Abstract

The exposure of scientific scandals and the increase of dubious research practices have generated a stream of studies on Questionable Research Practices (QRPs), such as failure to acknowledge co-authors, selective presentation of findings, or removal of data not supporting desired outcomes. In contrast to high-profile fraud cases, QRPs can be investigated using quantitative, survey-based methods. However, several design issues remain to be solved. This paper starts with a review of four problems in the QRP research: the problem of precision and prevalence, the problem of social desirability bias, the problem of incomplete coverage, and the problem of controversiality, sensitivity and missing responses. Various ways to handle these problems are discussed based on a case study of the design of a large, cross-field QRP survey in the social and medical sciences in Sweden. The paper describes the key steps in the design process, including technical and cognitive testing and repeated test versions to arrive at reliable survey items on the prevalence of QRPs and hypothesized associated factors in the organizational and normative environments. Partial solutions to the four problems are assessed, unresolved issues are discussed, and tradeoffs that resist simple solutions are articulated. The paper ends with a call for systematic comparisons of survey designs and item quality to build a much-needed cumulative knowledge trajectory in the field of integrity studies.