Resources for a newcomer to qualitative research?
December 4, 2011 7:11 AM Subscribe
I've been conducting a service review, and now that I've actually started getting the interviews done I realise that I haven't a clue how to analyse the data. Can anyone help a lost research newbie?
I'm a medical student, and I'm doing a service review as part of my coursework. The interview questions are fairly simple, basically all variations on "What do you think of [aspect of service]? Do you have any ideas how it can be improved?". There will probably only be around 20 interviews in all, so it's not a vast amount of data.
I have no background in research, and my supervisor doesn't have any background in this sort of research. Not knowing where to start I took out a couple of randomly chosen books on qualitative research from the library, but they left me more confused than when I started. I feel like all the resources are at grad school level and I'm operating at a preschool level!
I desperately want to produce something that's valid, not just to pass my coursework, but also to pay back the time of the people I've interviewed and make this a worthwhile process. Thanks in advance!
I'm a medical student, and I'm doing a service review as part of my coursework. The interview questions are fairly simple, basically all variations on "What do you think of [aspect of service]? Do you have any ideas how it can be improved?". There will probably only be around 20 interviews in all, so it's not a vast amount of data.
I have no background in research, and my supervisor doesn't have any background in this sort of research. Not knowing where to start I took out a couple of randomly chosen books on qualitative research from the library, but they left me more confused than when I started. I feel like all the resources are at grad school level and I'm operating at a preschool level!
I desperately want to produce something that's valid, not just to pass my coursework, but also to pay back the time of the people I've interviewed and make this a worthwhile process. Thanks in advance!
Response by poster: Thank you - I need to think about the research question a bit. It's going to be something along the lines of "Are we providing the service that people want/need?", but that probably needs some finesse.
posted by Coobeastie at 7:47 AM on December 4, 2011
posted by Coobeastie at 7:47 AM on December 4, 2011
If you really want to do it right, take a qual methods course. Someone on your campus is offering it. This way you'll learn about IRB and coding and everything.
posted by k8t at 8:12 AM on December 4, 2011
posted by k8t at 8:12 AM on December 4, 2011
20 interviews of qualitative data could be a lot of data, all depending on how you go about collecting the data. Are you doing oral interviews or are you doing a written survey? Interviews allow you to get a better sense of how participants construct the field themselves - but the encoding and analysis of the data are extremely time-consuming. An interview-based master's thesis, for instance, often relies on 20 interviews because the amount of work involved in transcribing 20 interviews is beyond the pale. Surveys allow you to get to what you're looking for rather rapidly, at the cost of perhaps missing out on some of the nuances interviews might bring. I'm pointing this so that you can plan accordingly.
I'm assuming from how you presented your question that you plan on doing 20 oral interviews. How long will these lasts? How will the data be recorded (notes? Digital recorder?)? How will the data be encoded (transcriptions? Transcription using a qualitative analysis software?). If possible, I recommend asking your supervisor to grant funds so the interviews can be transcribed by professionals in a short amount of time (you University likely has a few companies they deal with). That way, you can focus on the analysis.
As Pantarel70 points out, analyzing the data depends on your research questions and your premise. You can either have an exploratory research question (fake ex: "we notice non-compliance among health care practitioners and would like narratives that would allow us to explain the factors behind this phenomena") or you can have a hypothesis-type research ( fake ex: "Billig et al have demonstrated that the notion of expertise increase risk-taking factors in accounting firms. Since the risks have life and death consequences in the health sciences, we hypothesize that expertise in this field leads to less risk taking".) In the first (exploratory), the challenge is that you want the pattern to emerge from the data. In the second (hypothesis), you have a better sense of what you are looking for in the data.
Then there are numerous ways to go about validating the data. You can use a qualitative research software to delineate them (note, this, again, takes a lot of time, and requires that the data be transcribed with the software codes in mind). While software provide the security of "objectivity", they often lack in data triangulation (they analyze discourse and are poor at following cause and effects). You can go about doing critical discourse analysis (check out Norman Fairclough), where you will be the one to outline the most salient data by - then validate it by having someone else look at it (this is more subjective, but affords the possibility of following cause and effects outside of discourse). You can even, depending on the study, ask participants for retroaction on your analysis.
Best of luck!
posted by Milau at 8:19 AM on December 4, 2011 [2 favorites]
I'm assuming from how you presented your question that you plan on doing 20 oral interviews. How long will these lasts? How will the data be recorded (notes? Digital recorder?)? How will the data be encoded (transcriptions? Transcription using a qualitative analysis software?). If possible, I recommend asking your supervisor to grant funds so the interviews can be transcribed by professionals in a short amount of time (you University likely has a few companies they deal with). That way, you can focus on the analysis.
As Pantarel70 points out, analyzing the data depends on your research questions and your premise. You can either have an exploratory research question (fake ex: "we notice non-compliance among health care practitioners and would like narratives that would allow us to explain the factors behind this phenomena") or you can have a hypothesis-type research ( fake ex: "Billig et al have demonstrated that the notion of expertise increase risk-taking factors in accounting firms. Since the risks have life and death consequences in the health sciences, we hypothesize that expertise in this field leads to less risk taking".) In the first (exploratory), the challenge is that you want the pattern to emerge from the data. In the second (hypothesis), you have a better sense of what you are looking for in the data.
Then there are numerous ways to go about validating the data. You can use a qualitative research software to delineate them (note, this, again, takes a lot of time, and requires that the data be transcribed with the software codes in mind). While software provide the security of "objectivity", they often lack in data triangulation (they analyze discourse and are poor at following cause and effects). You can go about doing critical discourse analysis (check out Norman Fairclough), where you will be the one to outline the most salient data by - then validate it by having someone else look at it (this is more subjective, but affords the possibility of following cause and effects outside of discourse). You can even, depending on the study, ask participants for retroaction on your analysis.
Best of luck!
posted by Milau at 8:19 AM on December 4, 2011 [2 favorites]
Just saw your question: "Are we providing the service that people want/need?", but that probably needs some finesse.
That's an exploratory question. I suggest you analyze the data by using the data itself to set a limited number of factors. If you do the interviews yourself, try to sort them right after the interviews. Group those who are similar together; figure out which ones differ. Map out the terrain. Analyze one of each group to outline categories (try to pick the one you find are most telling). What categories emerge out of these interviews? Once you've laid out a first "schema" of the data, use the rest of the interviews to validate or ameliorate your first analysis. Then prioritize the needs: you might want to list out the ones mentioned the most often firsts. Otherwise, if you have two fairly delineated groups (those who are mostly happy with the services and those who aren't), map out why those groups exists and what explains the discrepancies).
posted by Milau at 8:35 AM on December 4, 2011
That's an exploratory question. I suggest you analyze the data by using the data itself to set a limited number of factors. If you do the interviews yourself, try to sort them right after the interviews. Group those who are similar together; figure out which ones differ. Map out the terrain. Analyze one of each group to outline categories (try to pick the one you find are most telling). What categories emerge out of these interviews? Once you've laid out a first "schema" of the data, use the rest of the interviews to validate or ameliorate your first analysis. Then prioritize the needs: you might want to list out the ones mentioned the most often firsts. Otherwise, if you have two fairly delineated groups (those who are mostly happy with the services and those who aren't), map out why those groups exists and what explains the discrepancies).
posted by Milau at 8:35 AM on December 4, 2011
This short article gives a very basic outline of two different approaches to qualitative content analysis.
posted by DiscourseMarker at 8:39 AM on December 4, 2011
posted by DiscourseMarker at 8:39 AM on December 4, 2011
Response by poster: Thank you so much everyone! Though I may have to have a follow-up question on how to deal with having massively bitten off more than I can chew.
posted by Coobeastie at 10:30 AM on December 4, 2011
posted by Coobeastie at 10:30 AM on December 4, 2011
I always suggest that people new to qualitative data collection read Learning From Strangers. It's a terrific overview of the entire enterprise of qualitative inquiry. Seems like there are various pdfs of it on the web, so you might be able to check it out before buying.
posted by jasper411 at 10:38 AM on December 4, 2011
posted by jasper411 at 10:38 AM on December 4, 2011
This thread is closed to new comments.
What you're going to be doing (which you've probably already gathered from the qual research books you've gotten) is "coding" the data. Looking for patterns of responses, the presence of certain themes, etc.
How you go about this is, obviously, hugely variable, but is going to depend largely on what you want to know, i.e. your research question.
Do you have some things you are looking to hear about in the interviews? Or do you just want to see what emerges as you work through it?
I would recommend 1.) first getting a firm handle on WHAT it is you're looking for (maybe talking to your supervisor) then 2.) listening to some of the interviews just to get a sense of them, without doing any coding at first.
Earl Babbie's Basics of Social Research is a good all-around social science methods book that treats things pretty simply. He has plenty of information on coding interview data that you might find useful.
posted by pantarei70 at 7:29 AM on December 4, 2011 [1 favorite]