Thursday, September 13, 2012

9.13 Analyzing interview data

NIH Training certificates are due Sept 18.  If you haven't already, please send the link to your certificate to the course email.

Update on the room situation:  I have been informed that we have a computer lab - but it needs to be confirmed.  I will post the room number here and on the postings board beside CAS 313 as soon as I have final authorization.

What we did in class:  We began with some talk about the data you posted on your blogs.  We observed that you each had a slightly different format, and that all of you did some good observing and writing down what happened.  So good job!

Next we talked through the process for analysis when the data you are working on are the messy interactions you find in an interview - rather than a set of shapes in a logic problem.  We noted that even though the data are quite different = the steps in the process are very much the same.



1. Coding: identify/name the features of your data.  For our interview data, some of the "features" or classifications you might note include:
  • questions + answers.  Identifying talk as either a question or answer can help answer questions about who is directing the interview (authority), the participants' investment or engagement with the process (hypothesis = if subjects ask more questions they are more engaged?), what kinds of information can be communicated through questions versus statements = and so on.  
  • speaker's distance or emotional connection to what s/he is saying.  noting/naming connection or the speaker's relationship to what s/he is saying can help answer questions about what is important in an interview
  • changes in attitude, perspective, or emotion
  • statements of "fact" and statements where the speaker interprets/responsd to fact
  • stories + parts of stories (introduction, presentation of problem, resolution of the problem, conclusion)
  • resolution (or lack of resolution) to stories: stories might have positive or negative or unresolved endings.  
  • the storyteller's language choices (are there any repeated phrases or words that signal characteristic feelings or ideas?)
I realize the notes we took will not have sufficient detail to provide solid data for most of these classifications - but that is OK.  This gives you a simplified, first opportunity to look at data and figure out how to name  what you see.

2.  Classifying:  place the different examples of questions and answers (or some other name/code) into groups.  Look for similarities and differences in your codes (as we looked for similarities and differences in the different kinds of interviews at the beginning of class).  In our analysis of Andrea's data - we kind of did this backwards.  We started by noticing the category = after the interview comments - and then we named different kinds of "after the interview comments": observations, interpretations, and evaluations.

3.  Identifying patterns.  As we talked throughAndrea's data, we identified a small pattern in terms of the order of the different kinds after the interview comments.   Corrine noted a small pattern where a subject became emotional - and that emotion seemed to influence the next question asked by the interviewer (were you confused. . . ?)  Other patterns that you noted in the data were connections between age and the the subject's perception of 911, and so on.

4. Developing hypotheses about what the patterns mean or how they fit together.  A hypotheses is a "guess" about what a pattern means or how it relates to other patterns.  In our discussion of the sequence of after the interview comments, I suggested that this was the same sequence most people use to tell stories.  They describe what happened, interpret what the events mean, and the offer some kind of meaning/moral/or evaluation.  To develop this hypothesis - i made a connection between a pattern I observed in the data - and some other pattern (story telling patterns) that is "out there" in the world.  

5. Testing the hypotheses. Continuing #4, we looked at the sequence of after the interview comments in some of the other data  - and the pattern we saw in Andrea's account was not really repeated.  So my hypothesis wasn't really a good one.  So then we tried some other hypotheses about what the patterns in our data meant.

6.  Creating a "theoretical story"is to put together an explanation of the relationships among the features of the interview you are interested in.    After you find some hypotheses that fit  the data - put them together to see if you can develop an explanation.  Several of you were looking at relationships between how old the subject was, the subject's emotions, where they were when they found out, and so on.  So we started working. on a theory - to explain why/how being a certain age shaped the experience of 911. 

I really enjoyed this class and was impressed with how much information we got out of your data!

For next class:
Blog 4: Set up an analysis of some of the oral history data posted for Blog 3.  You can analyze data from one blog or several.  The point of this exercise is for you to practice the analytic process we went through in class.  Name and classify what you see in the interview; look for and describe patterns; put forward a hypothesis and test it = and see if you can come up with a theory (explanation) that accounts for the patterns and relationships you see in the data.

On the calendar it says to read the literacy narratives, but instead, we are going to start with the shaggy dog stories.  They are posted at this link.

In class we will begin by talking about your experience working on the oral history data.  Come to class prepared to talk about what you learned (any tricks you discovered), where you got stuck, and what you'd like more practice doing.  Then we will work on a slightly different kind of analysis using the shaggy dog stories.  

See you next week! 


No comments:

Post a Comment