Peer Review
Peer review is the process in which other qualified scientists check a study for errors and sound methods before it is published.
Definition
Peer review is when other scientists read your work and check it before it is shared, looking for mistakes and asking hard questions. Formally, it is the evaluation of a report by independent experts who judge whether the methods were sound, the data support the conclusions, the analysis is correct, and the work fits with or fairly challenges existing knowledge. Reviewers assess the validity of the design, the adequacy of controls and statistics, and the reproducibility of the methods, which reduces errors and unsupported claims before publication. It is a quality filter rather than a guarantee of truth, since reviewers do not replicate the work and can miss fraud or subtle mistakes; replication by independent groups remains the ultimate test.
Example
Before a lab group presents its plant results, another group reads the report and notices that only two plants per condition were measured, which is not enough to be sure. A reviewer of a ramp report asks why the dots were connected on a graph with only three angles, points out that the $30^\circ$ range of $12\ \text{cm}$ is larger than the difference to the $20^\circ$ mean, and asks whether the cart was released the same way every time. A reviewer of a fertilizer study notes that no measure of spread was reported, that seedlings were not randomly assigned to trays, and that trays sat at different distances from the window, each a threat to internal validity that the authors must address.
Key Insight
Having someone find a mistake before you publish is a good thing; that is the review working. Peer review makes science self-correcting, because nobody has to trust one scientist's word when the claim must survive critics trying to find its weaknesses. It is a floor, not a ceiling: a reviewed paper has passed a check by a few experts, while a replicated finding has passed a check by nature.