A starting point that I use in surveys in urban settings when there are multiple occupier dwellings (e.g. blocks of flats / apartments and hotels) is to treat each access doorway ("front door") as a dwelling (i.e. each lock or hotel is many dwellings). In apartment buildings there may be many rooms behind each front door. In hotels there is usually a single room (excluding latrines / showers) behind each front door. Selection of households is then a matter of selecting front doors. I tend to do this with systematic sampling. There are complications. Sometimes a household may live in more than one room behind more than one front door. Sometimes more than one household lives in one apartment. You need to have rules for dealing with this. In the case of more than one front door for a household you might select the household (all rooms). In the case of more than one household behind a front door you might select a household at random. A lot will depend on how you define household. Your second issue is about eligibility. I would have a "filter question" that I ask to distinguish between "temporary guests" and IDPs. If you find a "temporary guest" then take the next nearest door (you may need to toss a coin) and apply the filter question again. I hope this is of some use.

Hi To follow up on Mark's point about household definition, this is a key point. If you use, for example, an economic definition of the household then you are defining the unit that shares the same food and cash income. In many cases this unit would be living behind 1 access door but the definition might help you in the situations where the unit is spread over more than 1 room in the hotel. As for the temporary guests, if your defined target population is 'people who are displaced' then they will not meet the necessary criteria. Best wishes Sonya