The goal of this research is to provide a viable model of people's evaluation of housing to explore ways of improving urban housing. To achieve this goal, two requirements were set up. First, individual differences in the housing evaluation should be respected. Second, the model should have reliability, i.e. it should be of statistically condensed outcome. Based on the personal construct theory, a two-step procedure which consists of the elicitation of the dimensions in the housing evaluation on the individual level, and the fixed format questionnaire which is statistically analysed along phenoinenologically obtained subgroups is proposed.