Prof. Dr. Larry AdamsAcademic, Author & Researcher
How to Conduct Quantitative Research · Chapter 9 of 33

Part I — FOUNDATIONS AND RESEARCH PLANNING

Chapter 5. Conceptual Framework and Hypothesis Development

5.1 Theory versus conceptual framework

A theory offers an explanatory account of relationships or mechanisms. A conceptual framework is the study-specific model that identifies the constructs, proposed links, boundary conditions and covariates relevant to the research questions. A diagram alone is not a framework: each arrow needs a theoretical and empirical rationale.

5.2 Building the framework

  1. Define every construct precisely and distinguish it from related concepts.
  2. Identify a theory or coherent set of theories that explains why a relationship should exist.
  3. Review evidence for each proposed link, including inconsistent findings.
  4. Specify the outcome, predictor(s), mediators, moderators and covariates only where conceptually justified.
  5. Show expected direction when theory supports it; do not add arrows merely to make the diagram look sophisticated.
  6. Write a short narrative explaining each relationship and its assumptions.
  7. Translate each relationship into a research question or hypothesis and an analysis plan.

5.3 Example framework

Illustrative model: perceived supervisor support (predictor) → work engagement (outcome). Organisational tenure and job type might be considered covariates if there is a theoretical reason and sufficient data. A mediator such as psychological safety or a moderator such as workload should be included only if the study has a defensible theory, valid measures, adequate sample size and a suitable analysis plan.

ElementMeaningKey caution
PredictorVariable used to explain or predict variation in an outcomeA predictor is not automatically a cause
OutcomePrimary variable the study seeks to explain or estimateDefine scale and scoring before analysis
MediatorProposed pathway through which a relationship may operateCross-sectional mediation has serious temporal limitations
ModeratorVariable that changes the strength or direction of a relationshipTest an interaction; explain conditional effects
CovariateVariable included for a defined analytical reasonDo not adjust indiscriminately; adjustment can introduce bias

5.4 Hypothesis writing

A directional hypothesis predicts the direction of a relationship (for example, higher perceived support is associated with higher engagement). A non-directional hypothesis predicts a relationship without specifying direction. The choice should follow theory and prior evidence, not convenience. A null hypothesis often states no association or no difference in the population, but reporting should focus on estimates and uncertainty rather than treating a p-value as proof that a null is true.

5.5 Causal diagrams and confounding

A causal diagram can help clarify assumptions about predictors, outcomes and confounders. A confounder is a common cause of the exposure/predictor and outcome. Including every available variable is not a solution: controlling for a mediator may remove part of the relationship of interest, while conditioning on a collider can create a spurious association. Covariate selection should be grounded in the causal question and domain knowledge.