[Pre-Assessment]

by

Pre-Assessment

Reliability and Validity

As per my understanding of biasness in research, the term bias is a historically unfriendly pejorative frequently directed at action research. As much as possible, the absence of bias constitutes conditions in which reliability and validity can increase. Most vulnerable to charges of bias are action research inquiries with a low saturation point (i.e., a small N), limited interrater reliability, and unclear data triangulation. Positivist studies make attempts to control external variables that may bias data; interpretivist studies contend that it is erroneous to assume that it is possible to do any research—particularly human science research— that is uncontaminated by personal and political sympathies and that bias can occur in the laboratory as well as in the classroom (Welch 1996). I feel that while value-free inquiry may not exist in any research, the critical issue may not be one of credibility but, rather, one of recognizing divergent ways of answering questions associated with purpose and intent. Action research can meet determinants of reliability and validity if primary contextual variables remain consistent and if researchers are as disciplined as possible in gathering, analyzing, and interpreting the evidence of their study; in using triangulation strategies; and in the purposeful use of participation validation. Ultimately, action researchers must reflect rigorously and consistently on the places and ways that values insert themselves into studies and on how researcher tensions and contradictions can be consistently and systematically examined. (Olt 2007)

Generalizability

In my personal understanding generalizability in any research is any claim of replication possible in studies involving human researchers and participants? Perhaps even more relevant to the premises and intentions that underlie action research is the question, Is this desirable in contributing to our understanding of the social world? Most action researchers are less concerned with the traditional goal of generalizability than with capturing ...