Please enable JavaScript.
Coggle requires JavaScript to display documents.
CASE RESEARCH (CASE STUDY METHOD) - Coggle Diagram
CASE RESEARCH (CASE STUDY METHOD)
Nature of Case Research
Definition
Intensive study of a phenomenon over time in its natural setting
Usually in one or a few sites
Data Collection
Multiple sources: interviews, observations, documents, archival/secondary data
Outcomes
Rich, detailed, contextualised inferences
Epistemological Uses
Positivist: theory testing
Interpretive: theory building
Disciplinary Use
More popular in business/IS than other social sciences
Strengths of Case Research
Flexible Role in Theory
Can be used for theory building and theory testing
Constructs Need Not Be Predefined
In interpretive case research, constructs can emerge from data
Flexible Research Questions
Research questions can be modified during the study (interpretive)
Rich Contextual Understanding
Captures complex context, history, processes
Multi-Level Perspectives
Individual, group, organisational levels
Multiple participants’ perspectives
Weaknesses & Common Problems
Methodological Weaknesses
No experimental control → weak internal validity
Heavy dependence on researcher’s integrative skill
Context-heavy → limited generalisability
Five Common Problems (Benbasat et al., 1987)
Vague or absent research questions
Sites chosen by convenience/access, not fit
Lack of triangulation between data sources
Poor documentation of data collection & analysis procedures
Failure to use longitudinal potential (only cross-sectional view)
Distinction from Teaching Cases
Teaching cases (e.g., Harvard) = classroom discussion tools
Case research = systematic, scientific method for explanation
Key Design Decisions in Case Research
Is Case Method Appropriate?
Exploratory questions
Early-stage theory building (construct discovery)
Focus on participants’ experiences and context
Understanding complex, temporal “how/why” processes
Unit of Analysis
Individual-level (e.g., decision maker)
Group-level (e.g., team, department)
Organisational-level (e.g., firm, division)
Multi-level combinations for richer insight
Single-case vs Multiple-case Design
Single-case suitable when:
Unique/extreme case
Revelatory access
Critical/contrary test of theory1.4.3.2 Multiple-case suitable when:
Theory testing
Generalisability
Replication logic (Yin)
Site Selection (Theoretical Sampling)
Must fit research questions
Dissimilar sites to increase variance (for generalisation)
Use theoretical sampling, not opportunistic convenience
Example: mix of firm sizes to study tech implementation
Site Selection (Theoretical Sampling)
Must fit research questions
Dissimilar sites to increase variance (for generalisation)
Use theoretical sampling, not opportunistic convenience
Example: mix of firm sizes to study tech implementation
Data Collection Techniques & Triangulation
Interviews (open-ended, focused) – primary method
Direct observation (meetings, sessions)
Documents & archival records
Physical artefacts (tools, outputs)
Triangulate across sources & respondents
Conducting Case Research (Process/Roadmap)
Define Research Questions
Theoretically & practically interesting questions
Preliminary constructs to guide initial design
Positivist: theory-based constructs ex ante
Interpretive: avoid fixed prior theory; allow change
Select Case Sites (Theoretical Sampling)
Choose sites to replicate, extend, or contrast theory
Minimise noise (industry, size effects)
Maximise variance in dependent variables of interest
Access via executives; clarify purpose, benefits, confidentiality
Create Instruments & Protocols
Interview protocol = list of questions
Can mix structured and unstructured questions
Follow protocol consistently; neutral tone, no leading
Plan supplementary sources (docs, reports, observations)
Select Respondents
Across levels, departments, roles
Prefer random; snowball acceptable if diverse
Criteria: involvement, knowledge, willingness
Start Data Collection
Record interviews (with consent)
Take notes: key points, behaviours, impressions
Transcribe interviews verbatim
Within-case Data Analysis
Analyse each case separately
Identify emergent concepts and patterns
Possible techniques
Intuitive sense-making
Grounded theory coding (open, axial, selective)
Diagrams, sequences, network charts
Cross-case Analysis (for multi-site)
Compare across cases for similar patterns
Ignore idiosyncratic, purely contextual noise
Within-group vs between-group comparisons (e.g., high vs low performers)
Pairwise firm comparisons
Build & Test Hypotheses
Develop tentative hypotheses from patterns
Iterate between data and theory (refine constructs/links)
Compare with prior literature
Reconcile conflicting findings creatively
Stop at theoretical saturation (no new insights)
Write Case Research Report
Describe sampling, data collection, analysis steps clearly
Allow readers to judge strength & credibility of inferences
Interpretive Case Research Exemplar – Eisenhardt (1989)
Research Focus
How executives make fast strategic decisions in high-velocity environments (HVE)
Do fast decisions help or hurt performance in HVE?
Context – HVE Personal Computing Industry
Rapid change: tech, competition, demand
Information incomplete, obsolete, noisy
Design & Sampling
Multiple-case design (8 firms)
Personal computing firms in Silicon Valley
Replication logic across cases
Similar industry & location → reduce noise
Data Collection
nitial CEO interviews → select key strategic decisions
Divisional head interviews (long, open-ended)
Courtroom-style questioning (“what happened?” “when?”)
Two-person interview teams; 24-hour rule for field notes
Questionnaires (conflict, power distribution)
Secondary data (reports, demographics, performance)
Personal observation (strategy sessions, meetings)
Data Analysis
Quantitative: patterns in conflict and power
Decision climate profiles from qualitative traits
Within-case: decision stories and timelines
Cross-case: compare firms by decision speed & performance
nductive constructs & propositions; compare with literature
Synthesise into inductive theory for HVE decisions
Key Findings (Contradicting Prior Literature)
Fast decision-makers used more information, especially real-time
Considered more alternatives simultaneously
Used two-tiered decision process (counsellors → quick selection)
Did not avoid conflict; used better conflict resolution
Fast decisions associated with superior firm performance
Positivist Case Research Exemplar – Markus (1983)
Research Focus
User resistance to new financial information system (FIS) at GTC
Why HQ accountants supported FIS but divisional accountants resisted
Design
Single-case study (unique phenomenon)
Testing Predictions
System-determined:
Prediction: fixing technical issues reduces resistance
Observation: resistance persisted → theory rejected1.7.4.2 People-determined:
Prediction: job rotation/co-opting reduces resistance
Observation: resistance persisted or increased → theory rejected1.7.4.3 Interaction theory:
FIS centralised data, reduced divisional accountants’ power
Corporate could bypass divisions, raising HQ power
Resistance aligned with loss of power → theory supported
Role of Case Evidence
Logical testing of rival hypotheses using observed behaviour
Supports interaction theory without formal statistics
Comparisons with Traditional Positivist Research
Controlled Observations vs Natural Controls
Case research lacks experimental control
Use “natural controls” (e.g., same person in different roles)
Also used in natural sciences (astronomy, geology)
Controlled Deduction
Case research often qualitative, lacks formal statistics
Use theoretical predictions → test against observed behaviour
Logical (non-mathematical) tests also valid
Replicability
Sites are unique, but theories & predictions can be replicated in new cases
Generalisability
Built via multiple case studies in varied contexts
Consistent findings across cases → stronger generalisability
Popper’s Four Criteria of Scientific Theories
Falsifiable
Logically consistent
Predictive ability
Better explanation than rival theories
Improving Scientific Quality in Case Research
Increase degrees of freedom
More case sites
More rival explanations
More levels of analysis
Markus: multiple groups + three rival theories + best-fitting theory