Hi Anonymous 1727 This classification of wealth is made using HEA (Household Economy Approach) method. Using participatory rural appraisal (PRA) techniques village leaders (key informants) identify different wealth groups in their community and defined the criteria for inclusion in each group. you can look at the the guides from the following links http://www.dppc.gov.et/, or http://www.dppc.gov.et/Livelihoods/livelihoodhome.html, if you are not successful, you can contact people from ENC or nutrition cluster.

(1) I think that a semi-quantitative approach might be your best bet. In this approach you could (e.g.) have some sort of trigger criteria in the survey questionnaire which would trigger the use of a semi-structured interview. You could also back this up with a SQUEAC type method casting IYCN as a coverage issue (e.g. a four month child not exclusively breastfed can be consider not to be covered by good IYCN practices). This would give you a wider and deeper dataset than a simple KAP type questionnaire (which can be difficult to get right anyway). (2) There are a number of approaches. HEA (and similar approaches) are semi quantitative and need some skill to do well. Wealth ranking methods in which you use sets of random pairs of households and ask respondents to say which is wealthier than which can work well. Data-analysis is somewhat specialised but tools (e.g. scripts / macros) are available. This method can be problematic in populations shy of "grading" their neighbours and will probably prove difficult in urban or camp settings. A fully quantitative approach is used in MICS (and well-described in the MICS literature) is to collect data about ownership of a wide-range of chattels and other indicators (e.g. housing floor and roof type, persons / room, &c.) and to use a data-reduction / dimension-reduction technique such as factor analysis (FA) or principal components analysis (PCA) to devise a scoring method. You apply the derived scoring method to the data and then use wealth-categories based on (e.g.) quintiles of the score for each HH. There is a considerable literature relating to FA/PCA to this problem which is broadly supportive of the methods. You'd need to do some work on questionnaires (i.e. to get suitable lists of chattels) before data collections. This is covered in the MICS literature. Analysis is relatively straightforward and can be done in SPSS, STATA, R, SAS and other systems. You may want to obtain the advice of a statistician for this if you are wholly unfamiliar with multivariate techniques. The MICS literature and web resources contain example analyses. I hope this helps.