The Validity of Research Studies

Research is presently divided by a predominant concern with phenomena and effects on one side and with theory and explanation on the other side. This divide has fostered acceptance of the idea that research can be valid in different ways, either by an emphasis on inferences about variable-to-variable relationships or by an emphasis on causal explanations featuring constructs. We propose a way of bridging this divide by introducing a unifying framework—a nomological network of variables and constructs employing the principle of inference to the best explanation (IBE). This framework, called Nomological Network IBE, is developed, and linked to previous conceptual thought about theory and effects. We also offer a Bayesian interpretation to make the logic of Nomological Network IBE formally clear. We then explore the consequences of our proposal for (1) for specific best research practices, (2) the role of replication, and (3) research application.