Statistical Significance

Wikis > Research > Statistics > Statistical Significance

P-Hacking and Other Statistical Sins

0.05 or 0.005? P-value Wars Continue

P Value Under Fire

Not Even Scientists Can Easily Explain P-values

P value ban: small step for a journal, giant leap for science

We’re using a common statistical test all wrong. Statisticians want to fix that.

Beware The P-Value

Psychology Journal Bans Significance Testing

Clinical and Statistical Significance

Statistically significant finding is a conclusion that there is evidence against the null hypothesis.

A ‘significant’ finding means that the difference was probably caused by something other than chance

A small difference between two large groups can be statistically significant but have little clinical meaning.
A large difference between two small groups can be clinically significant but not statistically significant.

When considering the significance reported in a study, consider:
• how likely is the significant difference due to chance (a type 1 error)
• if it was not due to chance, how likely was it due to flaws or biases in the study
• how likely was a type 2 error if the difference was not statistically significant
• if the differences are statistically different, are they also clinically significant

Sub Topics:

Comments are closed.