Extreme Cases

Conceptual Overview and Discussion

When starting a case study project, one of the most important decisions to be made is selecting a strategy for sampling the cases. As Bent Flyvbjerg has pointed out, the selection can be based on random sampling, as is done in more traditional quantitative studies. However, the very nature of the case study research, as an explorative mode of conducting research, more often leads to the use of information-oriented selection strategies instead of randomization. The information-oriented selection strategy is a deliberate way of selecting the cases for the study. It is based on the idea of trying to find paradigmatic cases, critical cases, maximum variation cases, or extreme or deviant cases with respect to the study population and the questions being posed to this group. When using the maximum variation strategy, cases that reflect both the highest and lowest status with respect to the variable(s) being measured will be presented to capture the extent of the diversity that has been observed in the study population.

Extreme cases may be selected during two different phases of case study research: (1) when determining the method of data collection and selecting the cases from a larger pool of possible cases available for study, and (2) after doing the preliminary analysis of a larger pool of cases and deciding to concentrate on a more in-depth analysis of one or more extreme or deviant cases, which typically leads to reporting these cases as a separate part of the study or as an independent substudy.

Application

The main purpose of selecting extreme cases in the data gathering phase is to find a starting point for the study, and to determine the methods of measurement and data analysis. The researcher will select cases that demonstrate the most obvious differences from the majority of the sample, in terms of their available background data, that is relevant to the study questions. He or she will then inquire further, and in a very detailed way, about these participants in order to develop a holistic picture of these exceptional cases. A very common way of making decisions about which cases to select is to consider what previous data (e.g., test results) that have been acquired from the study population suggest about which cases will prove most informative. For example, in a study concerning student characteristics that are related to dropping out of high school, by Christine Christle, Kristine Jolivette, and C. Michael Nelson, the schools selected for closer examination were selected based on the lowest and highest rates of student attrition reported during previous phases of a larger study.

It is important to draw upon theoretical assumptions of which individual traits (e.g., gender, socioeconomic status) may influence variation in the research phenomena under question. In some studies, it is relevant to sample extreme cases from both males and females in a sample. Similarly, when comparing different countries, extreme cases may need to be selected from populations that experience the same kinds of economic circumstances (e.g., the same level of industrialization) in order to draw meaningful comparisons between them.

...