IB Docs (2) Team
Generalizing from case studies
The view that one cannot generalize on the basis of a single case is a common misunderstanding seen on examinations. It is often used as an evaluative strategy. For example, “HM was a case study. One limitation is that case studies cannot be generalized.” This is actually not true.
More discoveries have arisen from intense observation [of individual cases] than from statistics applied to large groups.
William Beveridge (1951)
Why bother with case studies?
It is true that cases are not randomly selected. But then, they are not intended to be random. Purposive sampling is used for case studies. Case study approaches do not produce statistically representative data in this manner. The question is, does this matter?
One problem is that when we think of case studies, people often think of them as a single piece of data. This is a significant error. Think about the following two studies from health psychology. Which one do you think tells us more about stress in the general population?
- Study 1: Cohen (1993) carried out a study to see if stress leads to colds. He gave 276 healthy participants a series of tests to determine their recent exposure to stress. The participants were then randomly allocated to one of two conditions: exposure to either a cold virus or a simple saline solution, which served as the placebo. The researchers found that those who had high levels of stress for a month or more were over twice as likely to get ill as those who had no such stress.
- Study 2: Marmot et al (1997) carried out a case study of over 7000 British civil servants over a period of fives to determine if their level of stress in the workplace had a negative effect on their cardiovascular health. The researchers found that risk factors such as smoking, obesity, and hypertension could account for some of the differences, but the most significant factor was the degree of control that participants felt they had.
The problem with the first study is that there is a single data point - they either did get the cold or they did not. In the case study, however, there are several data points taken over time with a large sample. The fact that the sample is purposive and the study is not experimental in nature does not mean that we cannot then draw inferences about the role of stress on one's health. Although it is true that case studies cannot be replicated, there is often a lot of data that has been consistently and reliably measured.
Statistical generalization is only one of many ways by which people gain and accumulate knowledge. Just because the findings of a case study cannot be generalized to the general population does not mean that it cannot be generalized at all. More importantly, it does not mean that the knowledge that we gain from a single case does not add to our collective understanding of human behaviour. Often such case studies lead to new hypotheses about behaviour and lead to more comprehensive research.
Case studies are also important because they allow for the falsification of theory. In a very famous example, Karl Popper argued that if he had the hypothesis that "all swans are white," it would only take one case of a black swan to challenge this hypothesis. This would mean that the original hypothesis either needs to be rejected or revised. Case studies play a key role in theory development.
A single case study that is supported by other case studies is usually considered more dependable. When another case reaches the same conclusions, it confirms the findings of the first case. Researchers make use of meta-analyses in order to come up with an aggregate conclusion based on the findings of several case studies.
- Case studies allow us to gain information that would not be obtained in any other way.
- Case studies are naturalistic and have high ecological validity
- Most case studies collect both qualitative and quantitative data - and even include experimental research.
- Method triangulation is used. If these different methods yield consistent results, this increases the credibility of the research. We know it was not the choice of how the research was done that resulted in the findings.
- They can be used to falsify existing theory and to generate new hypotheses
- They can be used to explain or predict behaviour in similar contexts - this is called transferability.
- They are longitudinal so they are able to measure the consistency of data over time.
- Many cases studies are prospective in nature, allowing researchers to identify risk groups before a behaviour or health concern develops.
- If they are carried out over a long period of time, researchers may make unexpected discoveries.
Preparing for exams
So, back to HM. You could say that one of the problems with HM is that although he sustained damage to the temporal lobe - specifically, the hippocampus, it may be difficult to transfer the results of the case for at least two reasons: First, prior to the surgery he was suffering from epileptic seizures which could have theoretically caused memory loss; Second, HM was put on medications which could have resulted in memory loss. However, it is important to note that the case study of HM lasted for almost fifty years. There was a very large amount of data that was collected and the data was highly consistent. In addition, further case studies of individuals with hippocampal damage have supported the findings; so in fact, this study may be “generalized.” This study helped psychologists to understand the role of the hippocampus in the transfer of STM to LTM.
Twitter
Facebook
LinkedIn