(Created page with "== Paired t-test == This variation of t-test is used for Paired Data === Paired Data === Two set of observations are ''paired'' if each observation in one set has exactly...") |
(No difference)
|
This variation of t-test is used for Paired Data
Two set of observations are paired if each observation in one set has exactly one corresponding observation is another set.
Examples:
library(openintro) data(textbooks) t.test(textbooks$diff, mu=x.bar.nul, alternative='two.sided')
or
t.test(textbooks$uclaNew, textbooks$amazNew, paired=T, alternative='two.sided')
Test
Calculations
library(openintro) data(textbooks) hist(textbooks$diff, col='yellow') n = length(textbooks$diff) s = sd(textbooks$diff) se = s / sqrt(n) x.bar.nul = 0 x.bar.dif = mean(textbooks$diff) t = (x.bar.dif - x.bar.nul) / se t p = pt(t, df=n-1, lower.tail=F) * 2 p
or
t.test(textbooks$diff, mu=x.bar.nul, alternative='two.sided')
Let $\mu_d = \mu_0 - \mu_1$ be the difference between two methods
Our test:
Say, we have:
Test statistics:
Then we compare it with $t_{397}$
And we conclude that the difference between the two methods is not 0