====== Publication bias ======

📌 Definition:

Publication bias refers to the tendency for studies with positive, statistically significant, or novel findings to be more likely published than those with negative, null, or inconclusive results.

🧠 Why does it happen?

Journals often prefer to publish “exciting” or positive findings.

[[Author]]s may feel discouraged from submitting studies with negative or null results.

[[Editor]]s and [[reviewer]]s may reject papers lacking significant findings.

Industry [[sponsor]]s might suppress unfavorable outcomes due to conflicts of interest.

🧪 Scientific consequences:

[[Overestimation]] of treatment effects in the literature.

Skewed evidence base used for clinical guidelines and decision-making.

Undermines the validity of meta-analyses and systematic reviews.

Reduces [[transparency]] and reproducibility in science.

🔬 Common example:

Suppose 10 studies are conducted on a new drug:

3 show significant benefit and get published.

7 show no benefit and remain unpublished. A meta-analysis including only the 3 published studies will overstate the drug’s effectiveness.

🧰 Tools to detect/prevent publication bias:

Pre-registration of [[clinical trial]]s (e.g., ClinicalTrials.gov).

Funnel plots to detect asymmetry in meta-analyses.

Reporting guidelines (PRISMA, CONSORT).

Encouraging journals to publish negative or null findings.

Open science initiatives and data sharing.