A conceptualization of generalization and its place in educational research is proposed. Problems with existing approaches to generalization include limited concepts, misleading or simplistic results, perception variance or change, and the discouragement of sensitivity due to convenience. Researchers should distinguish between conclusions which are apparently specific to a study, and broader generalizations. Research should be conducted in natural settings, using methodology which interferes as little as possible with the characteristics of the setting. Researchers need to be meticulous in describing their sample, setting, treatments and instrumentation. To gain an adequate level of understanding, research efforts need to be more intensive and deeper than is customary. The choice of research focus (research questions, and hypotheses) is as important as methodology in carrying out the investigations. Extensive prior investigation is needed to identify key variables in the setting in which the research is to be carried out. Hypothesis testing should be de-emphasized, and the focus of research should be shifted to description, exploratory data analysis, estimation of the magnitude of effects, and replication of interesting findings. (Author/PN)