Number of respondents justifiable for case based research?

Home Forums Default Forum Number of respondents justifiable for case based research?

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
  • #3273
    Marty Reilly

    Hi all,

    I am a PhD student using a multiple case study methodology across 5 organisations within one industry sector. Finding enough respondents in each organisation was problematic to say the least.

    Ideally I was looking at obtaining 6-8 respondents from each org but in some instances have only managed to get 4 or 5.  Has anyone seen research within management which justified this small number of respondents? Whilst I am confident that I’ve got a valid story to tell in each org I am somewhat worried that the small number of respondents may not stand up to scrutiny.

    Any advice is greatly appreciated.


    Kate Golden

    Hi Marty, I hope this is helpful… From what I believe I read in both Stake and Yin’s work on case study is that the case study’s integrity is viewed in terms of the breadth and depth of methods utilized to explore the case(s) in depth. This would include but is not limited to the number of respondents and type of data collected (e.g. focus group, interview, survey), nature of observations, use of additional documentation (e.g. document review, meeting minutes, employee evaluations) and exploration of other extant data sources (e.g. description of program history, annual budget, policies and procedures). I am of the belief that with some or all of these elements included in your work, you have a very valid approach that will absolutely be viewed favorably amongst those who have an understanding of (or who can be convinced) the determinants of rigorous qualitative work. I admit that I am unsure of how this may play out in a management context but if my understanding is correct, rigor is best judged by its ability to tell a complex story from multiple angles. Maybe its more of a marketing challenge…

    Roger Gomm

    To have your findings taken seriously you will need to be able to claim that those people you did interview are representative of much larger numbers of people. How many respondents you need to recruit for this isn’t something which can given as a fixed formula. It depends on the diversity within the population with regard to the matters in which you are interested which presumably in your case is beliefs and opinions of respondents about what happens in the organisation of which they are members, and these will depend on the the types of people they are and their standpoints within the organisation  The best guide to diversity is usually a survey, or even better a census of the actual population of interest or one rather like it. Then you can guage how many different kinds of (whatever it is) there are likely to be and select respondents to represent each kind – though not necessarily in proportion to their occurrence. Such data however are unlikely to be available so the next best strategy is to select respondents according to some scheme of demongraphic diversity: some males- some females, some of each of a scheme of age groups some from each of a scheme of seniority within the organisation and so on – all this as a proxy for the diversity you are really interested in. And failing this, when you are trying to estimate how far you think your findings apply to people you haven’t studied you will need to look at how those you did study fit within this kind of demographic profile and see what kinds of people got missed out and which were over-represented.

    Of course all the additional data sources suggested by Kate will be useful too.

    Greg Guest

    Hi Marty,


    You may want to check out:

    Guest, G., A. Bunce and L. Johnson.  2006.  How many interviews are enough? an experiment with data saturation and variability.  Field Methods 18:59-82.


    We provide some evidence-based recommendations for QL sample sizes. In general though, case studies are chosen based on unique attributes they embody relative to the larger population.  They are used to inform a larger research problem and/or theoretical model.  If this is the rationale behind your choice of case studies, 4-5 should be enough, at least from a methodological perspective.  Whether these number hold up to scrutiny really depends on your audience and how well you make the case for their selection.  As an fyi, in medical research a single case study (i.e., n=1) is often enough to get published.





    Luz Marina


    Soy de Venezuela y estoy interesada en ¿Cuántas entrevistas son suficientes? un experimento con la saturación de datos y la variabilidad.  Métodos de campo 18:59-82.

    En las investigaciones cualitativas que se realizan actualmente, los estudiantes de doctorado aún tiene esa confusión de ¿cuántos sería las entrevistas para justificar el trabajo? a lo que he respondido de que, en la medida que realicen las entrevistas, el investigador se dará cuenta hasta donde llegar, siempre y cuando haga el análisis  de la primera entrevista,  la saturación, es decir, la repiticion continua y reiterada de la misma percepción por parte del entrevistado será el indicativo de que debe cerrar el ciclo de las entrevistas, sin embargo, inician las entrevistas sin realizar nigún análisis de ellas y llegan a tener hasta 20 entrevistado o más, lo cual abruma la cantidad de datos que le proporcionan y también se pierden en la información.

    Sería interesante tener a mano esta publicación para dar respuestas a esa interrogante en los investigadores que comienzan a incursionar acerca de este método de análisis 

    Saludos cordiales

    Luz Marina

Viewing 5 posts - 1 through 5 (of 5 total)
  • The forum ‘Default Forum’ is closed to new topics and replies.