Multiple testing

Question 1

Issue with multiple testing

Answer 1

• using statistical tests multiple times with a given dataset is prone to errors
• if p value is 0.05, when one does 20 calculations with data, you would get one positive just by chance

Question 2

Bonferroni correction

Answer 2

• to correct for this multiple testing the Bonferroni method can be used (it is strict and may lead to false negatives_
• significance level for multple tested data is altered as (normal significance value/number of statistical analyses carried out)
• makes p value very small

Question 3

Family-wise Error

Answer 3

-represents the probability that any one of a set of comparisons or significance tests is a type 1 error

Question 4

False discovery rate

Answer 4

• new approach to the multiple comparisons problem

- instead of controlling the chance of any false positives, FDR controls the expected proportion of false positives

Question 5

Other multiple testing corrections

Answer 5

• Scheffe test
• Tukey’s honestly significant difference test
• Dunnet test