🧪 This platform is in early beta. Features may change and you might encounter bugs. We appreciate your patience!
dissemination_bias
Dissemination bias is the umbrella term for all processes by which research findings are selectively made available based on the nature of their results. It encompasses publication bias, time-lag bias, location bias, citation bias, and language bias as specific mechanisms. The common thread is that the accessibility and visibility of research depend not on its quality or importance, but on whether its findings are positive, significant, novel, or aligned with powerful interests.
A pharmaceutical company conducts ten clinical trials of a new drug. Three show the drug is effective and are published in major journals, presented at conferences, and promoted in press releases. Seven show no effect and are filed away in regulatory archives, never published or discussed publicly. Doctors and patients see only the positive evidence.
A tech startup funds six usability studies on its new productivity app. The two studies showing users complete tasks faster are turned into white papers, featured in press releases, and presented at an industry summit. The four studies revealing user frustration and high error rates are quietly shelved and never shared beyond the internal team.
A government agency commissions eight independent evaluations of a new youth rehabilitation program. The two evaluations showing reduced reoffending rates are published on the agency's website and cited in a ministerial speech. The remaining six evaluations, which show negligible or mixed results, are classified as internal documents and never made publicly accessible.
Binary (yes/no) questions an LLM must answer to identify this aspect:
Are some research findings being selectively shared or publicized based on their results?
Type: binaryIs the dissemination of findings influenced by whether they support a particular position?
Type: binaryAre null or unfavorable findings being suppressed, delayed, or under-promoted?
Type: binaryDoes the publicly available evidence present a skewed picture because of selective dissemination?
Type: binaryDissemination bias is the umbrella term for all processes by which research findings are selectively made available based on the nature of their results. It encompasses publication bias, time-lag bias, location bias, citation bias, and language bias as specific mechanisms. The common thread is that the accessibility and visibility of research depend not on its quality or importance, but on whether its findings are positive, significant, novel, or aligned with powerful interests.
Multiple actors in the research ecosystem — authors, reviewers, editors, sponsors, and media — all have incentives that favor the spread of positive, significant, or novel findings. No single actor intends to create a biased evidence base, but the cumulative effect of many individually rational decisions produces systematic distortion.
Mandate registration and reporting of all studies, including results. Support open-access data repositories. Require publication of all completed trials as a condition of regulatory approval. Fund systematic reviews that actively seek unpublished evidence. Value and reward null results.
Recognized as a major threat to evidence-based medicine, policy, and science. Entire initiatives like AllTrials and ClinicalTrials.gov were created specifically to combat dissemination bias in clinical research.
Studies with statistically significant or positive results are more likely to be published, while null results remain unpublished. This distorts the published literature and inflates apparent effect sizes in meta-analyses.
Significant results are published faster, distorting the evidence base at any point in time.
Significant results appear in higher-impact journals, amplifying their visibility.
Studies with significant results are cited disproportionately more often.
Research funded by parties with financial interests tends to produce favorable results.
Splitting a single study into multiple publications to inflate publication count.
Use these tools to detect, analyze, or train this aspect.