This course begins with the premise [paraphrasing Donald Campbell] that the fundamental goal of methodology is to rule out plausible rival hypotheses that make our research findings ambiguous and tentative.  Accordingly, this seminar explores the development and application of qualitative research designs and methods for the analysis of small-N studies. We develop practical tools for improving validity, reliability, and inference in research where the number of observations, cases, subjects, etc. is small compared to that commonly encountered in quantitative studies.

This course focuses on methods for ensuring that collected data allow the researcher to answer the questions posed by the research agenda, where mistakes in data collection can lead to false inferences in data analysis.

The goal of the course is two-fold: (1) to enable students to evaluate and critique studies employing qualitative methods and (2) to provide students with the skills to create rigorous qualitative designs to guide their own research.

The seminar examines substantive examples from American politics, public policy, comparative politics, and international relations.

SEMINAR REQUIREMENTS:

1. The course will be structured as a seminar. This places a substantial burden on students to come to the sessions prepared to discuss the readings and to actually discuss the topics for the day.   All readings assignments relevant to a given week's class discussion must be read prior to that class.

Active and creative participation in class discussion is an essential part of the seminar. Students will be responsible for the assigned readings, for taking part in class discussions, and for leading the class discussion. (25%)

2. Students will write five three-page analytic papers that apply an assigned set of analytic tools to a recently published of political science (assigned by the instructors). (75%)

The required course readings will come from two basic sources. First, four textbooks should be purchased at the MIT COOP:

The web-based syllabus for this course contains hyperlinks to all the other readings for the course (i.e., those not in the four books).  Some materials are linked from the syllabus directly through the MIT library to the appropriate web source.  You will need an MIT certificate to access these materials.

Additional reading materials will be found on the course STELLAR website.  This web site hosts the courses electronic reserve materials.  [For basic access go to: http://stellar.mit.edu/S/course/17/sp05/17.878/index.html.

We begin with a review of the goals of social research: description, descriptive inference, hypothesis testing, prediction, theory development, and causal inference. What is it we are trying to do in social research?  What do we want our research projects to accomplish?

Next we review the basic concepts and terminology of social research. This discussion provides a common language for the remainder of the course.

Required Reading:

Recommended Reading

In trying to understand the range of uses of case study research we start with a look at the traditional comparative case method in the social sciences -- a "variable-oriented" approach to analysis. The analytic logic here is derived directly from quantitative (statistical) analysis and is often described as some variant of "Mill's method."

Next we contrast this with the "case-oriented" approach, where the case itself, rather than the variables, is the focus of the research effort.  What does case-oriented research aspire to teach us?

Required Reading:

Defects in Research Design as Plausible Rival Explanations

The purpose of this course is to improve the credibility and rigor of qualitative analyses.  This begs the question: What undermines the credibility and rigor of qualitative analyses?  We examine a framework devised for experimental and quasi-experimental quantitative research and adapt it to qualitative -- case-based -- work.

Before we begin to collect data how can we devise better research designs to reduce the number of plausible rival explanations (threats to validity) confronting our findings?

Threats to Validity in Qualitative Data Collection

To understand better how plausible rival explanations (threats to validity) arise in qualitative research we review the main forms of qualitative data: interviews, focus groups, observation, documents & archives, secondary sources, and unobtrusive measures.

We devise a checklist of most likely threats to validity by data-type and devise data collection strategies to nullify those threats.

Strategies for Purposeful Case Selection

We look at options for purposeful case selection and consider how each raises or removes the threats posed by plausible rival explanations. We ponder some of the more troublesome practices in qualitative research: For example, can we select cases on a single value of the dependent variable and still hope to have anything credible to report?

We also consider the basic issue of "bias" in the analysis that may result from case selection. The first charge is simple: How do we chose cases in a manner that insulates we from the accusation that we intentionally chose our cases to fit the result we wanted?

The second charge is more complex:  How  do we prevent unintentional forms of "selection bias" in case selection?

Required Reading:

Here we learn to devise the study hypothesis and its corresponding "null" hypothesis.  This sets the stage for applying the most robust analytic techniques to the collected data.

In this sessions we consider two of the three basic strategies for learning from "within case analysis:"  The first is "pattern matching:" delineating all the observable implications of our argument.  This amounts to proliferating non-equivalent dependent variables. [Yin (1994; 106)].

The second strategy involves parsing the case into a larger number of "observations." This may be accomplished by (1) moving the "unit of analysis" to a lower level (e.g., from state to county) where the research question permits it. Or, (2) we could sub-divide the case into sequential time units (e.g., U.S. national security policy pre-9/11 and post-9/11) and employ the qualitative equivalent of time-series analysis. Or  (3) we could attempt both.  The resulting observations  might be called a nested-cases or embedded cases.

When is it appropriate to treat a case study as more than a single "observation?"

The third strategy, process tracing, we consider later in the course.

A mechanism is the systematic means by which A produces a change in B.  Mechanisms are the basis for inferring causation. Lacking a plausible mechanism all correlational findings must be considered spurious.   We explore the concept of mechanisms in social research.

Choose a narrative account of an important historical event in your area of research. Devise a design that employs "mechanisms" to provide additional observable hypotheses for analysis. Describe your design in a three page paper. Due April 30 in class

In this session we develop a practical understanding of, and method for, implementing that great icon of case study work:  "process-tracing."  Reflexively (and defensively) invoked by case study researchers whenever they are asked about their methodology, there remains no practical outline of this alleged method.

We will attempt to give it form and substance.  Defining a process as the sequential concatenation of mechanisms that link causal variables the "new" process tracing is a series of steps by which component mechanisms are operationalized,  their links specified, and observable traces investigated.

Mill's method, classical comparative methods, and the focused-comparison method are each variable-based approaches for analyzing multiple case studies.

Grounded theory and analytic induction (iterative analytic induction) are two case-based approaches for analyzing multiple case studies.

What are the methodological strengths and weaknesses of these different approaches?  Which plausible rival explanations (threats to validity) do these methods raise or negate?

Improving Rigor and Credibility

In this final session we review a final series of methods that are applicable to all research projects for further improving the rigor and credibility of qualitative studies.  We look at triangulation strategies for data, investigators, theories, and methods.  We discuss requirements for procedural auditing -- e.g., documentation of interviews, and maintaining a research journal.

Required Reading: