CHAPTER 2: DESCRIPTIVE STATISTICS

Lecture Notes for Introductory Statistics1

Daphne Skipper, Augusta University (2016)

1. Stem-and-Leaf Graphs, Line Graphs, and Bar Graphs

The distribution of data is how the data is spread or distributed over the range
of the data values. This is one of the first and most important aspects one would
want to know about any data set.
Stem-and-leaf graphs provide a quick way to view the distribution of a small
data set by hand.
Example 1. Stem-and-leaf. The following are the ages of 29 actors at the time
that they won the Best Actor award:
18 21 22 25 26 27 29 30 31 33
36 37 41 42 47 52 55 57 58 62
64 67 69 71 72 73 74 76 77
Data from Example 2.18 in the textbook.
Explore the distribution of the best actor data by constructing a stem-and-leaf
graph.

We already saw bar graphs in Chapter 1. Frequency polygons and time series
graphs are two types of line graphs that we will see in the next section.

2. Histograms, Frequency Polygons, and Time Series Graphs

A histogram is similar to a bar graph. The difference is that bar graphs are for
categorical data, whereas histograms are for quantitative data. Since quantitative
data aren’t naturally sorted into categories, we must create “classes” of data: equal
width intervals of data values. We record classes and frequencies in a frequency
table. We constructed a frequency table and histogram using height data in Chapter
1.
A frequency polygon displays the same data as a histogram, but in line graph
form. This format is useful for comparing the distributions of multiple datasets by
overlaying frequency polygons.
Example 2. Overlaying frequency polygons. Figure 1 is an overlay of fre-
quency polygons of men’s and women’s pulse data. By overlaying frequency poly-
gons, we are able to easily compare the distributions (histograms) of two data sets.
(This chart is from Elementary Statistics by Mario Triola.)
A time series graph displays the trend of a variable over time. The x-axis is
time (years, minutes, seconds, etc.). The y-axis is the range of data values.
Example 3. Time series graph. The time series graph in Figure 2 allows us to
easily see the trends in housing prices in Australia versus the US.
1
Chapter 2 Notes Descriptive Statistics D. Skipper, p 2

Figure 1. Frequency Polygons

Figure 2. Time Series Graph

This TED talk contains an amazing example of data visualization: http://www.

ted.com/talks/hans_rosling_shows_the_best_stats_you_ve_ever_seen. (Start
at around the 2:30 mark.) Notice how Hans Rosling’s methods make it easy to see
patterns in extremely complicated time-series data.

1
These lecture notes are intended to be used with the open source textbook “Introductory
Statistics” by Barbara Illowsky and Susan Dean (OpenStax College, 2013).
Chapter 2 Notes Descriptive Statistics D. Skipper, p 3

3. Measures of Location of Data

The median of a data set is the “middle value”, but it does not have to be one
of the data values.
Example 4. Median. Find the median.
(1) 5 5 6 8 12
(2) 5 5 6 8 12 17

The kth percentile, Pk , of a data set is the value that separates the lower k% of
data values from the upper (100 − k)% of data values.
Example 5. Understanding percentiles.
(1) The 35th percentile, or P35 , separates the lower % of data values
from the upper % of data values.
th
(2) “Median” is another name for the percentile.

Percentiles are calculated on data that is sorted from lowest to highest. Low
percentiles correspond to low data values. High percentiles correspond to high data
values.
Example 6. Interpreting percentiles.
(1) Joe ran a 5K and his finishing time is the 5th percentile. Interpret the 5th
percentile in this context. (What percent of racers did Joe beat? Is this a
good finishing time?)
(2) If Abby takes the SAT, would she prefer for her score to be P10 or P88 of
SAT scores for this year? Explain.

Locating the k th percentile. The position in the sorted data set of the k th
percentile is
k
i= (n + 1).
100
If i is a whole number, Pk is the data value found at position i of the sorted data
set. If i is a decimal, round i up and round i down; Pk is the average of the data
values found at these two positions in the sorted data set.
Example 7. Calculating percentiles. The following are the sorted heights in
inches of 40 students in a statistics class.
61 61 62 62 63 63 63 65 65 65
66 66 66 66 66 67 67 67 68 68
68 68 68 68 68 68 69 69 69 69
69 69 70 71 72 72 72 73 73 74
Data from TryIt 2.24 in the textbook.

(1) Find the 80th percentile of the class heights. Use appropriate notation to
express the answer. Interpret the answer.
(2) Find the 38th percentile of the class heights. Use appropriate notation.

Finding the percentile of a data value. let x be the number of data values
below that data value and let y be the number of data values equal to that data
Chapter 2 Notes Descriptive Statistics D. Skipper, p 4

value in the data set. The the percentile k of the data value is
x + 0.5y
k= (100).
n
Example 8. Finding the percentile of a data value. Use the class height
data from Example 7. Use appropriate notation to express your answer.
(1) Kara is 67 inches tall. At what percentile is Kara’s height?
(2) Charles is 73 inches tall. At what percentile is Charles’ height?

4. Boxplots
Quartiles are numbers that separate the data into quarters. Like all percentiles,
quartiles may or may not be actual data values. The first quartile, Q1 , is the middle
of the bottom half of data; Q1 = P25 . The second quartile, Q2 , is the median;
Q2 = P50 = median. The third quartile, Q3 is the middle of the top half of the
data; Q3 = P75 . The five number summary of a data set is: minimum, Q1 ,
median, Q3 , maximum.
We can calculate quartiles “by hand” just as we calculate any other percentiles.
However, the calculator will calculate quartiles for us directly.
Example 9. Using the calculator to find quartiles. The following are
the heights in inches of 20 boys in a statistics class.
66 66 67 67 68 68 68 68 68 69
69 69 70 71 72 72 72 73 73 74
The following are the heights of the 20 girls in the same statistics class.
61 61 62 62 63 63 63 65 65 65
66 66 66 67 68 68 68 69 69 69
Data from TryIt 2.24 in the textbook.

(1) Enter the 20 boys’ heights into L1 .

(2) Find the 5 number summary for the boys’ heights. (Use stat, calc, 1
var stats.)
(3) Enter the 20 girls’ heights into L2 .
(4) Find the 5 number summary for the girls’ heights.

A boxplot is a graph of the 5 number summary scaled on a numberline, showing

the concentration of data. The spread of the middle 50% of data values, the data
values between Q1 and Q3 , are indicated by a box. The median is marked in the
box. The bottom 25% and top 25% of data values are indicated by the “whiskers”.
Example 10. Constructing a boxplot. Use boys’ and girls’ height data from
Example 9.
(1) Construct a boxplot for the boys’ height data by hand.
(2) Use the calculator to construct boxplots for both data sets on the same
scale.

Example 11. Interpreting boxplots. Use the boxplots from Example 10.
(1) 25% of girls are shorter than inches. What quartile is this?
(2) 50% of girls are shorter than inches. What quartile is this?
Chapter 2 Notes Descriptive Statistics D. Skipper, p 5

(3) 25% of boys are taller than inches. What quartile is this?
(4) In which quartile are the girls’ heights most concentrated? The boys’
heights?

The interquartile range, or IQR, is the range of the middle 50% of data values:
IQR = Q3 − Q1 . The IQR is often used to identify outliers in the following way.
Data values that are below Q1 − (1.5)IQR or above Q3 + (1.5)IQR are considered
outliers.
Example 12. Identifiying outliers. Use the girls’ class height data from the
last few examples.
(1) Calculate the IQR of the girls height data.
(2) Find the boundary height below which a girls’ height would be considered
a “short” outlier.
(3) Find the boundary height above which a girls’ height would be considered
a “tall” outlier.
(4) According to these boundary values, does this data set contain any outliers?

5. Measures of the Center of Data

The mean is the most commonly used measure of the “center” of data. The
mean is calculatated by adding all the data values and dividing by n, sample size,
or the number of data values. Notation for mean:
x̄ = sample mean,
µ = population mean.

The median is the second most commonly used measure of the center. Since
the median does not take the actual exact values of data into account, it is a better
choice when there are extreme data values. In this case, the mean can be skewed
toward the extreme data value(s) and be misleading with respect to the center of
the bulk of the data. For example, we hear about “median income” rather than
mean income, because there are some extremely large incomes that would make the
ordinary income, as measured by the mean, appear larger than it really is for the
average person.
The mode is the most frequently occuring data value. There can be more than
one mode if there is a tie for the data value with the highest frequency. The mode
is used primarily for categorical data, for which the mean and median don’t make
sense.
Example 13. Mode. Find the mode of the boys’ height data:
66 66 67 67 68 68 68 68 68 69
69 69 70 71 72 72 72 73 73 74

Example 14. Mean using a frequency table. Use the same boys’ height data
as above.
(1) Make a frequency table of the height data.
(2) Use the frequency table to calculate the mean of the height data by hand.
Chapter 2 Notes Descriptive Statistics D. Skipper, p 6

(3) Using the calculator, enter the frequency table into L1 and L2 .
(4) Calculate the mean and median using 1 Var Stats.

We can’t get the exact mean of data from a grouped frequency table, but we can
estimate the mean by using the midpoint of each class in place of each data value
in that class.
Example 15. Mean of a grouped frequency table. Maris conducted a
study on the effect that playing video games has on memory recall. As part of her
study, she compiled the following data.
Hours spend on video games Number of teens
0-3 3
4-7 7
8-11 12
12-15 7
16-19 9
(1) Find the midpoint of each class.
(2) What is the best estimate of the mean number of hours spent playing video
games?
TryIt 2.30.

The Law of Large Numbers says that as sample size increases, sample mean
(x̄) gets closer and closer to the population mean (µ).

6. Skewness and the Mean, Median, and Mode

Data is skewed to the left if there is a “tail” of data to the left in the histogram,
or on the low end of the data. Data is skewed to the right if there is a “tail” of
data to the right in the histogram, or on the high end of the data.
The mean will be pulled in the direction of the skewing due to extreme data
values. If there is more than mild skewing, the median is a more approprate measure
of center than the mean for a more accurate summary of the bulk of the data.

7. Measures of the Spread of Data

Standard deviation is the most commonly used measure of the spread of data.
The standard deviation tells us approximately how far data values are from the
mean, on average. A larger standard deviation indicates that the data values are
farther from the mean, on average.
Example 16. Interpreting standard deviation. Suppose the average (mean)
wait time in line at both Publix and BiLo is 5 minutes. However, the standard
deviation of the wait times at Publix is 1 minute and the standard deviation of the
wait times in BiLo is 3 minutes.
(1) Which supermarket has more variation in wait times?
(2) At which supermarket would you be able to predict your wait time more
precisely?

The standard deviation can be used to determine if a data value is close to or far
away from the mean relative to the rest of the data. As a rule of thumb, more than
Chapter 2 Notes Descriptive Statistics D. Skipper, p 7

two standard deviations from the mean is considered “unusual”. (Unusual data
values are not as extreme as “outliers.” More than three standard deviations from
the mean is a more reasonable boundary for outliers, if we want to use standard
deviation instead of the IQR formula.)

Example 17. Comparing data values using standard deviation. Suppose

the wait times at Kroger are 5 minutes with a standard deviation of 2 minutes.
(1) Draw a number line and mark the mean at 5 minutes and mark 1, 2, and
3 standard deviations above and below the mean. Label the unusual data
value range(s) on the number line.
(2) Rosa waits 3 minutes at Kroger. How many standard deviations from the
mean is her wait time?
(3) Binh waits 11 minutes at Kroger. How many standard deviations from the
mean is his wait time?
(4) Is either wait time unusual at Kroger?

The standard deviation formulas are a bit much to calculate by hand, so we

will use the calculator to find standard deviation. The formulas for sample and
population standard deviation are different and your calculator provides both, so
it is very important to be able to identify the notation for each when using the
calculator:

s = sample standard deviation,

σ = population standard deviation.

The calculator uses subscripts on these symbols to indicate which variable is being
used: sx and σx , for example. We will almost always have sample data (as
opposed to population data), so we will almost always use sx from the
calculator for standard deviaton.

Example 18. Standard deviation using calculator. Use your calculator to

find the standard deviation of the boys’ height data:

66 66 67 67 68 68 68 68 68 69
69 69 70 71 72 72 72 73 73 74

Consider the data to be a sample from the population of male students at the
school. (Is it reasonable to consider this to be a representative sample?) You may
wish to enter the data in frequency table format, if you don’t still have it in your
calculator.

The deviation of a data value from the mean is the signed distance of the data
value from the mean:

deviation = data value − mean.

The z score of a data value is the number of standard deviations from the mean
that data value is. If the data values is below the mean, the z score is negative. If
Chapter 2 Notes Descriptive Statistics D. Skipper, p 8

the data value is above the mean, the z score is positive.

deviation
z =
standard deviation
data value − mean
=
standard deviation
x − x̄
=
s
Example 19. Notation. Write the formula for a population z score, using the
appropriate population symbols.

Example 20. Z score. Use the same boys’ height data, which has mean x̄ = 69.5
and standard deviation s = 2.5.
(1) Find the z score of a height of 74 inches. Should we consider 74 inches to
be unusually tall among these boys?
(2) Find the height that has a z score of -0.7.

Example 21. Compare data values from different data sets. Find the
z score for each girl’s time relative to her team. Which girl has the fastest time
relative to her teammates?
Swimmer Time(sec) Team Mean Time Team Standard Deviation
Angie 26.2 27.2 0.8
Beth 27.3 30.1 1.4
TryIt 2.35

The following two rules give us more precise ideas of how far the data are spread
from the mean based on the standard deviation. Chebyshev’s Rule is for ANY data
set:
(1) At least 75% of data is within two standard deviations of the mean
(2) At least 89% of the data is within three standard deviations of the mean
(3) At least 95% of the data is within 4.5 standard deviations of the mean
The Empirical Rule, also known as the 68-95-99 Rule, is ONLY for data with a
BELL-SHAPED and SYMMETRIC histogram/distribution:
(1) Approximately 68% of data is within one s.d. of the mean
(2) Approximately 95% of data is within two s.d.s of the mean
(3) Approximately 99% of data is within one s.d.s of the mean

The formula for sample standard deviation is

r
Σ(x − x̄)2
s= .
n−1
The formula for population standard deviation is
r
Σ(x − µ)2
σ= .
N
We won’t use these formulas in practice, but it would be good to get a feel for
how the formula approximates the average distance of data values from the mean,
and to recognize that the formulas for sample standard deviation and population
standard deviation differ.
Chapter 2 Notes Descriptive Statistics D. Skipper, p 9

Example 22. The standard deviation formula. In groups of 3 or 4 students,

guess the age of the person in the provided photograph. Submit your group’s
guess. Instructor note: Provide a photograph of someone that you know, but who is
unknown to the students for this short activity. It’s also fun with more photographs.
(1) List each guess in the first column a table.
(2) In the second column, list the deviation of each guess from the true age:
guess - true age.
(3) Does a negative deviation indicate a low or high guess?
(4) Are the guesses biased young? Biased old? Relatively unbiased?
(5) In the third column, list the absolute value of the deviation from the pre-
vious column. The most natural average distance from the true age is the
mean absolute deviation. However, absolute values are not practical in alge-
braic formulas. Calculate the mean absolute deviation. This is the average
distance of the guesses from the true age.
(6) In the fourth column, list the square of each deviation.
(7) Now calculate the variance (the square of the standard deviation): Add up
the values in the 4th column and divide by the n − 1. The units of variance
is “years squared”.
(8) Take the square root of the variance. This is the standard deviation of the
guesses from the true age. The units of standard deviation is “years”, the
same as the data values.
(9) Compare the mean absolute deviation and the standard deviation. (Are
they the same?)
(10) Find the z score of your group’s guess. Which group had the best guess?
Did any group have an “unusual” guess?
Adapted from “Teaching Statistics, a bag of tricks” by Gelman and Nolan.
guess deviation |deviation| (deviation)2
(guess - true age)