Public Interest Anthropology at Penn

In ethnography, data analysis most usually takes place throughout the project. That is to say, we learn from the data we gather during one visit to the field helps us learn what to watch for, notice, or ask during the next visit. As fieldwork progresses, constantly refining our ideas of what might be happening at the site. At this level, data analysis is ongoing and helps fieldwork gain momentum towards useful information.

Presumably, however, there eventually comes a point when we turn our attention more fully to working with the data we have gathered already, often after leaving or limiting visits to the fieldsite. What does our data mean? What have we learned? What can we say regarding our guiding question, or others that we may know how to ask now based on the research? In short, how might we best analyze the data we have gathered?

While there is no single canonical way to approach ethnographic data, the following points may be useful in helping us arrive at some conclusions:

Read through the fieldnotes, notes on interviews, interview transcripts, site documents, or whatever data has been gathered several times. Becoming very familiar with the information at the start helps to to proceed.

Mark the data and take notes on any patterns, connections, similarities, or contrastive points in the data. Does anything stand out as a usual way of doing things at the site? What seems unusual, and why? What becomes clear analytically that was not clear before?

Consider facilitating the above process, called coding, by using a computer-assisted data analysis program like Nud/ist or Ethnograph. Students should see the computer lab TA for assistance in getting started.

Follow up on what you noticed above by looking for "local categories of meaning" in the data. What terms do the informants have for things? What can you as a researcher identify as themes, even if the informants don't? Remember that the main purpose of ethnography is eliciting "native points of view"; these "local categories" are its components. Try to come up with a list of "local categories" from the data.

Test the categories and explanations you have started to draw out of the data against the variety of cases you have recorded. Are there alternative explanations for what you think you have seen so far? What can you learn from looking at the data from a variety of perspectives?

Try triangulating among the various forms of data you have gathered. If a point or an explanation holds across several sources you have gathered - if, for example, it can be supported by fieldnotes, interviews, and/or site documents - then you can be more sure that you have found something integral to understanding your site.

Consider trying "respondent validation", or explaining your developing conclusions to your informants. The informants might be in a position to share additional things which help to confirm or complicate what you have learned. Remember, of course, that the informants are still socially positioned, and may or may not agree with the analysis in part based on their positions or perspectives within the social network have investigated. Agreement from informants doesn't necessarily mean we're right, and disagreement doesn't necessarily mean we're wrong.

Once we have arrived at some conclusions regarding the data gathered, we must consider the question of how to focus on the guiding question which drove the research. Can that question be answered from what we learned? Is another question more appropriate? What other questions has the research provoked? Remembering that the thesis sentence must be an answer to the guiding question, it is important to work back and forth between our emerging conclusions and guiding question to produce a cohesive paper.

References

Hammersley, Martyn and Paul Atkinson
1995 "Documents" In Ethnography: Principles in Practice. Pp. 157-174. Second edition. New York: Routledge.