PDF pending restoration
ERIC Number: ED141514
Record Type: RIE
Publication Date: 1977-Apr
Reference Count: 0
Representative and Quasi-Representative Designs for the Improvement of Adult Education Research and Evaluation Studies.
Scissons, Edward H.
Traditional research literature distinguishes between two general types of threats to the generalizability of experimental findings: Internal validity and external validity. Relatively minor importance has been attached to external validity in educational literature. Bracht and Glass elaborate on external validity and deal with two types: Population validity and ecological validity. Snow indicates that the biggest threat to external validity occurs when the experiment does not fit the phenomena being studied. Traditional experiments are nonecologically oriented, tend to manipulate the subject for experimental convenience, and yield results which are non-representative. Historically, basic research has emphasized internal validity and applied research has been concerned with immediate problem solving (external generalizability). Generalizability must be the sole factor in judging all good research, and the fundamental design principle governing generalizability is representativeness. There are a number of strategies educational researchers may use to build significant and generalizable studies, for example, imbedding experimental procedures in existing structures in as nondisruptive a manner as possible, replicating experiments, using placebo treatment groups, examining the utility of proposed research, and not overgeneralizing from results. Adult educators also must be concerned with the generalizability of their research. They must be applied researchers, emphasizing external validity over internal validity. (LMS)
Publication Type: Speeches/Meeting Papers
Education Level: N/A
Authoring Institution: N/A
Note: Paper presented at the Adult Education Research Conference (Minneapolis, Minnesota, April 1977)