The literature on Bonferroni correction methods is confusing me...
I have compared means of multiple variables (some dependent, some independent) with Fisher's exact tests, Kruskal-Wallace tests, and Mardia-Watson-Wheeler tests. All in an attempt to identify important variables in the presence or absence of a species. Should I use the sequential Bonferroni correction now on all of my p-values?
It appears based on the literature that this may be useful since the high number of variables I have measured may increase the "incorrect" probability of getting a significant P-value in one of my variables. However, I am not sure if the "multiple tests" that the literature keeps referencing is referring to multiple statistical tests on one data set, i.e. temperature affecting presence/absence; or instead the multiple variables I have measured in order to predict presence/absence of a species, i.e. temperature, elevation, distance to water, etc.
I only used one statistical test for each variable.
So, in summary: is it logical to now use the sequential Bonferroni correction method on my list of P-values that are associated with different variables? (I am using R if this is relevant to any answers!)
Thank you so much in advance for your patience!
Sincerely,
Confused & frustrated grad student.
I have compared means of multiple variables (some dependent, some independent) with Fisher's exact tests, Kruskal-Wallace tests, and Mardia-Watson-Wheeler tests. All in an attempt to identify important variables in the presence or absence of a species. Should I use the sequential Bonferroni correction now on all of my p-values?
It appears based on the literature that this may be useful since the high number of variables I have measured may increase the "incorrect" probability of getting a significant P-value in one of my variables. However, I am not sure if the "multiple tests" that the literature keeps referencing is referring to multiple statistical tests on one data set, i.e. temperature affecting presence/absence; or instead the multiple variables I have measured in order to predict presence/absence of a species, i.e. temperature, elevation, distance to water, etc.
I only used one statistical test for each variable.
So, in summary: is it logical to now use the sequential Bonferroni correction method on my list of P-values that are associated with different variables? (I am using R if this is relevant to any answers!)
Thank you so much in advance for your patience!
Sincerely,
Confused & frustrated grad student.