Use this URL to cite or link to this record in EThOS:
Title: Disclosing research findings : determinants and impact of incomplete and partial scientific knowledge in the context of evidence-based medicine
Author: Salandra, Rossella
ISNI:       0000 0004 7969 8543
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
This dissertation examines the disclosure of research findings in the context of evidence- based medicine. The determinants and impact of incomplete and partial scientific knowledge are explored in three interrelated studies. Study I (Chapter 4) explores the contextual factors associated with the selective reporting of research findings in clinical trials, defined as the publication of only part of the findings originally recorded during a research study, based on the results. I find that the chances of selective reporting are higher for industry-funded studies than for publicly-funded studies; yet, this effect is restricted to studies where at least one author is industry-affiliated. I also find that selective reporting is more likely in projects exploring radical innovation, compared to those investigating incremental innovation. Study II (Chapter 5) investigates the poor reporting of products' adverse effects in scientific publications, specifically focussing on the misreporting of drugs' adverse events in clinical trials. I find support to the hypothesis that negative relative performance at the level of the project increases the likelihood of misrepresentation of adverse effects in the focal project's final publication. The results also partially support my hypothesis on the influence of competition: in head-to-head competitions (i.e., trials where two drugs are confronting each other), the effect of negative performance on the misreporting of adverse effects is weaker for leaders. Study III (Chapter 6) aims at identifying the causal impact of bias in publication introduced by selective reporting. Using a matched control sample and diff-in-diff methodology, the study suggests that bias (detection) causes a significant decline in future citations. These findings offer support to the view that evidence appraisal systems (e.g., systematic reviews) are effective in signalling incomplete knowledge to the scientific community, in turn contributing to reducing avoidable research waste. The common thread connecting my studies is the investigation of the issues relating to the accuracy and truth of the research record in medical publication; the 'extreme case' of medical research speaks to broader challenges currently faced by scientists while they attempt to build a trustworthy knowledge-base, in publications and elsewhere.
Supervisor: Criscuolo, Paola ; Salter, Ammon Sponsor: Engineering and Physical Sciences Research Council
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral