Publication bias: which results made it into the visible record?

When we look only at the papers we can readily find, we can ask how that visible set was assembled. The question is not answered simply by counting the references in an article.

A fictional shelf of reports

Imagine six invented projects comparing ways of labeling craft-storage boxes. In this example, three projects report clear differences and three do not. Suppose only the projects reporting clear differences are written up for a public newsletter because of their results.

A reader who sees those three articles does not see the full six-project set. That deliberately constructed example illustrates selection into the visible record. It provides no evidence about how often this happens in ADHD research or any real field.

It also does not tell us what the six projects would show if their data were combined. We have not supplied measurements, methods or sample sizes. Filling in a numerical answer would be another unsupported step.

Missing is not automatically biased

A particular missing report may have an unknown status. If you cannot find a paper, do not assume its authors suppressed an unwanted result. The term in the definition concerns a relationship between results and publication; a broken link alone does not establish that relationship.

You can record the absence while leaving its reason open. That restraint matters when describing real researchers or organizations.

A visibility note for an existing review

When reading a review, look for:

• Where and when the authors searched.
• Which kinds of records they considered.
• Whether protocols or registrations are discussed.
• What they say about unpublished or unavailable results.
• Which limitations they identify in the visible evidence.

This is a reading aid, not a complete systematic-review procedure. An unanswered row should not become an accusation or an invented estimate of missing findings.

Different readers can contribute

An adult may want the original review behind a confident headline. A parent may notice that the article cites only favorable examples. A grandparent might help locate the search-date statement. A teen can ask whether all relevant projects were considered without having to adjudicate motives.

Selection of published studies is different from selecting participants for one study. It is also different from retracting an article after publication. The related pages keep those processes distinct.

This page does not conclude that a particular treatment literature is biased or that all published findings should be dismissed. Its practical contribution is a precise question: what is known about the route from completed work to the evidence this summary makes visible?

Source

[1] U.S. National Library of Medicine. Publication Bias — MeSH Descriptor D017594. U.S. National Library of Medicine. Descriptor last updated July 14, 2009; original publication day not asserted. Accessed September 6, 2026. Full scope note and descriptor history. Supports the adjacent attributed claim. All illustrative scenes, numbers and optional note formats are original inventions, not source findings.

Related guides

• systematic review
• preregistration
• research retraction
• selection bias