Abstract
Data fabrication and falsification are clear breaches of research ethics, but have been shown to be insidious factors in various research disciplines. It would be naïve to believe that data fabrication and falsification do not affect the validity and reliability of business research. It behoves all users of such research to militate against these unethical practices and ensure that they do not go undetected. This paper briefly reviews the motivations for researchers, interviewers or surveyors to falsify or fabricate research data. This is followed by a discussion of techniques in the literature for detecting such unethical and fraudulent practices. Typically, these rely on the premise that falsification or fabrication of data results in anomalies in the dataset that cannot be attributed to sampling or methodology.
A number of business case studies are discussed involving subtle data anomalies that could be attributable to fabrication or falsification or data. It is demonstrated that tried and tested parametric or non‑parametric statistical tests are often more than sufficient to identify these anomalies that characterise bogus data. However, data fabrication and falsification are not necessarily self‑evident and it may therefore require an unconventional and innovative approach to determine the appropriate variables of analysis. Analysis of the phenomenon leads to the conclusion that data fabrication and falsification are most easily detected by carrying out analyses on apparently extraneous variables, as these would tend to be neglected by the errant interviewer or surveyor. This leads to proposing a generic approach to detecting bogus data and a corresponding protocol to militate against it.
A protocol is proposed that separates the essential research functions by adopting the trias politica principle, or separation of powers, analogous to the three branches of government: the legislature, the executive and the judiciary. The protocol requires the three functions of research design plus substantive analysis, data collection, and data verification to be separated. Suggestions for presenting data analyses and research findings that will ensure greater transparency, militate against data fabrication and falsification, improve reliability, and promote research integrity are included. The paper concludes with a specific recommendation to academics, consultants, reviewers, examiners, and other users of business research to hold researchers more accountable for their validity and reliability of their research outputs.