An abstract is a few hundred words chosen by the people whose result it describes. Reviewers see it, and it is the least constrained part of a paper, because it has room for conclusions and no room for the qualifications that support them.
An abstract is a few hundred words chosen by the people whose result it describes. Reviewers see it, and it is the least constrained part of a paper, because it has room for conclusions and no room for the qualifications that support them.
A large proportion of abstracts omit the number analysed, which bounds every claim that follows. Where it appears it is often the number enrolled rather than the number completing, and those diverge most in exactly the trials where it matters.
Studies comparing abstract conclusions with the results underneath find a consistent pattern of non-significant findings described in positive language. This is documented rather than alleged, and it is why the results section is worth the extra minute.
Abstracts tend to report relative changes, which sound larger. The absolute figures, which determine whether anything matters practically, generally appear only in the tables.
Space pressure falls hardest on adverse effects. A trial reporting harms thoroughly in its body may mention none in the abstract, and a reader stopping there has seen only one side of the ledger.
Work presented at a conference is often never published in full, and where it is, the final numbers frequently differ. Citing a conference abstract as a study overstates what exists.
Deciding whether to read the paper, and establishing what question was asked. Those are real uses, and they are not the same as establishing what was found.
Never cite a finding from an abstract alone where the full text is available. If the full text is not available, say that the abstract is the source.
Where an abstract must compress, adverse effects are the first casualty. A trial reporting harms thoroughly in its body may mention none in its abstract, and a reader stopping there has seen one side only.
Much conference material is never published in full, and where it is, the final numbers often differ. Citing a conference abstract as a study overstates what exists.
The claims on this page about the summary most people read is the part the authors had most freedom over can be checked against sources outside this site: