Mitch's Memorandum #6 on Ercikan

Memorandum on…	Submitted by: Mitch Bleier
Ercikan, K. Limitations to sample-to-population generaliztations. In K. Ercikan & W.-M. Roth (Eds.), Generalization in educational research. Mahwah, NJ: Taylor & Francis.	U ED 70300 Prof. Tobin Spring 2008 March 19, 2008

Kadriye Ercikan addresses using research done on ostensibly representative sample populations to make claims about the larger population from which they were taken.

S/he presents three arguments for making generalizations and indicates the type of research associated with each (Table 1), then provides and describes examples of instances where sample-to-population generalization is inappropriate or misleading and explains how various, apparently justifiable generalizations may be erroneous. In each case, Ercikan discusses what researchers must consider in developing a research design, and interpreting and applying the results of their research to educational practice or policy.

Arguments for generalization	Associated research type
Sample-to-population extrapolation	quantitative methods using representative random samples and statistical analysis
analytic generalizations	experimental and quasi-experimental methods; claims are made in relation to a theory.
Case-to-case translation	qualitative methods employed; explicit arguments made to justify the application of findings to different groups.

Table 1: Generalizations and Associated Types of Research

Ercikan explains that

[s]ample-to-population generalizability is an indicator of the degree to which total sample findings generalize to the total population. This generalizability does not address whether the findings are meaningful for individuals or groups of individuals (p. 12).

and points out that assumptions about randomness and representativeness of samples, the effect of sample size on generalizability of findings, and the ability to identify and control the myriad factors that may render sub-groups incomparable on the particular areas addressed in a study, must be taken into account in research design and in interpretation and application of findings. Although this paper makes a strong case for alternatives to sample-to-population generalization, Ercikan concludes by addressing similar limitations for analytic and case-to-case generalizations.