In any case however using territorial markers tagging is important (and to certain extent – the only reliable) approach that can provide acceptably relevant estimate of the absolute number of the population in question (and not just shares as poverty rates and unemployment rates). The absolute number is crucial for needs assessment and hence for defining numeric targets. If targets (and resources) are determined on the basis of census data, the real needs will be inevitably underestimated. 3. Ethnic minority boosters in sample surveys Household budget surveys (HBS) and labour force surveys (LFS) are the most important surveys when looking at issues of poverty, unemployment and social exclusion and the respective policies to address those issues. Unfortunately, these surveys in most countries fail to include a representative sample of ethnic minorities (especially when a minority is small or lives segregated from the majority) or when it is solely based on census data. To overcome this sampling problem and use HBS/LFS as a regular and precise data collection mechanism for minorities, sampling boosters of the respective minorities (i.e. increasing the sample size of minorities) or separate minority samples would be necessary. However, this is very costly and impossible when several minorities exist in one country. Constructing the random sample boosters may be a problem, mostly because of the unclear number of the ethnic population. One possible compromise is accepting the self-identification principle (during the census) and constructing a random sample based on the population selfidentified or having declared a respective mother tongue (ideally both). In this case a minority booster would bear the “genetic” features (and problems) of the PIN-based methods for statistical data disaggregation and shares both its benefits and detriments. An alternative could be constructing a sample on the basis of territorial mapping of the ethnic population – assuming that such mapping is in place. Similar to the latter is using GIS (Geographic Information System)based sampling, which to large extent is a variety of territorial tagging. 4. Custom surveys among social services recipients This approach entails anonymous questionnaires (usually brief, consisting of just few questions) filled in by recipients of social services on voluntary basis. For example, unemployed person registering at the labour office is invited to fill in a questionnaire in addition to the regular forms. The questionnaire may include the field “ethnicity” and is dropped in a sealed box to make linking of the questionnaire with the standard application impossible. Such approach can be a good source of information, both for the ethnic profile of the recipients of social services and for the way in which their providers work (for example, are there any ethnic-based prejudices?). In the best case scenario (assuming there is no duplication of questionnaires and their number is close to that of the recipients of social services) such survey could be representative just for the recipients, not for the whole ethnic group. 5. Community-based monitoring Community-level data is particularly important with regard to monitoring social exclusion and poverty. Such a system could provide basic information on the communities in question based on standard questionnaires completed on regular basis by a designated member of the community after receiving training on basic data collection and reporting techniques. The system would provide:  Quantitative information on the community status (number of households, their housing conditions, number of children attending school, their age and grade, number of drop-outs, number of new-born, number of vaccinated children etc.). Chapter 9: Data Collection Tools 133