I presume you refer to data collected in stage 1 (identification of areas of low and high coverage as well as reasons for coverage failure) and data from stage 2 (confirmation of the location of areas of high and low coverage and the reasons for coverage failure identified in Stage 1). These data are used to build the Prior and thus it contributes to the coverage estimation using the beta-binomial conjugate analysis based on Bayesian method. To answer your question about having “a guidelines on numbers of key interviews to be done or FGD to be done”, the “guideline” (FANTA-SQUEAC technical reference) clearly stated how to use qualitative data as well as the principles of the SQUEAC: triangulation and redundancy. You need to collect data until there is no new information and information become redundant (many sources report the same element). How many? Well............ if your barrier/booster is very important, it will come out quite quick and clearly. Not sure what you mean by “analysis are not done statistically” (which sounds as having a negative connotation). The way to calculate the mode of the prior is “to use scores or weights that reflect the relative importance or likely effect on coverage of each finding” and this based on the above triangulation and redundancy. In other words, this procedure allows to translate a qualitative information (e.g. regular shortage of RUTF) into a weight affecting coverage (information triangulated using several key informants and different methods and redundant – everybody pointed out this barrier). If not done yet, I suggest you download the FANTA SQUEAC technical reference http://www.fantaproject.org/ where analysis of qualitative and quantitative data is largely discussed. I hope this answer your question.

Short answers ... Qualitative data do contribute to the final coverage estimate. They are used to inform the prior in a Bayesian analysis. How this is done is extensively documented. There are clear and thorough guidelines WRT the spread of data sources, collection methods, and the number of interviews to be done. The guidelines are the classical qualitative approach of triangulation by source and method and sampling to redundancy. Statistical data analysis is covered in SQUEAC documentation. There is specialist software for sample size calculations and data analysis. The material is too large to be carried by a forum post. I think that [url=http://en.wikipedia.org/wiki/RTFM]"RTFM"[/url]) applies here. I suggest you download a copy of the [url=http://www.fantaproject.org/publications/squeac-sleac.shtml]SQUEAC Technical Reference[/url]. This covers all of your questions in some depth. A French version of this reference will be available in a month or so.