Designing Case

Grupp 2
Jukka Mäki-Turja, Johan
Andersson, Joel Huselius

Case Studies from Chapter 1

A case study is an empirical inquiry that

Investigates a contemporary phenomenon within its real-life
context, especially when
the boundaries between phenomenon and context are not
cleraly evident

When to use a CS?

Many more variables of interest than data points
Relies of multiple sources of evidence
Benefits from prior theoretic propositions, guiding data collection
and analysis.
In answering ”how” and ”why” questions

Multiple case design  Conclusion and Advice  .Outline – Research Design  What is a Research Design?  The role of Theory Criteria for high quality research design  Single vs.

not a logistical problem!   Research design can be seen as a blueprint of research What question to study?  What data are relevant?  What data to collect?  How to analyze the results?   Case studies require its own research design  Not a special case of.g. experiment. e. ..What is a Research Design  Research Design is a difficult part of doing Case Studies   No roadmaps exists… Logical plan to go from A to B A = initial set of question to be answered  B = conclusions of study  Logical.

5 Components of Research Design Questions  Propositions  Unit of analysis  Linking data to propositions  Criteria for interpreting the findings  .

Questions and Propositions  Questions  The high level questions of the Case Study.  Propositions  Possible  (partial) answers ( propositions  State purpose instead .  Case studies suitable for ”how” and ”why” questions.a hypotheses) Directs attentions on what to examine in the study  More concrete than questions  Forces the study in the “right” direction  In exploratory studies .

one might be tempted to cover “everything”.Unit of Analysis  What is the ”case”?     Relates to research questions and proposition    An individual? A decision? A program? Without clear propositions. . Non-favoring research questions – too vague or too numerous Different units of analysis requires different research design and data collection strategy.

Linking data to propositions Least well developed  Pattern Matching   Identify effects/no effects patterns  Which pattern matches best? ”effects” pattern ”no effects” pattern Observation .

The criteria for Interpreting the findings How close does a match have do be in order to be considered a match?  No general solution…  Hope that patterns of rival propositions are sufficiently constrasting  .

Outline – Research Design What is a Research Design?  The role of Theory  Criteria for high quality  Single vs. Multiple case design  Conclusion and Advice  .

Statistical generalisation  Replication  .  Important to have a theoretical framework providing guidance   Existing work Analytical vs.The Role of Theory Covering these 5 aspects force you to begin constructing a preliminary theory.

Criteria for high quality Judging the quality of Research Design  Four tests   Construct Validity  Internal Validity  External Validity  Reliability .

Demonstrate that the selected measures of these changes do indeed reflect the specific type of change that have been selected. To meet Construct Validity. . 2. 1.Construct Validity    ”Establishing correct operational measures for the concepts being studied” Case studies are often criticized that subjective judgement is used collecting data. e. Select the specific type of changes that are to be studied.g.

whereby certain conditions are shown to lead to other conditions.  Inferring theory   Study x leads to y  What happens if unknown z affects y? .Internal Validity “Establishing a causal relationship. as distinguished from spurious relationships”  For explanatory or causal studies only.

External Validity ”Establishing the domain to which a studies findings can be generalized”  Critics state that single cases offer a poor basis for generalization.   Analytical  generalization rather than statistical Generalization by replication  Replication logic same as for experiments .

Reliability   ”Demonstrating that the operations of a study can be repeated with the same results” The goal of reliability is to minimize the errors and biases in a study.  Case  study protocols to document General approach: conduct research ”as if someone were always looking over your shoulder”  compare with accounting .

Tests Tactics Phase Construct Validity Multiple sources of evidence Establish chain of evidence Review draft CS report Data collection Data collection Composition Internal Validity Pattern matching Explanation building Adress rival explanations Use logic models Data analysis Data analysis Data analysis Data analysis External Validity Use theory in single-case studies Use replication logic in multiple case Research design Research design Reliability Use case study protocols Develop case study database Data collection Data collection .

Multiple case  Single case appropriate in certain conditions  Multiple case design better in general  Embedded vs. Holistic  Holistic = one unit of analysis  Emdedded = several units of analysis .Case Study Designs  Single vs.

Basic types of Designs Single-case Designs Context Holistic (single unit of analysis) Context Context Case Case Context Context Case Case Context Context Case Case Case Context Embedded (multiple units of analysis) Multiple-case Designs Case Embedded Unit of Analysis 1 Embedded Unit of Analysis 2 U1 U2 U1 U2 Context Context Case Case U1 U2 U1 U2 .

Critical case: clear set of propositions Extreme/unique case Representative/typical case Revelatory case  5. Longitudinal case   6. 4. 2.Single-case Design  Five rationales 1. 3. Previously inaccessible phenomena Same things at different points in time Assumes that conditions changes over time As a pilot case for multiple case studies  Not considered as a case study of its own .

loses higher level (holuistic) aspects.  study might be conducted on a too abstract level  Research question slippage  Embedded  Avoids design slippage  Extensive analysis  Might focus too much on subunits.Embedded vs. . Holistic Designs  Holistic  When design no logical subunits can be identified.

Multiple-case Designs More robust results and compelling arguments  Require more resources  Replication rather than ”sampling” logic  Each case can be holistic or embedded  .

Replication vs. Sampling logic  Replication – analytical generalization  Analogous to that used in multiple experiments  Goal is to duplicate results from previous work  Convergent evidence is saught  ”Sampling”– statistical  Analogous to that used in surveys  Goal is to gather general information from large amounts of data .

you can be accused of distorting or ignoring the discovery to accommodate your design.Literal vs. . Theoretical Replication  Literal replication   Theoretical replication    Contrasting results for predictable reasons If cases are contradictory initial proposition must be revised   Similar results Without redesign. A prerequisite of successful replication is a rich theoretical framework Number of cases is very fuzzy.

Rationale for a multiple case design Comes from understanding theoretical and literal replication  Simplest multiple case design   Literal  replication among two cases More complicated multiple case design  Theoretical replication between different types of conditions  Literal replication within each type of condition .

Conclusion and Advice  When you have a choice (and resources) choose multiple case design  Two cases is significatly better than a single one – allows for replication.  Drastical improvment of generalizability  Theoretical replication even stronger argument  Avoids critisism and skepticism  If you use single case  prepare to make an extremly strong argument in justifying your choice of case. .