A form of cherry picking where the initial hypothesis is replaced by one suggested by a pattern in the data. Although this may be less than ideal it is pragmatically necessary in some contexts. However, where a post-hoc hypothesis is used its use should be openly acknowledged and any potential impact on results should be assessed.
Defined on page 82
Used on pages 82, 83, 88, 92, 100
Also known as post-hoc hypotheses, post-hoc hypothesis testing