STA 2 - R LAB 2

Review on Basics of R (cont.) + Paired-Samples t Test

1. Objectives
 Basic graphical and tabular methods
 Editing graphs
 Paired-samples t-test
2. Basic tabular methods
Remember that before we apply a graphical or statistical method on a variable that is to be treated as a
categorical variable, we should be sure that it has been converted into a factor.
We can convert a character variable into a factor by the factor() function. For example,
 data1 <- read.table("mtcars.csv",header=TRUE,sep=",",quote="\"",
stringsAsFactors = FALSE) #Read data into R
 data1$am <- factor(data1$am,ordered=FALSE,levels=c(0,1),labels =
c("automatic","manual")) #Convert into categorical variable

Exercise 1:
a. Explain why the conversion of the data1$am variable is necessary in the above code.
b. Let’s check whether the data1$am variable has been converted into a factor correctly.
c. Go back to the code for importing a text file. What happens if we use stringAsFactors =
TRUE? Try this:
 data2 <- read.table("mtcars.csv",header=TRUE, sep=",",quote="\"",
stringsAsFactors=TRUE)

Now, let’s create a frequency table. We can try:

 am.table <- table(data1$am)
 prop.table(am.table)
 prop.table(am.table)*100

To create a contingency table, use the following format of the table() function:
tableName <- table(row variable, column variable)

Exercise 2: Create a contingency table named gearVSam.table2 showing the relationship between
gear and am. Are you happy with the output? Let’s discuss how to improve it.

3. Basic graphical methods

3.1 Simple bar graph
Based on the frequency table produced previously, we can now produce a simple bar graph. The
following listing shows different ways to plot a bar graph.
 am.table <- table(data1$am)
 barplot(am.table)

1|P a ge
 STA 2 - R LAB 2

 barplot(am.table, main="Bar graph of Transmission", xlab="Types of

Transmission", ylab="Frequency",ylim=c(0,20))
 barplot(am.table, main="Bar graph of Transmission", xlab="Types of
Transmission", ylab="Frequency", horiz=TRUE)
 barplot(am.table, col="skyblue",main="Bar graph of Transmission ",
xlab=" Types of Transmission ", ylab="Frequency")

3.2 Clustered bar graph

Let’s type in the following code:
 trans.vs.gear<-table(data1$am,data1$gear)
 trans.vs.gear
 barplot(trans.vs.gear, beside=TRUE)
 barplot(trans.vs.gear, col=c("red", "yellow"),
beside=TRUE,ylim=c(0,20))

Exercise 3: For the above clustered bar graph

a. Add a title for the graph and labels for the two axes.
b. Use different colors for the bars
c. Convert gear variable to a factor before producing a contingency table and clustered bar graph
and observe the difference.
3.3 Stacked bar graph
The code below will produce a stacked bar graph
 barplot(trans.vs.gear,col=c("red","yellow"))
We can use the spineplot() function to produce a spine plot, a generalized version of the stacked bar
graph. Let’s observe how the spine plot differs from the previous stacked bar graph.
 spineplot(trans.vs.gear, col=c("blue", "green", "pink"))

3.4 Stem and leaf display

 mpg <- data1$mpg
 stem(mpg)

3.5 Histogram
 hist(mpg)
 hist(mpg,breaks=5,col="red")
 hist(mpg, freq=FALSE, breaks=5,col="red")

3.6 Boxplot
The following command is to work with boxplot (for numerical data):
 boxplot(data1$mpg)
 boxplot.stats(data1$mpg)
 boxplot(data1$mpg ~ data1$gear)

2|P a ge
 STA 2 - R LAB 2

4. Editing graphs
4.1 Adding title and axis labels
The function title() adds title and axis labels to a graph. The general format is:
title(main="my title", sub="my sub-title", xlab="x-axis label", ylab="y-
axis label")

The title() function works with the currently active graph.

4.2 Adding a box outside the graph
Use the box()function

5. Paired-samples t Test
We can use the following code to conduct a paired-samples t test to see if the population mean
difference is not zero:
t.test (y1, y2, paired=TRUE, alternative = …,conf.level=0.95)

where y1 and y2 are numeric vectors for the two matched groups and conf.level argument allows us
to specify the confidence level of the reported CI.
Exercise 4. Load the GolfScores.csv dataset. The dataset contains scores of the first and final rounds
for a sample of 20 golfers who completed in PGA tournaments. Suppose you would like to determine
if the mean score for the first round of a PGA Tour event is significantly different than the mean score
for the fourth and final round. Use R to generate the test output. Use  = 0.1.
a) What is the mean difference between in scores for the two rounds? For which round is the
sample mean score lower?
b) What is the p-value? Was the mean score significantly different for the two rounds?
c) What is the 90% confidence interval estimate for the difference between two population
means? Does this CI support your conclusion in part (b) (Does the interval include 0)?
d) Remember that in practice we have to check assumptions for each test we perform. Is the data
distribution for the paired differences reasonably normal?
Note: To check the normality of a dataset, a histogram can be used (but a QQ plot is more useful). In
case of small sample size, however, it is better to use the stem and leaf display and the qq plot to check
if data is normally distributed. The R code for a qq plot is as follows.
 qqnorm(data)#Compare quantiles of our data with theoretical normal
quantiles
 qqline(data)# Add a line to a normal quantile-quantile plot passing
through the first and third quartiles

If the data is normally distributed, the data points should fall in a straight line. Departures from the line
are indicative of a lack of normality.

3|P a ge
 STA 2 - R LAB 2

The R output for this exercise is provided below. You are expected to write R code that produces the
same output:
Paired t-test

data: data3$First and data3$Final

t = -1.416, df = 19, p-value = 0.173
alternative hypothesis: true difference in means is not equal to 0
90 percent confidence interval:
-2.3322058 0.2322058
sample estimates:
mean of the differences
-1.05

Stem and Leaf Display of Golf Score Differences

The decimal point is 1 digit(s) to the right of the |

-0 | 7765
-0 | 43221
0 | 00111112234

Note: If the assumption of normality is violated, the t test may provide misleading results (you should
refer to the practical guidelines regarding how to use one-sample t-test in the Probability and Statistics
course). In such cases, we should use a nonparametric test (to be taught later in this course).
Exercise 5. Load the PriceChange.csv dataset. In early 2009, the economy was experiencing a
recession. The dataset contains data price per share of stock for a sample of 15 companies on January 1
and April 30 (The Wall Street Journal, May 1, 2009).

4|P a ge
 STA 2 - R LAB 2

a. What is the change in the mean price per share of stock over the four-month period?
b. Provide a 90% confidence interval estimate of the change in the mean price per share of stock.
Interprete the results.
c. How was the recession affecting the stock market? Use  = .1
The R output for this exercise is provided below. You are expected to write R code that produces the
same output:

Paired t-test

data: data4$Jan and data4$Apr

t = 2.0043, df = 14, p-value = 0.06478
alternative hypothesis: true difference in means is not equal to 0
90 percent confidence interval:
0.2970457 4.6029543
sample estimates:
mean of the differences
2.45
Stem and Leaf Display of Price Changes
The decimal point is 1 digit(s) to the right of the |

-0 | 432211
0 | 12344
0 | 778
1 | 2

5|P a ge