Generalizability

Generalizability is the degree to which the results of a research study reflect what the results would be “in the real world,” with another sample of participants or with the variables operationalized in other ways. In other words, research results are generalizable when the findings are true generally speaking in most contexts with most people most of the time.

In the classic quantitative research framework of experimental design, researcher design theorists such as Thomas Cook and Donald Campbell have emphasized external validity as a necessary criterion for concluding that research results are generalizable. Threats to external validity include how a sample was selected from the broader target population to which one wishes to generalize, the situational specifics of the experimental manipulations, and the measurement choices made when assessing the independent and dependent variables. Generalizability is optimized when samples are chosen randomly, the research environment and researcher behaviors are carefully controlled so as not to affect the outcome, and [Page 725]constructs are defined and measured in ways that validly and reliably represent the broad ways that variables operate.

In the qualitative research framework, there is a somewhat different understanding of generalizability. Although some qualitative researchers argue that it is inappropriate to assume that generalizability is even an appropriate goal of social science research, there are some generally accepted generalizability criteria if one wishes to understand research results in a wider context. However, qualitative researchers are often more interested in vertical generalization, the extent to which research findings add to building or understanding theory, than they are interested in horizontal generalization, the more traditionally quantitative wish to conclude that there would be similar results with another sample drawn from the same population. The qualitative framework known as grounded theory, for example, is more focused on whether theory that has been induced from the data collected is a fair representation of the data than whether a sample of participants is a fair representation of some abstract population.

See also Grounded Theory; Qualitative Research Methods; Quantitative Research Methods; Random Selection; Threats to Research Validity