labour could be carried out either as a "free-standing" or "stand-alone" inquiry, or as a module attached to other ongoing household-based surveys. The latter approach is much more efficient in many ways, particularly if the module is implemented as a supplement to an established programme of a labour force survey (LFS) conducted on a sample basis at the national level. This undertaking will not only result in substantial cost savings, but the operation could be achieved in less time. Operationally, it means that the module would be piggybacked onto one of the rounds of the LFS and that the interviews for both the LFS and the module would be carried out at the same time. Since the LFS questionnaire always seeks to enumerate the demographic and socio-economic composition of household members, there would be no need to repeat this part in the module for children. Through empirical studies it has been demonstrated that the incidence of child labour and the demographic and socio-economic characteristics of the "adult" household members are correlated positively or negatively depending on the different variables considered. Therefore, information on other household members has to be collected as well.
The attractiveness of the modular option stems from the fact that there would be no need to list all the households selected in the initial stage of sampling, representing the primary-stage sampling units (PSUs), or to collect the basic information required for the stratification and selections of the second stage sampling units (SSUs). Also, since there would be no repeat questions on the demographic and socio-economic characteristics of the children, the entire module would be considerably shorter. In addition, where the ILO and national statistics offices have collaborated closely in attaching a comprehensive child labour module to ongoing household-based surveys, significant savings have been realized. For example, in Turkey, where the module was attached to one of the two rounds of the national LFS, the cost of the operation amounted to only one-fifth (20 per cent) of the total resources that would have been needed if a stand-alone child labour survey had been conducted. A similar approach was carried out in Cambodia where the cost of the child labour survey component was about one-tenth (10 per cent) of the estimated total resources.
A sample questionnaire which could be used for a child labour modular inquiry attached to a household-based labour force survey can be provided on request by the ILO's Bureau of Statistics.
Coverage and classification of child labour
To avoid limiting the coverage of the incidence of various forms of child work, all types of activities (schooling and non-schooling, economic and non-economic activities) of children within a specified age group would be represented through sampling and enumerated, and the volume or workload of their activities quantified, so that the assembled statistical information could be cross-tabulated by the different characteristics of the variables included in the questionnaire. Depending on the level and nature of the quantified activities or variables, those which are judged or expected to have negative effects or consequences on the health, education and normal development of the working children could be considered as falling within the boundary of "child labour". The data should then be further dissected into various categories of affected children based on the degree of harm caused by the quantified activities.
Since the comprehensive household survey would investigate all the activities of children in the particular age cohort, the data collected would make it possible to identify the specific occupations of the working children, their working conditions, accidents/injuries/illnesses suffered - including their frequency and gravity - problems related to the workplace environment, particulars of employers using children, the specific industries in which children work, the effect of their work on their normal life (including their schooling), and other related matters which would assist in assessing more fully the extent, nature and causes of child labour. Such detailed information would also become instrumental for focused study of a particular category of working children, or occupation, industry and the like, and for formulating and implementing policies and programmes for the immediate elimination of the most harmful of children's activities, as well as for the complete eradication of child labour in the long term.
Given that some international conventions, declarations and resolutions seek to protect all children under 18 years of age from different types of harmful activities, it is strongly recommended that the survey of child labour should cover children from 5 to 17 years of age. One advantage of using this broad age cohort is that it would allow breakdown of the data obtained by different age groups, thereby satisfying the requirements of the various instruments, as well as national needs for respecting compulsory schooling regulations, labour codes and other legal requirements of individual countries. Examples of groupings are: 5-14, 15-17 and 5-17 years, the 5-14 age group broken down further by 5-9 and 10-14 years, all commonly used by most countries. It may be useful to have the results also separately classified for children under 11, 12-13, and 14 years, in view of ILO Convention No. 138 and Recommendation No. 146 concerning Minimum Age for Admission to Employment (both adopted in 1973). Convention No. 138 generally prohibits economic activity of children under the age of 15, while making exceptions for those aged 14 years and for those between 12 and 13 years, depending on different circumstances in individual countries.
It is also recommended that work of a domestic nature (household chores) performed by children in their own parents' or other relatives' homes where they actually reside should be included in the investigation of children's schooling and non-schooling activities. This is to measure the time spent on this type of work to identify those children who are working more than the daily number of hours that may be considered as normal to learn common household chores and related activities. The final data compiled on these children should then be tabulated separately from those relating to children who are economically active (as defined in accordance with international standards). Non-economic work of a domestic nature in the parents' or guardians' household would then be classified and tabulated into various ranges according to the number of hours during which such work was performed. A threshold could then be established beyond which the activity could be deemed as constituting child labour.
The above is based on the argument that many non-school going children perform housekeeping activities in their parents' or guardians' households for various reasons, one being to make adult household members available for economic activity elsewhere. For many of these children this is a full-time occupation involving preparing and serving meals, washing clothes, cleaning floors, taking care of younger siblings, serving as messengers in and around the household, and so on, all this at the sacrifice of the education and playtime to which each child is entitled under the United Nations Convention on the Rights of the Child. Even those who attend school are found to be spending several hours a day performing such activities which are detrimental to their schooling, health and normal development to adulthood. Such children suffer fatigue which affects their school performance and many are exposed to hazardous situations, for example cooking food over an open fire. Children who are put under the guardianship of relatives or other persons are especially susceptible to much abuse in these areas of work. Behind the guardianship status there are often other arrangements which amount to child labour, including bondage which is among the worst forms of the practice.
The sampling approach to be adopted for surveys on children's activities should in principle be a multi-stage stratified sampling design to capture a good number of working children. In areas where the child labour incidence is suspected to be rare, or where children's activities are highly diversified, an over-sampling is recommended. Conversely, where there is a homogeneity of activities (in rural areas, for example), under-sampling (a smaller sample) could be considered.
Where statistical software packages are not readily available, a self-weighting systematic sampling design with probability proportional to size (PPS) of the population being considered should be adopted, since this will have a uniform weight for estimating totals and will also facilitate the computation of percentages, means and ratios of the population parameters directly from the sample data.
The questionnaire should be designed in two parts.
Part I should be directed to the head of the household or the adult responsible for the household and should be concerned with characteristics of each household member, housing particulars, household living standards, and schooling and non-schooling activities of children under 18 years, as follows:
Part II is related to detailed information on children under 18 years as provided by the children themselves for an in-depth study of their activity patterns and the impact on them. Many of the questions addressed to the household head or the parents/guardians are repeated to the children both for comparison and validation. Also, the children can be more specific than their parents on matters such as working conditions, injuries and illnesses, workplace hardships, and relations with the employer. The information to be obtained from the children should include the following:
Pilot test of survey instruments
Once the content of the questionnaire is agreed upon, the guide for enumerators and supervisors should be prepared. The questionnaire and the guide need to be tested and finalized on the basis of a pilot survey conducted with a small sample of households.
Sampling and stratification
The sampling and stratification procedures to be adopted depend on the availability of basic information on various factors and at different levels. Up-to-date information or demarcations regarding the general characteristics of the areas to be covered, the composition of the population and/or the households within each area, and so on, would be required to serve as a master frame in designing: several phases of sampling and stratification. In the experimental surveys, the various elements or classifications considered for the stratification included: development levels of the selected rural and urban areas (e.g. poorly developed/well developed, slum/non-slum blocks); income classes (low, middle, high); overall rates of literacy/illiteracy of the population; school attendance levels; family size, and so on.
Again, depending on the sample size and the other factors, an over-sampling may be needed to capture adequately the incidence of child labour, which is more rare than finding an adult in the labour force.
[Ukrainian] [English] [Russian]