“Our methodology has not kept up with our ontologies…” we heard in our opening lecture of Case Study Research Methods, as part of the Oslo summer school course in comparative social science studies. This statement was originally made by political economist Peter Hall with regards to comparative work in political science. This is likely to be even more true in our relatively nascent field of sustainability research. While there is a general recognition that mixed methods are often necessary to adequately understand complex social ecological systems, many of the methods and approaches commonly used remain more quantitative and statistical. I believe that the methods taught from a political science perspective can be incredibly valuable for our work as sustainability scientists, especially as we aim to find and create generalizable theories and lessons learned from comparative cases.
First of all, the summer school was incredibly well organised and also managed to create a great community! So for the time that we were not on city tours, or swimming in the sea, the course covered everything from inferences about causal effects and mechanisms, to concept formation and measurement, to typological theory, fuzzy set analysis, process tracing and congruence and counterfactual analysis and multi-method research. The course was taught by Professor Andrew Bennett, who wrote the well-known methods book Case Studies and Theory Development in the Social Sciences (George and Bennett, 2005). I highly recommend the book for anyone interested in mid-range theory development and multiple methods, particularly formal modelling, statistics with case studies.
The case for case study analysis
We began the course by situating case study methodology in the history of social science methods and covering philosophical premises. The view presented by Professor Bennett, in line with the above-mentioned book, is written in part as a critique to the widely cited book “Designing Social Inquiry” (King, Keohane and Verba, 1994). King, Keohane and Verba’s main point is there is one logic of inference and social scientists would do well to adopt a more standardised approached borrowing from more quantitative tools.
The starting point for our course was that case study analysis can help us understand causal mechanisms that a more random statistical approach could not. Goertz and Mahoney (2006) neatly outline 10 main contrasts between qualitative and quantitative research:
In particular, I think the focus of case analysis on the “causes of effects” approach can lend itself valuably to social ecological system research (point 1). We are often interested in questions such as: “what is the cause of this regime shift?” or “what is the cause of a poverty trap in a given social ecological context?”
Case studies can allow us to look at a large number of intervening variables and inductively observe unexpected aspects of the operation of a particular causal mechanism (George and Bennett 2005, p 21). George and Bennett defines causal mechanisms as “unobservable physical, social, or psychological processes through which agents with causal capacities operate, but only in specific contexts or conditions, to transfer energy, information or matter to other entities” thus leading to certain outcomes (ibid p.137).
Professor Bennett’s approach to research design was helpful, and provides a starting point for thinking about steps for theory development in social ecological system research.
- Specify the problem, what is your puzzle, and what novel contribution can you make? Two main ways to make a contribution: background theory novelty (if your theory predicts something no other theory predicts) and use novelty (different information is used to test a theory than was used to generate the theory)
- Specify the variables: what are the independent and dependent variables?
- Case comparison: most similar cases have similar independent variables and different outcomes; least similar have different independent variables and similar outcomes.
- Within case: use process tracing to understand causal processes
- NOTE: coming up with these variables will likely require coming up with new concepts
- Case selection: in case studies one cannot have any 0,0,0 cases (where the dependent variable is present).
- Could have most similar or least similar research design, plus it’s always interesting to study outliers
- 12 criteria
- Describing variance
- Questions to test hypotheses
One method particularly interesting for the SES LINK group, and social ecological systems researchers more generally is typological theorising.
Typologies are nothing new to sustainability scientists, creating ideal types, and characterising variants of observable phenomena in order to help our explanatory and predictive power in future similar cases. Typological theory takes this a step further, and can be useful in clarifying which cases and research designs are possible. Bennett defines typological theory as a theory that specifies independent variables and provides hypotheses on how these variables operate individually and in conjunction with each other under certain conditions, acting on a specified dependent variable. Importantly, it’s a way to model complexity.
Typological theories see cases as combinations of variables, or types, usually creating through the clustering of dichotomous nominal variables.
Developing a typological theory:
- Identify the possible outcomes on the dependent variable (you could typologies the DV with various combinations of outcomes)
- Identify the independent variables, put them in nominal categories, and create a typological space of all their possible combinations
- Theorize on the expected outcome for each type
- Place the historical cases in the typological space, including negative cases
- Compare the cases’ actual outcomes with your initial theorizing:
- Identify Anomalies: unexpected outcomes for a type, or different outcomes for the same type
- Consider whether “empty boxes” are impossible or just unlikely types
Such typologising can also be very important in informing case selection. For example Professor Bennett used this approach in defining opportunities for intervention in Soviet-Russian military interventions from 1973-96. He defines 6 independent variables for 6 different cases where the Soviet Union exploited the opportunity, and 5 in which they did not exploit the opportunity to invade. From this typology he is able to conclude that Soviet interventionism declined in the 1980s (e.g. some of the cases had the same values for independent variables and different outcomes), and according to his hypothesis because they learned from their mistakes.
One of the main issues I still struggle with is how to deal with context, and to be confident that the 3-4 variables you’ve selected are the most important ones. E.g. had the opportunity to invade Afghanistan in the 1980s arisen, how can we be sure that the USSR would not have invaded? The history of the Great Game in Afghanistan between Imperial Russia and England may play an important causal role for intervention. Nevertheless, this method seems like it could be very useful in creating a typology of social ecological system interactions, as well as playing a crucial part in cross case comparison research design as we further develop and generalise knowledge and theory from cases.
I haven’t been able to even start to do this dense course justice, but hope to have sparked some interest in using this methods for comparative work in social ecological system research. I look forward to continue to work through more of these methods, including not only typology theorising, but also process tracing and qualitative comparative analysis in advancing our theories on social ecological traps!
For further reading:
- Alexander L. George and Andrew Bennett, Case Studies and Theory Development in the Social Sciences (MIT Press 2005).
- Henry Brady and David Collier, Rethinking Social Inquiry (second edition, 2010)
- Gary Goertz, Social Science Concepts: A User’s Guide, (Princeton, 2005).
- Gary King, Robert Keohane, and Sidney Verba, Designing Social Inquiry (Princeton University Press, 1994).