![]() ![]() I'm interested to know if there are other reasons why it's bad beyond the higher possibility of spurious results and if there are ways, such as splitting data first, of changing a hypothesis post hoc but avoiding the increase in Type I errors. I appreciate the interest in my question, but the answers and comments are mostly aimed at what I thought I established as background information. I understand that one reason why changing a hypothesis to fit observed data is problematic is because of the greater chance of committing a type I error due to spurious data, but my question is: is that the only reason or are there other fundamental problems with going on a fishing expedition?Īs a bonus question, are there ways to go on fishing expeditions without exposing oneself to the potential pitfalls? For example, if you have enough data, could you generate hypotheses from half of the data and then use the other half to test them? update Many basic statistics books warn that hypotheses must be formed a priori and can not be changed after data collection otherwise the methodology becomes invalid. It is well known that researchers should spend time observing and exploring existing data and research before forming a hypothesis and then collecting data to test that hypothesis (referring to null-hypothesis significance testing). ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |