observer,” involved with the experience and action of those observed (Blumer, 1966; Denzin, 1978; Kelly and Lesh, 2000). In such “naturalistic research paradigms” (Moschkovich and Brenner, 2000), investigators do not seek to distance themselves from research participants, but rather to immerse themselves in the participants’ lives, with conscious attention to how the investigator affects and contributes to the research process. Such strategies were developed to allow the researcher to observe, analyze, and integrate into the research process unexpected, constantly changing, and other potentially influential aspects of what is being studied. These approaches are often particularly important in studying how changes in school subject matter or the development of new technologies can be incorporated into educational practice. In collecting and coding such qualitative data, convergence can be demonstrated with repeated instances, more than one observer, and multiple raters. Also essential to the process is the examination of competing interpretations, contrasting cases, and disconfirming evidence. Regularity in the patterns across groups and across time—rather than replication per se—is a source of generalization. The goal of such scientific methods, of course, remains the same: to identify generalized patterns.
Uses of theory also tend to distinguish work in the social and physical sciences. Theory in the physical sciences leads to predictions about things that will happen in the future. Strong theories include causal mechanisms that predict what will happen and give insights into why. Theory in the social sciences is predictive, but more often it serves to understand things that happened in the past, serving a more diagnostic or explanatory purpose. Understanding the past often enables social science researchers to explain why things happened. Though understanding the past can sometimes predict the future, it does so only in broad outline and with a lesser degree of certainty. For instance, researchers have documented the regularity of certain misconceptions and patterns of error as students learn scientific or mathematical ideas. Although one cannot predict exactly when they will occur, awareness of them permits teachers to interpret student comments more effectively and to create assessment items to test for evidence of them.
A related and final point is that the level of certainty with which research conclusions can be made is typically higher in the physical sciences than in