Bla ordlisten:
Planned analyses are planned and specified in study protocols before starting data collection.
These contrast with unplanned (post-hoc) analyses or data fishing, which may be suggested by inspecting the data. Multiple statistical comparisons result in misleading findings resulting from the play of chance. This problem is made worse by reporting bias; i.e. only reporting findings that are “statistically significant”.
If you feel that this definition hasn't helped you to understand the term, click on our monkey to let us know.