14. Doctoral Research in Strategic Management

Introduction and Context

Strategic management research examines how organizations create, sustain, and transform value under conditions of uncertainty, interdependence, competition, and institutional change. At doctoral level, the purpose of research is not merely to describe managerial practice or apply familiar frameworks. It is to produce defensible knowledge: knowledge that clarifies an important problem, extends or challenges theory, uses evidence appropriately, and withstands critical examination.

Doctoral research in strategic management is difficult because its subject matter is multilevel and historically situated. Organizational outcomes may reflect the decisions of executives, the capabilities of teams, industry structures, institutional pressures, technological change, and events that cannot easily be controlled by the researcher. Moreover, strategy research frequently involves reciprocal relationships. For example, capabilities may influence performance, but successful performance may also enable firms to invest in capabilities. Establishing what causes what therefore requires more than observing statistical association.

A strong doctoral study begins with an important puzzle rather than with a fashionable method. The researcher must connect a substantive phenomenon to a theoretical conversation, formulate a question that can be investigated, select a design appropriate to the question, and make claims proportional to the evidence. This chapter develops that logic and evaluates major approaches used in advanced strategic management research.

Learning Objectives

By the end of this chapter, the reader should be able to:

1. Formulate significant and researchable questions in strategic management.

2. Construct theoretical arguments involving mechanisms, boundary conditions, and levels of analysis.

3. Conduct and critically evaluate a systematic literature review.

4. Compare quantitative, qualitative, mixed-method, longitudinal, comparative, and case-study designs.

5. Distinguish association from causality and assess validity, reliability, replication, and generalizability.

6. Identify ethical, governance, and data-management responsibilities in doctoral research.

7. Evaluate the use of digital data, computational methods, analytics, and artificial intelligence in strategy research.

From a Managerial Problem to a Researchable Question

A managerial problem is not automatically a research question. “Why is our digital transformation failing?” is a legitimate organizational concern but is too broad for a doctoral investigation. A research question specifies a phenomenon, a relationship or process, a context, and an intellectual purpose. It may ask how a phenomenon unfolds, why an effect occurs, when a relationship is stronger or weaker, or under what conditions an established explanation fails.

Good questions often emerge from a tension. A theory may predict that a resource produces advantage, while empirical studies show inconsistent outcomes. Two theoretical perspectives may explain the same phenomenon differently. A practice may be widespread although its consequences are poorly understood. A construct may be used extensively but measured inadequately. Such tensions create opportunities for contribution.

A useful question-development sequence is:

• Identify a consequential phenomenon or anomaly.
• Map the relevant literature and its assumptions.
• Locate an unresolved theoretical, empirical, methodological, or practical gap.
• Specify the unit and level of analysis.
• Define the intended contribution and feasible evidence.
• Test whether the question is answerable within available time, access, data, and ethical constraints.

A research question should not be judged only by novelty. Significance also involves explanatory value, practical importance, methodological rigor, and the ability to alter how scholars understand a phenomenon. A narrow study can make a substantial contribution if it identifies a previously unrecognized mechanism or boundary condition.

Theoretical Argument and Research Contribution

Theory is more than a list of variables. A theoretical argument explains why particular conditions generate particular outcomes. It identifies constructs, specifies relationships, proposes causal or process mechanisms, and states assumptions and boundary conditions. A conceptual model may show that dynamic capability affects innovation performance through organizational learning, with environmental turbulence moderating the relationship. The model becomes theoretically meaningful only when the researcher explains why learning is the relevant mechanism and why turbulence changes its operation.

Several theoretical traditions are especially influential in strategic management. The resource-based view explains advantage through valuable, rare, difficult-to-imitate, and organizationally embedded resources. Dynamic capability theory focuses on the capacity to sense, seize, and transform in changing environments. Transaction cost economics emphasizes governance choices under bounded rationality and opportunism. Institutional theory examines how organizations respond to regulative, normative, and cultural-cognitive pressures. Agency theory analyzes conflicts arising from divergent interests and information asymmetry. The knowledge-based view treats knowledge integration and transfer as central sources of organizational performance. These perspectives can complement one another, but combining theories without explaining their distinct roles produces theoretical ambiguity.

A doctoral contribution may involve extending a theory to a new context, specifying a mechanism, identifying a boundary condition, reconciling competing predictions, challenging an assumption, developing a new construct, or explaining a process that existing variance-based models overlook. Contributions should be stated precisely. “This study adds to the literature on innovation” is weak; “this study demonstrates that interorganizational trust improves innovation only when knowledge is codified sufficiently to permit transfer but not so extensively that experimentation is constrained” is more informative.

Theories operate at different levels. Individual cognition, team routines, organizational capabilities, interorganizational networks, industries, and institutions should not be treated as interchangeable. A cross-level argument must explain how constructs at one level influence outcomes at another. The ecological fallacy occurs when relationships observed at a higher level are assumed to apply to individuals; the atomistic fallacy occurs when individual-level observations are used to make organizational claims without justification.

Systematic Reviews as Research and Theory-Building

A systematic review is a transparent and reproducible process for identifying, evaluating, and synthesizing relevant studies. It differs from a selective narrative review because its search strategy, inclusion criteria, coding procedures, and analytical choices are explicitly documented. A systematic review may be conducted to aggregate evidence, assess the maturity of a field, identify conceptual fragmentation, or develop a new research agenda.

The process normally includes definition of the review question, development of search strings, selection of databases and sources, screening against inclusion and exclusion criteria, assessment of study quality, extraction and coding of evidence, synthesis, and reporting. The review protocol should specify the population, phenomenon, context, outcome, and study designs of interest. Citation counts may help identify influential work, but influence is not equivalent to methodological quality.

Quantitative meta-analysis estimates the distribution and magnitude of effects across studies. It requires attention to effect-size comparability, sampling error, measurement differences, publication bias, heterogeneity, and dependence among observations. A statistically significant average effect may conceal substantial variation across contexts. Qualitative evidence synthesis, thematic synthesis, and bibliometric or science-mapping techniques are valuable when constructs and methods are heterogeneous. Such techniques reveal patterns but do not substitute for theoretical interpretation.

Common review failures include searching only one database, excluding unpublished or null-result studies without justification, changing criteria after seeing results, treating all studies as equally credible, and confusing frequency of publication with theoretical importance. A high-quality review makes its decisions auditable and explains what the evidence can and cannot establish.

Choosing a Research Design

Research design is the logic connecting a question to evidence and inference. No design is universally superior. The appropriate design depends on whether the question concerns prevalence, causal effects, meanings, processes, configurations, change over time, or comparison across contexts.

Quantitative Designs

Quantitative research is appropriate when constructs can be operationalized and relationships can be examined systematically across cases. Cross-sectional surveys can estimate associations and test models, but temporal ambiguity and common-method bias limit causal interpretation. Panel and longitudinal designs observe units repeatedly and can examine change, lagged effects, path dependence, and reciprocal relationships. Archival research may use financial, patent, alliance, employment, or textual data, although archival availability does not guarantee construct validity.

Experimental and quasi-experimental designs provide stronger leverage on causality. Randomized experiments establish treatment assignment independently of potential outcomes, but organizational experiments may be difficult to implement. Natural experiments, difference-in-differences, regression discontinuity, instrumental variables, and matched comparisons can strengthen causal inference when randomization is unavailable. Each depends on assumptions that must be defended rather than merely named.

Measurement requires particular care. Reliability concerns consistency, while validity concerns whether a measure represents the intended construct. Content validity asks whether important dimensions are included. Convergent and discriminant validity assess relationships among measures. Measurement invariance is necessary when comparing groups, countries, or time periods; otherwise observed differences may reflect changes in interpretation rather than changes in the construct.

Qualitative and Case-Based Designs

Qualitative research is especially valuable for underexplored phenomena, process questions, complex mechanisms, and situations in which meanings and interpretations shape action. Interviews, observations, documents, digital traces, and field notes can be combined through triangulation. A case may be selected because it is typical, extreme, revelatory, critical, or theoretically useful. Case research does not become rigorous simply by presenting a rich narrative. It requires a transparent case boundary, systematic data collection, coding, comparison of evidence, attention to negative cases, and a clear chain of inference.

Process research commonly examines how sequences of actions, events, and interpretations produce outcomes. Temporal bracketing, event analysis, and process tracing can help distinguish stages and mechanisms. Researchers must avoid retrospective rationalization, in which participants construct coherent explanations after the fact. Contradictory accounts and documentary evidence are analytically valuable rather than inconvenient.

Comparative case research examines similarities and differences across organizations, industries, regions, or institutional settings. The logic may be most-similar systems, most-different systems, or configurational comparison. Generalization is usually analytical rather than statistical: findings refine concepts and theoretical explanations rather than estimate population parameters.

Mixed-Method and Comparative Designs

Mixed-method research integrates quantitative and qualitative evidence within a single inquiry. An exploratory sequential design may use qualitative findings to develop constructs or measures, followed by quantitative testing. An explanatory sequential design may use qualitative investigation to explain unexpected statistical patterns. Convergent designs collect different forms of evidence in parallel and integrate them during interpretation.

Integration is the defining requirement. Merely placing a survey and interviews in the same dissertation does not create mixed-method research. The researcher must explain how the methods address complementary parts of the question, how findings are compared, and how contradictions are handled. Mixed methods can improve inference, but they require additional time, methodological competence, and disciplined integration.

Causality, Validity, and Replication

Causal reasoning asks whether a change in one condition produces a change in an outcome, through what mechanism, for whom, and under what circumstances. A correlation may arise from reverse causality, omitted variables, selection effects, measurement error, or common exposure to a third factor. A credible causal claim therefore requires a temporal ordering, a plausible mechanism, and evidence that alternative explanations have been addressed.

Internal validity concerns whether the observed relationship is credible for the studied units. External validity concerns transferability to other populations, settings, or times. Construct validity concerns the adequacy of conceptualization and measurement. Statistical conclusion validity concerns whether the analysis supports the stated inference, including issues of power, model specification, missing data, and inappropriate significance testing.

Longitudinal designs improve temporal reasoning but do not automatically establish causality. Fixed-effects models can control for time-invariant differences, while lagged models can represent temporal ordering; neither resolves all forms of endogeneity. Researchers should distinguish robustness checks from proof. A result that survives alternative specifications is more credible, but it remains dependent on assumptions and data quality.

Replication is central to cumulative science. Direct replication repeats a study as closely as possible; conceptual replication tests the underlying proposition using different measures, samples, or contexts. Strategic management presents special replication challenges because firms, industries, technologies, and institutions change. This makes transparent documentation, data and code sharing where permitted, preregistration when suitable, and precise reporting particularly important. Null findings and failed replications can reveal boundary conditions and prevent overgeneralization.

Research Ethics and Governance

Strategic management research often involves commercially sensitive information, employee data, executive interviews, and identifiable organizational records. Ethical review should address informed consent, voluntariness, confidentiality, data minimization, secure storage, access controls, retention periods, and the management of conflicts of interest. Participants may face professional risks if their statements are disclosed, even when names are removed. Small samples and distinctive organizational descriptions can make anonymity difficult; this limitation must be communicated honestly.

Researchers have obligations to organizations as well as individual participants. Access agreements should not permit sponsors to suppress unfavorable findings or control interpretation. Researchers must disclose funding, relationships, and analytical decisions that could influence conclusions. Fabrication, falsification, plagiarism, undisclosed duplicate publication, selective reporting, and inappropriate authorship are serious breaches of research integrity.

Digital and AI-enabled research introduces additional concerns. Automated text analysis, web scraping, machine learning, and generative AI can process large volumes of data, but researchers must examine data provenance, consent, algorithmic bias, reproducibility, privacy, and model hallucination. AI-generated coding or summaries require human verification. A method is not ethically acceptable merely because data are publicly accessible.

Digital Methods and Emerging Directions

Digital technologies have expanded the empirical possibilities of strategy research. Platform data can illuminate network formation, ecosystem dependence, and user engagement. Natural-language processing can analyze strategic disclosures, earnings calls, patents, and public communications. Machine learning can improve prediction and discover patterns, while causal machine learning may help estimate heterogeneous treatment effects.

Prediction and explanation should not be confused. A highly predictive model may provide little insight into why a relationship exists, and an interpretable theoretical model may predict less accurately while offering stronger explanation. Researchers should define the role of computation in the research design, preserve data and code workflows, test model sensitivity, and avoid presenting algorithmic output as theory.

Emerging research increasingly addresses grand challenges, sustainable transitions, geopolitical disruption, artificial intelligence, ecosystem governance, and responsible innovation. These topics invite multilevel and interdisciplinary designs. They also heighten the need to examine distributional consequences: a strategy may improve firm performance while imposing costs on workers, communities, suppliers, or the environment.

Illustrative Case: A Mixed-Method Study of Platform Dependence

Background

The following fictional case illustrates how a doctoral researcher might investigate strategic dependence on a dominant digital platform. Meridian Home, a medium-sized manufacturer, sells through its own website and through a large online marketplace. Marketplace sales grow rapidly, but the platform periodically changes search algorithms, commission rates, and data-access rules.

Situation and Research Problem

Managers disagree about whether marketplace participation strengthens or weakens long-term strategic autonomy. A simple comparison of sales before and after joining the platform cannot determine the effect because the firm may have joined precisely when demand was increasing. The researcher therefore formulates the question: how do platform governance changes alter the relationship between marketplace dependence and a supplier’s innovation and performance outcomes?

Analysis and Design

The study uses three stages. First, a longitudinal archival analysis tracks marketplace dependence, product launches, margins, and platform policy changes over six years across a sample of suppliers. Second, a comparative case study examines Meridian Home and two firms with different levels of dependence. Interviews with executives, product managers, and channel partners are combined with platform documents and internal planning records. Third, the findings are integrated to examine whether data access and bargaining power constitute mechanisms linking dependence to innovation outcomes.

The design addresses several threats. Firms are matched on size and product category where possible; alternative measures of dependence are tested; policy changes are analyzed as time-specific events; and interview claims are compared with documents and performance records. The researcher does not claim that platform participation alone caused every observed outcome. Instead, the evidence supports a conditional explanation: dependence can accelerate market access while weakening experimentation when governance changes reduce visibility into customers and increase the cost of channel adaptation.

Outcome and Learning Points

The case yields an analytically useful distinction between access value and autonomy value. Platform participation creates access value through reach and demand generation, but autonomy value depends on the supplier’s ability to retain customer knowledge, develop alternative channels, and negotiate governance changes. The study contributes by identifying data access and channel redundancy as mechanisms that condition the consequences of dependence.

The case demonstrates that a significant research question may require several forms of evidence. It also shows why causal language must be calibrated, why process evidence complements statistical patterns, and why organizational outcomes should be interpreted within changing institutional and technological contexts.

Common Problems in Doctoral Strategy Research

Several recurring errors weaken dissertations. Researchers sometimes begin with a method rather than a question, select variables because they are readily available, or claim novelty merely because a study uses a new sample. Others construct overly complex models with weak theoretical links, treat control variables as automatic solutions to endogeneity, or infer organizational mechanisms from individual-level data.

Additional problems include inadequate attention to missing data, post hoc hypothesis generation presented as prior prediction, insufficient consideration of rival explanations, and failure to distinguish statistical significance from substantive importance. Qualitative studies may suffer from unsystematic sampling, unclear coding, or excessive reliance on illustrative quotations. Mixed-method studies may present parallel findings without integration.

Best practice is disciplined alignment. The research question, theory, design, measures, analysis, and claims should support one another. Researchers should document decisions, search for disconfirming evidence, report limitations without defensiveness, and explain what future evidence could change their conclusions. Doctoral quality is demonstrated not by claiming certainty but by making inference transparent and intellectually accountable.

Chapter Summary

Doctoral research in strategic management converts important organizational puzzles into theoretically informed and empirically defensible inquiries. Strong questions identify a phenomenon, an unresolved issue, a context, and a feasible basis for inference. Theoretical arguments explain mechanisms, boundary conditions, assumptions, and levels of analysis rather than merely listing variables.

Systematic reviews provide transparent methods for synthesizing evidence and developing theory. Quantitative, qualitative, mixed-method, longitudinal, comparative, and case-study designs answer different kinds of questions and carry different assumptions. Causal inference requires attention to temporal order, mechanisms, rival explanations, endogeneity, and validity. Replication, transparent reporting, ethical governance, and responsible use of digital and AI-based methods are essential to cumulative knowledge. The central principle is alignment: claims must remain proportionate to the design and evidence that support them.

Key Terms

Research question: A focused, answerable inquiry concerning a phenomenon, relationship, process, or condition.

Theoretical contribution: A substantive extension, refinement, challenge, or clarification of existing theory.

Mechanism: The process through which a condition produces or influences an outcome.

Boundary condition: A circumstance that limits when or for whom a theoretical relationship applies.

Level of analysis: The organizational or social level at which a construct is defined and examined.

Systematic review: A transparent, documented process for identifying and synthesizing relevant research.

Meta-analysis: Statistical aggregation and analysis of effect estimates from multiple studies.

Construct validity: The degree to which an operational measure represents the intended theoretical construct.

Internal validity: Credibility of causal inference for the units and setting studied.

External validity: The extent to which findings transfer to other settings, populations, or times.

Endogeneity: A condition in which an explanatory variable is related to unobserved influences on the outcome.

Longitudinal design: Research that observes phenomena or units across multiple time periods.

Case study: An intensive investigation of a bounded contemporary phenomenon in context.

Triangulation: Use of multiple data sources, methods, researchers, or theories to strengthen interpretation.

Mixed-method research: Intentional integration of quantitative and qualitative approaches in one inquiry.

Replication: Repetition of a study or theoretical test to assess the stability of findings.

Measurement invariance: Evidence that a construct has comparable meaning across groups or time.

Research integrity: Honest, transparent, responsible conduct of research and reporting.

Analytical generalization: Extension or refinement of theory rather than statistical projection to a population.

Causal inference: Reasoned judgment about whether and how one condition produces an outcome.

Review Questions

Short Answer Questions

1. What distinguishes a managerial problem from a doctoral research question?

2. Why are mechanisms and boundary conditions important in a theoretical argument?

3. What is the purpose of a systematic review protocol?

4. Distinguish internal validity from external validity.

5. Why does mixed-method research require integration rather than simple coexistence of methods?

Long Answer Questions

1. Explain how a researcher can move from a strategic management puzzle to a significant doctoral research question.

2. Compare quantitative, qualitative, and case-study designs in relation to their assumptions, strengths, and limitations.

3. Discuss the principal threats to causal inference in organizational research and the methods used to address them.

4. Explain how systematic reviews can contribute to theory development as well as evidence aggregation.

5. Evaluate the ethical and governance issues involved in using organizational, employee, and digital-platform data.

Analytical / Application Questions

1. A researcher finds that firms with greater digital capability have higher innovation performance in a cross-sectional survey. Develop three rival explanations and propose a design for assessing causality.

2. Design a mixed-method study examining why an alliance succeeds in one institutional environment but fails in another. Explain the sequence and integration of methods.

3. Using the Meridian Home case, identify the likely mechanisms, boundary conditions, and threats to validity in the proposed explanation of platform dependence.

22. Multiple Choice Questions with Answers

Question 1

Which feature most clearly distinguishes a doctoral research question from a broad managerial problem?

A. It focuses exclusively on financial performance.

B. It specifies a phenomenon, context, intellectual issue, and feasible basis for evidence.

C. It uses advanced statistical software.

D. It is approved by an organizational sponsor.

Correct Answer: B. It specifies a phenomenon, context, intellectual issue, and feasible basis for evidence.

Question 2

A theoretical argument is strongest when it primarily explains:

A. How many variables appear in a conceptual model.

B. Why a relationship occurs, through what mechanism, and under what conditions.

C. Why a study uses a large sample.

D. Why a topic is popular among managers.

Correct Answer: B. Why a relationship occurs, through what mechanism, and under what conditions.

Question 3

What is a central purpose of a systematic review protocol?

A. To guarantee statistically significant findings.

B. To replace assessment of study quality.

C. To make search, selection, and synthesis decisions transparent before or during the review.

D. To restrict the review to highly cited articles.

Correct Answer: C. To make search, selection, and synthesis decisions transparent before or during the review.

Question 4

Which design is most directly suited to examining organizational change and temporal ordering?

A. A single cross-sectional survey.

B. A longitudinal panel design.

C. A one-time expert interview.

D. A citation-count analysis alone.

Correct Answer: B. A longitudinal panel design.

Question 5

A relationship between capability and performance may be endogenous because:

A. The variables are always measured qualitatively.

B. Performance may also influence capability investment or both may be affected by an omitted factor.

C. The sample contains more than one industry.

D. The study reports confidence intervals.

Correct Answer: B. Performance may also influence capability investment or both may be affected by an omitted factor.

Question 6

In qualitative case research, analytical generalization refers to:

A. Estimating the exact prevalence of a finding in a population.

B. Repeating interviews until every participant agrees.

C. Extending or refining theoretical explanations from the case.

D. Selecting only the largest organizations.

Correct Answer: C. Extending or refining theoretical explanations from the case.

Question 7

Why is measurement invariance important in comparative research?

A. It ensures that statistical significance is always achieved.

B. It helps establish that a construct has comparable meaning across groups or time.

C. It eliminates the need for sampling.

D. It converts qualitative data into quantitative data.

Correct Answer: B. It helps establish that a construct has comparable meaning across groups or time.

Question 8

Which statement best describes replication in strategic management?

A. Only exact repetition is legitimate, regardless of context.

B. Replication can be direct or conceptual and helps assess stability and boundary conditions.

C. Replication is unnecessary when a study is published in a leading journal.

D. Replication means adding more control variables to the original model.

Correct Answer: B. Replication can be direct or conceptual and helps assess stability and boundary conditions.

Question 9

In the Meridian Home case, which finding represents a boundary condition?

A. Marketplace participation exists.

B. The firm sells products online.

C. The benefits of platform participation depend on data access and channel redundancy.

D. The firm has product managers.

Correct Answer: C. The benefits of platform participation depend on data access and channel redundancy.

Question 10

Which practice is most consistent with responsible use of AI in strategy research?

A. Treating generated coding as automatically valid.

B. Using publicly accessible data without considering privacy.

C. Documenting data provenance, checking outputs, and preserving human verification.

D. Replacing theoretical reasoning with predictive accuracy.

Correct Answer: C. Documenting data provenance, checking outputs, and preserving human verification.