I don't think so. I have read "tail probability", but I can't recall in which paper. I just scanned recent papers and found those two that might serve as examples:
http://www.ncbi.nlm.nih.gov/pubmed/26891189
http://www.ncbi.nlm.nih.gov/pubmed/26390921
"Probability of observing the dataset MORE EXTREME than the current one under the Null Hypothesis (H0)."
Thank you all.
I think it is called : "Observed significance level" .
From Algorithms to Z-Scores: Probabilistic and Statistical Modeling in Computer Science
Norm Matloff, University of California, Davis
Page no: 302
I think that "observed significance level" is a very bad name, because it seems to mix to incompatible philosophies of statistical tests. A particular "level" is defined for error-rates in hypothesis tests (that do not need any explicit calculation of p-values), whereas p-values are a statistic used in the context of significance tests (that do not control error-rates).
I would feel better if "level" would be dropped from the name: p-value = "observed [statistical] significance". However, I still find this an suboptimal name, because teh significance is condiered higher when the p-value gets lower, and the mapping between significance and p is non-linear. I would use the term "[statistical] significance" for -log(p), and from this it follows that p-value = "exp(-significance)" or "1/exp(significance)".
Because the relation between "significance" and "p" is complicated I still think a better name is just "tail probability".
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers