Post hoc tests

Question 1

what do post hoc tests work out?

Answer 1

where the differences lie between levels and whether these differences are significant

Question 2

post hoc tests are _______ that are used to follow up a significant ANOVA

Answer 2

t tests (either independent t tests or paired t tests)

Question 3

what p value is used for only 1 statistical test?

Answer 3

Alpha 0.05

Question 4

why can’t alpha 0.05 be used for more than 1 statistical test?

Answer 4

5% of the time we make an error so 3 t tests each give 5% chance of error. this makes 15^ of the time we will make a type 1 error of thinking we have an effect when we don’t. This is too high!!! this is called FAMILY WISE ERROR RATE

Question 5

divide the alpha 0.05 by the number of t tests we need to conduct e.g. 0.05/3 = 0.016. what is this called?

Answer 5

a bonferroni correction

Question 6

each post hoc t test is assessed against an alpha of ______ so that we accept that 1.6% of the time we will make a type 1 error

Answer 6

0.016

Question 7

if the p value is less than 0.016 then we have a ______ outcome

Answer 7

significant

Question 8

why is it necessary to correct the <0.05 alpha we use to assess significance?

Answer 8

if we do not correct the alpha, we increase the chance of making a type 1 error. each test runs the risk of being wrong 5% of the time so 3 tests > 15% and 10 tests we would be wrong 50% of the time

Question 9

what values need to be reported in the write up of post hoc tests?

Answer 9

after each of the factors/conditions report M (mean) and SD (standard deviation) also t (df) = [t value], p = [p value]