This paper examines how potential conflicts produced by ethical issues such as confidence and confidentiality can affect the research methodology of educational research and evaluation projects. Despite the differences in site selection, clientele, and relationship to treatment, both evaluators and researchers must make adjustments in method to accommodate ethical concerns. Based on a research study of the political socialization process of Colombian children and several educational program evaluation projects in New England, participant observation systems are considered for methodological modification for ethical reasons. In both the research and evaluation studies, the investigators felt that the risk of the Hawthorne effect or other biasing effects must be subordinated to the observed subjects' right to know the nature of the observations. In addition, honoring assurances of confidentiality to one group of subjects enhances the confidence that other groups will have in the investigators' promises. Examples of ethical conflicts that arise uniquely in muliclientele program evlauations are also described. (Author/DE)