Sunday, December 9, 2012

Sample Size and Qualitative Research

The Great Wave off Kanagawa
The Great Wave off Kanagawa (Photo credit: Wikipedia)
Someone had to do it.  Do what? Make a definitive answer to what the sample size should be in a qualitative research study published by a peer reviewed journal.

In this case the someone was Shari L. Dworkin, Associate Editor of the Archives of Sexual Behavior, the individual responsible for qualitative research submissions at this journal.  She takes her stand in an editorial (2012) 41:1319-1320.  Henceforth, the policy of her journal will be to recommend that:

25-30 participants is the minimum sample size required to reach saturation and redundancy in grounded theory studies that use in-depth interviews. 

A part of me admires her for putting a number to her answer.  I, too, have struggled many times when people ask me that question--"How many interviews does it take?" "What's enough data?"  "How do qualitative researchers figure out what is the right sample size?"  The answer that most qualitative researchers give:  "It depends," is not very satisfying for the novice or the researcher who is struggling to gain an understanding of this methodology. How easy it would be just to say "25-30; take it or leave it". 

In her editorial, she does a good job of talking through the rationale for her answer and discussing the literature upon which she drew for her answer (the primary grounded theory gurus).  She identifies this number as about right if you are seeking to find new leads...not much new coming up. 

Just as I admire her...I also fear for her, because I worry that, in making her editorial decision, there is a lot of qualitative research territory that she hasn't taken into consideration.  So, let me unpack what some of the "it depends" concerns would be:

  • Grounded theory studies are not the whole of qualitative research, and in relying so heavily on this one "flavor" in our field have important views been ignored.  (Now I personally believe that the whole qualitative research flavor discussion is over-rated...see earlier blog posts...but I think it may be relevant here.)
  • Grounded theory studies are not always solely composed of interviews (as the editorial appears to suggest).  I would want to know how interviews are mixed with other forms of data before I could say that, for instance, 10 interviews, was insufficient.  [How many observations?  Of what kind and length?  Was there significant document analysis?  Was visual data obtained?  What about the use of journals from the participants?]  I would need to know how the interviews were contextualized in the sea of data that is possible to gather in a qualitative research study. 
  • What do you mean by "in-depth interviews"?  From the editorial, I have the sense we are talking one interview per individual (maybe 1 hour?).  In my training with Buddy would have meant 4-5 interviews (yes 4-5 hours divided up over a period of time) with one individual.  Whose in-depth are we using? 
  • What does 25-30 mean?  In the editorial it appears that this would be 25-30 interviews with separate individuals.  If you had 5 focus group interviews with 5 individuals in each (5 x 5 = 25)...would that count? I am working on a study that relied primarily on focus groups--I can report that we interviewed hundreds...but in groups. 
  • What will a ruling like this do to the incredibly important smaller exploratory start-up study, and/or the dissertation study?  These are often of smaller size than a very well funded study by a senior researcher.  Will these valuable efforts become unpublishable?  Are we starting a new kind of ranking order for peer reviewed journals?  
When I shared my concerns about the policy described here with a friend who is a qualitative researcher, her response was--"...setting a number of interviews seems to me like giving up the expert's responsibility to vet the methodology.  Maybe the problem is that the methodology wasn't adequately described.  Shouldn't we review the number of interviews as part of reviewing the methodological description.  We still have that responsibility."  

I invite all comment on this issue.  This is the first time I have seen a journal take a stand of this sort.  I am glad that this has happened, because it opens up space for a well needed discussion. 

Enhanced by Zemanta

1 comment:

julie anderson said...

More even patience and how to tackle to different people are also very important as it is a dealing with the customers!
Sample jobs