ABSTRACT
The method of randomization dispenses with the nagging problems of normality, and probability sampling. Tests of significance are seen as low-cost approximations of ths model, which provide greater precision at the cost of more assumptions. The inability to use true experimentation results in a lack of control over a wide variety of largely unknown and unenumerable potential influences. The empirical sociologist is confronted with the necessity to identify and to find justification for rejecting as many of these potential influences as he can. As the authors considers the logical problem of scrutinizing the validity of the relationship between the researcher’s variable of classification and his quantitative results, it is useful to distinguish between internal and external validity. Internal validity addresses the question as to whether or not the experimental variable made a difference in the specific instance under consideration. External validity asks about the range of generalizability.
