What is the purpose of ethnography – Are we guessing what people do?

I think I’ve mentioned this a few times in the class, but I’m often afraid that my understanding of American culture (including race, ethnicity, gender, class, etc., since those are very different from Japanese ones). After I read Becker’s article, I started thinking, “what if I’m merely guessing what American culture is, and what if I’m guessing wrong?”

As Becker says, some “descriptions of drug use are pure fantasy on the part of the researchers who publish them” (p. 59). Another example he mentions in his article was that college students looked at letter carriers from a “stratification point of view” (p.63) which is far from what the letter carriers opinion about their preference for towns they work. Considering these two examples, do you think we tend to fantasize what we want to see as researchers?

I have been reading about the Naturalistic Inquiry, which I already mentioned it in the previous blog post. I had a chance to read more about it last week and I would like to share how the Naturalistic inquirers see qualitative research methodology and how they ensure the credibility of such research outcomes. Researchers, Lincoln & Guba (1985) and Wolf & Tymitz (1979) talk about the trustworthiness/dependability of qualitative methods in their articles. Especially, Lincoln and Guba (1985) use four categories: credibility, transferability, dependability, and confirmability.

Here are the descriptions of the categories:

• Credibility: The credibility standard requires a naturalistic study to be believable to critical and to be approved by the persons who provided the information gathered during the study. The credibility of the NI can be tested by prolonged engagement, persistent observation, triangulation, peer debriefing, negative case analysis, progressive subjectivity checks, and member checking (Lincoln & Guba, 1985).

• Transferability: This criterion refers to the applicability of findings in one context to other contexts or settings. The target context must be compared to the research context to identify similarities. The transferability analysis is facilitated by clear descriptions of the time and context in which working hypotheses are developed by the naturalistic inquirer (e.g. Thick description by C. Geertz).

• Dependability: A researcher looks to see if the researcher has been careless or made mistakes in conceptualizations the study, collecting the data, interpreting the finding and reporting results. A major technique for assessing dependability is the dependability audit in which an independent auditor reviews the activities of the researcher (as recorded in an audit trail in fieldnotes, archives, and reports) to see how well the techniques for meeting the credibility and transferability standards have been followed.

• Confirmability: It refers to the quality of the results produced by an inquiry in terms of how well they are supported by informants who are involved in the study and by events that are independent of the inquirer. Reference to literature and findings by other authors can strengthen confirmability of the study as well as information and interpretation by people other than inquirer from the research site itself. The comfirmability audit is a way to as if the data and interpretations made by the inquirer are supported by material in the audit trail, are internally coherent, and represent more than “figments of the inquirer’s imagination” (Lincoln & Guba, 1989).

In the Credibility part, Lincoln & Guba mentioned the importance of member checking: data collected during the research are reviewed by the participants and members who provided the data. I’m sure some people do not like to share the data with the participants, but I think this is better than guessing without confidence or guessing wrong. Becker touched on this issued at the end of his article and this is something that we, researchers, need to think about. I think that we should not interpret or infer what we research. Rather, I think that we need to leave our data open-ended and let the readers interpret in their ways.