This paper explains the principles and procedures of second-order theoretical analysis of observation data. The second-order theoretical procedure is illustrated by analysis of the student-teachers seminar on the subject of the Holocaust, focusing on dilemmas of Holocaust survivors. Analysing observational data allows two distinctive method of analysis: first-order analysis and second-order analysis. First-order analysis of observation is a process based on the direct descriptions and explanations of the informants who participated in the observed event. Second-order theoretical analysis of observation is used when we cannot construct a full description and explanation of the observed event directly from the informants. While the data gathered from participants indeed focus on the informants, they are not accompanied by their direct descriptions and explanations. Second-order theoretical analysis is used to analyse the data we collected on and from our informants which contain valuable indicators suggesting a possible theoretical explanation with the potential to express the informants' worlds. The paper suggests procedures for verifying the trustworthiness of descriptions, explanations and theory derived from observation.