What is “reliability” in ethnographic research?

In research which does not use qualitative instruments, how do you know that your research is reliable? Generally speaking, research is reliable if the instrument yields the same results on repeated trials. The National Research Council report also argues that educational research must be scientific and can “yield findings that replicate and generalize across the studies” (Feuer, Towne, Shavelson, 2002, p. 5), but how can we replicate the findings of ethnography which mainly looks at human behavior?

I had a chance to talk to a museum program evaluator who has Ph.D in Education. Dr. Deborah Perry uses a methodology called “naturalistic inquiry” which studies a group of people in natural settings. She recommended me to read a book by Dr. David Williams, which is available online (http://education.byu.edu/ipt/williams/index.html). In Chapter 5, he talks about standards for judging natural inquiry and touches on the issue of reliability.

First, he defines naturalistic inquiry

• focuses on discovery and participants interactions
• claims that realities are multiple, constructed, and holistic
• holds that only-time and context-bound working hypothesis
• claims that the knower and the known are interactive and inseparable

In addition, he says that each of us constructs a view of reality different ways. Thus, it is essential to seek for each person’s interpretation/construction of reality, instead of finding out a “true” definition of everyone’s experience. In other words, we cannot generalize human behavior. Therefore, I think that reliability of naturalistic inquiry and qualitative research depends on how researchers accept their own and others’ subjectivity and project it onto the interpretation of constructed reality. I think it is fun to think about how ethnographers think about the reliability of their research projects.