Independent samples t-test

Independent samples t-test

Working Directory

setwd("/Volumes/EDMS451/Software Instructions")

Manually Enter Data

group1_dat <- data.frame(X = c(4.1, 5.2, 7.5, 8.2, 9.3, 3.1, 1.6,
                               6.0, 7.4, 6.5, 5.1, 9.6, 2.8, 5.9),
                         group = rep(1, 14))
group2_dat <- data.frame(X = c(5.1, 1.2, 2.5, 5.4,
                               7.7, 2.9, 3.0, 2.3),
                         group = rep(2, 8))
dat <- rbind(group1_dat, group2_dat)
colnames(dat) <- c("June1234", "group")   # change the variable name

dat                           # view the data

   June1234 group
1       4.1     1
2       5.2     1
3       7.5     1
4       8.2     1
5       9.3     1
6       3.1     1
7       1.6     1
8       6.0     1
9       7.4     1
10      6.5     1
11      5.1     1
12      9.6     1
13      2.8     1
14      5.9     1
15      5.1     2
16      1.2     2
17      2.5     2
18      5.4     2
19      7.7     2
20      2.9     2
21      3.0     2
22      2.3     2

Check the descriptives

aggregate(June1234 ~ group, data = dat, mean)

  group June1234
1     1 5.878571
2     2 3.762500

aggregate(June1234 ~ group, data = dat, sd)

  group June1234
1     1 2.418257
2     2 2.124643

Independent samples t-test

• Null hypothesis: \(H_0: \mu_1 - \mu_2 = 0\)
• Two sided test: alternative = "two.sided"
• 99% Confidence interval: conf.level = 0.99
• formula testing the effect of group membership on outcome: June1234 ~ group

t.test(June1234 ~ group, data = dat, alternative = "two.sided", conf.level = 0.99)

    Welch Two Sample t-test

data:  June1234 by group
t = 2.1354, df = 16.37, p-value = 0.04817
alternative hypothesis: true difference in means between group 1 and group 2 is not equal to 0
99 percent confidence interval:
 -0.7696543  5.0017972
sample estimates:
mean in group 1 mean in group 2 
       5.878571        3.762500 

# alternative code, yielding the same results
# supply the outcomes in each separate groups
with(dat, t.test(June1234[group == 1], June1234[group == 2], alternative = "two.sided", conf.level = 0.99))

    Welch Two Sample t-test

data:  June1234[group == 1] and June1234[group == 2]
t = 2.1354, df = 16.37, p-value = 0.04817
alternative hypothesis: true difference in means is not equal to 0
99 percent confidence interval:
 -0.7696543  5.0017972
sample estimates:
mean of x mean of y 
 5.878571  3.762500