The Quality of Accounting Research and Internal, External, and Construct Validity

The “Mathews Committee” (Mathews, 1990) concluded that the research performance of the accounting discipline in Australia was weak. This paper is motivated by the need to improve the quality of accounting research in Australia and probably some other countries as well. It discusses three issues of crucial importance to research quality ‐ internal, external, and construct validity with the discussion illustrated with examples from existing accounting research. Internal validity is concerned with the ability to make causal statements from a piece of research. External validity is concerned with the ability to generalize from the research and construct validity is interested in the validity of the variables measured. Internal and external validity are discussed within the context of five common research designs. No one design is strong in both types of validity and tradeoffs between internal and external validity are necessary. When designing a new research project, the particular tradeoffs made should depend on the research objectives. The paper discusses how these choices might be made and how, if internal validity problems exist, they might be minimized. With construct validity, no tradeoffs are appropriate ‐ the researcher should attempt to eliminate this type of invalidity. The paper discusses how this might be done.