Question

Why is it necessary to adjust alpha (α) when preforming multiple F-tests following ANOVA to identify where the differences lie?

Answer 1

Answer :

• It's an entirely far from being obviously true theme really. At the point when there are numerous gatherings, a few people say that they have distinctive Power and thus, they ought to be balanced while computing their effect as a unit.
• For instance, Bonferroni test diminishes the estimation of the power by the quantity of tests.
• On the off chance that there are 3 different sets, at that point the estimation of intensity 0.5 will be isolated by 3 to make it 0.017.
• Nonetheless, some school of considerations state that if the sets are pretty much indistinguishable and there is no prominent distinction in intensity of effect, we can basically run Fisher's LSD with α=0.05.
• Numerous researchers are against the variety modification in light of the fact that expanding the intensity of the test (lessening the alpha dimension) builds the likelihood of Type 1 and Type 2 blunders (Rejecting the invalid speculation or tolerating the substitute theory).
• Be that as it may, everything relies upon the arrangement of correlations also and the techniques for utilizing assortment alteration.
• for instance, Bonferroni strategy opens the entryway for auxiliary endpoints where as it doesn't expect to augment the power and conversely, we have Hochberg technique which means to boost the power however barely opens door for optional endpoints.
• So the inevitable point of the investigation ought to be to guarantee that p <= an alongside amplified factual power