Case Study Methodology

I’ve been reading Yin’s Case Study Research (2003) to get some grounding in putting together my research. Here’s my initial take from it.

Yin lists the 5 main components of research design. This is how I think my work relates:

  1. A study’s questions – I’m happy that my research questions are relevant and useful and feedback from Guy is positive (note – need to post the revised proposal to the blog!)
  2. My propositions based on the questions – my initial direction is that storytelling as part of an  will have a positive qualitative impact on an organisation’s communictaion and internal working practices but the counter proposition is that any gains may be marginal or there will be no significant change in the state of affairs given the effort required to deveop skills and approaches. These propositions need sharpening up, they feel flabby. Yin says that there are valid reasons for not having valid propositions if the case study is more exploratory in nature
  3. The unit of analysis – bit trickier this one. The reseach is looking at Netskills (although there are some “dotted lines” to people in other organisations)but the study itself will mainly be focussing on the experiences of a subset of employees. Individual experiences are important to that but the case study is trying to look at the orgainisational level. It will be important to include key people not involved in the storytelling development activities  but who are stakeholders (e.g. the director, the training manager, “publicity officer”)
  4. The logic linking the data to the proposition – Again, problematic at this stage. I’m hoping that the interviews and observations will bring out some emergent themes; I certainly don’t want to dictate these, given my participant observer status. It’s hard to say at this stage what the logic will be for linking these data back to the propositions but I will need to make sure that my interview questions are closely linked to the research questions. 
  5. Criteria for interpreting findings – The data is going to be predominantly qualitative and narrative in nature (I’ll be asking for participants to describe their own journeys and add their own interpretations to events). As such there will be a significant subjective interpretation of the data and much emphasis on the participants’ opinions as to the value of the work they’ve undertaken. Another issue is timescales, which are not huge! In reality, the effects, positive or not, of this development work will be cumulatively visible over an extended period of months or years so this case study may just provide the basis for further research or form the template for case studies on other organisations

The role of theory

Yin quotes Sutton and Shaw (1995, p378) who describe a theory as “…a [hypothetical] story about why acts, events, structure and thoughts occur.” It’s nice to see the work story cropping up there but I’m not entirely sure what the difference between theory and proposition is.

I’ve tried to articulate my theory of story in this context to myself but I’m not coming up with anything that I find particularly satisfying yet. This is a question I want to take to Guy at our next meeting.

Perhaps the theory will emerge as a result of the research and trying to manufacture one now isn’t going to be helpful. Another thing I’ll need to bear in mind is how much I’ll be able to generalise from this cae study to a universal theory. I’ll have to be very caeful there.

Criteria for judging quality of reseach design

According to Yin there are 4 tests of quality

  1. Construct validity – can I demonstrate the data I’m collecting and measuring in the case study actually reflects the focus of the study. I’ll need to show this in my analysis but at this stage I think that looking for quantitative measures in things like social media activity (follows, mentions, retweets etc) or numbers of people booking on workshops for example wouldn’t reflect the focus of my study.
  2. Internal validity – If I’m establishing cause an effect between x and y am I able to show that it’s not something that has been caused by z? It will be important to look at the wider context at Netskills. There has been quite a lot of activity recently in team and personal development and there are also a number of staff taking on different roles in the team. This might have an impact on the study, as well as drivers coming from the changes that are happening to Jisc, from whom we receive a proportion of our funding.
  3. External validity – are the findings generalisable beyond the cae study. I anticipate that I’ll not be able to say definitively that what has happened in the Netskills context  will be directly applicable to other organisations. Although there are similar organisations to Netskills, even with Jisc, much of what we’ve been doing relates closely to individuals and the Netskills working culture. This isn’t to say it won’t be applicable, just that there will have to be clear and serious caveats written into the analysis.
  4. Reliability – would the results of the case study be replicable by another researcher following the same process? This links closely to point 3 but it’s a reminder to me to make sure that my approach is clearly documented and explained.

What type of case is Netskills’ storytelling development?

Going by Yin’s terminology I think that the Netskills case is representative, rather than a unique or revelatory case, in that I’m trying to establish what lessons could be learned for similar organisations. Given a longer time frame of years a longitudinal approach might yield more reliable results so again, this could form the basis for further study.


I need to be aware of Yin’s advice that multiple case studies are preferred over single as it means replication is possible. I’m not going to have that opportunity so that will necessarily colour my findings.

I also need to keep the case study design open and be willing to change the design as the study progresses.

Tagged with: , ,

Leave a Reply

Your email address will not be published. Required fields are marked *