Amy B. Smoyer, PhD

Incarcerated Lives, Health & Social Work


As President Obama's speech highlighted last night, there are 14 million unemployed people in the USA, about 9% of the population.  In Spain, the rate of unemployment is over 20%, almost 5 million people.  For these folks, there is no need to explain the concept of redundancy.  Redundancy as in, "We need to eliminate redundancies in order to control costs and so you are laid off."  In this tough economy, companies can't afford to have two people, or one person and a computer, doing the same job.  Still, I think the use of the word "redundancy" in this context is callous.  Every individual is, by definition, a unique person.  The suggestion that one person's contribution can completely encompass another's denies the self that we bring to everything, even screwing widgets into place.

Anyway, this is all semantics.  The talking heads are constantly searching for new words to gloss over the realities of the world and they probably don't mean it.  More problematic for me, is that qualitative social scientists use this word as well - when talking about how many people should be included in a research project.

The question of sample size - how many people should be interviewed? - is a point of ongoing discussion among qualitative researchers.  Most seem to agree that the answer to this question is, "It depends."  A little bit vague and unsatisfying, right?  

Here is a more extensive and often cited answer from Lincoln & Guba (1985, p. 202):
Participants should be sampled, "to the point of redundancy...the size of the sample is determined by informational considerations.  If the purpose is to maximize information, the sampling is terminated when no new information is forthcoming from the new sampled units."
Hmmm....a little less vague, but still unsatisfying.

In the case of my dissertation project, I originally proposed interviewing 30 formerly incarcerated women.  This number seemed reasonable given my resources and reflects the experience of my committee members and other qualitative researchers about how many is generally "enough."  As it turned out, I interviewed 25 women before leaving for Spain.  The arrival of Irene, and subsequent power and phone outages in the days that followed, forced two potential participants to cancel.  But really I can't "blame" the sample size on Irene, there was a lot going on and this was the best I could do under the circumstances.

The question is, did I reach the "point of redundancy"?  On the one hand, yes.  I found that I often knew the descriptive information that women shared with me about their prison food experiences.  I heard about the endless trays of cake, early morning hot cereals and concoctions created in the dorms from commissary snacks.  The same issues about trust and control came up again and again.  But, honestly, each woman brought her own impressions and reflections on the prison experience.  No two women were completely alike.  Every conversation shed new light on what it can mean to be incarcerated.

It may be helpful to use a Venn diagram to think about the sample size question.  I like this one that I found at on the Cornell Math Department website:

There is no redundancy here.  Each story is unique but there are points of overlap.  Points where all 4 stories overlap and points where just 2 or 3 overlap.  Perhaps we have "enough" stories when that center point were all the stories connect is thick and dark.  That and life nipping at your heels - time to pull up the tents and move on.  I honestly believe that if I kept at it until "no new information is forthcoming," I'd never be done.  And so I am not done, just done for now.   As they say in elementary school, "You get what you get and you don't get upset."