Pot hoc tests and comparisons Flashcards
When do we conduct a post hoc test?
- After a significant ANOVA is conducted.
- You reject the Null Hypotheses
- there are 3 or more treatments
- to determine which means are significantly different.
Go back and compare two at a time ( pairwise comparisons )
What is the risk of conducting pairwise comparisons?
Type 1 error 5% Pairwise Error
What is the familywise or experiment-wise error?
As you do more separate tests the risk of Type 1 error accumulates and is called the experiment-wise or familywise alpha level. e.g 6 test at 5% each = 30% chance of a type 1 e
Name some post- hoc tests
LOTS exist ( they are all different types of T-tests) Tukey, Scheffe, LSD, Bonferroni, Sidak, Duncan, Gabriel, Walter-Duncan, REGWQ
What is the benefit/s of using Scheffe Test ( post hoc test)?
It’s not just pairwise test. It can make all sorts of linear comparisons. e.g flying vs non-flying superheroes?
How does Tukeys (post hoc) test for significance?
If the mean difference is larger than Turkeys HSD there is a significant difference, smaller - not significant
Bonferroni correction - adjusted alpha 0.05 divided by no. comparisons making ( planned comparison test)
A t-test with adjusted alpha level to decide significance, Very conservative. It may lead to type 2 errors (non significant result when their really is an effect
What are the disadvantages of using Scheffes Test?
- conservativeness means it has less statistcal power than tukey test and its harder to find a significant result
- uses extreme cautious method to reduce type 1 error allowing greater probability of type 2 error.
Tukey benefits vs disadvantages
Tukey test allows you to compute a single value that determines the min difference between treatment means that is necessary for significance.
If your ANOVA is significant and you are interested in all possible pairwise compariosn what test could you do?
A post hoc test - , such as Tukeys HSD
If your ANOVA is significant and you are only interested in some comparisons, what test would you use?
A planned compariosn, such as Sidak correction
Sidaks test is a modification of Bonferroi test, is it more conservative or less conservative
Sidaks test is less conservative than Bonferroni t ( adjusting alpha)
What type of error occurs when we fail to reject a null hypothesis when it is, in fact, false ( believing that the groups do not differ when they actually do)
Type 2 error
What error is it when we may reject the null hypothesis when it is, in fact, true ( we think there is a difference in our results but there isn’t)
Type 1 - this error can be minimized by selecting an appropriate alpha level
Factors that can influence power include
sample size, effect size. alpha set by researcher
