Chapter 3: Numerical Descriptions of Data

Section 3.3: Ranking (or Position)

Along with the center and the variability, another useful numerical measure is the ranking
of a number. A percentile is a measure of ranking. It represents a location measurement
of a data value to the rest of the values. Many standardized tests give the results as a
percentile. Doctors also use percentiles to track a child’s growth.

kth percentile – the data value that has k% of the data at or below that value

Example #3.3.1: Interpreting Percentile

a.) What does a score of the 90th percentile mean?

Solution:
This means that 90% of the scores were at or below this score. (A person did
the same as or better than 90% of the test takers.)

b.) What does a score of the 70th percentile mean?

Solution:
This means that 70% of the scores were at or below this score.

Example #3.3.2: Percentile Versus Score

If the test was out of 100 points and you scored at the 80th percentile, what was
your score on the test?

Solution:
You don’t know! All you know is that you scored the same as or better than
80% of the people who took the test. If all the scores were really low, you
could have still failed the test. On the other hand, if many of the scores were
high you could have gotten a 95% or so.

There are special percentiles called quartiles. Quartiles are numbers that divide the data
into fourths. One fourth (or a quarter) of the data falls between consecutive quartiles.

To find the quartiles:

1) Sort the data in increasing order.
2) Find the median, this divides the data list into 2 halves. The median is also the second
quartile, Q2
3) Find the median of the data below the median. This value is the first quartile, Q1.
4) Find the median of the data above the median. This value is the third quartile, Q3.
Ignore the median in both calculations for Q1 and Q3

If you record the quartiles together with the maximum and minimum you have five
numbers. This is known as the five-number summary. The five-number summary
consists of the minimum, the first quartile (Q1), the median, the third quartile (Q3), and
the maximum (in that order).

61
 Chapter 3: Numerical Descriptions of Data

The interquartile range, IQR, is the difference between the first and third quartiles, Q1
and Q3. The middle half of the data (50%) falls in the interquartile range. If the IQR is
“large” the data are spread out and if the IQR is “small” the data are closer together.

Interquartile Range (IQR)

IQR  Q3  Q1

Determining probable outliers from IQR: fences

A value that is less than Q1 1.5* IQR (this value is often referred to as the lower fence)
is considered an outlier.
Similarly, a value that is more than Q3  1.5* IQR (the upper fence) is considered an
outlier.

TECHNOLOGY: QUARTILES, 5-NUMBER SUMMARY, IQR

Using your TI84:
 First push STAT 1 and enter the data into L1 (or any other list)
 Push STAT  1 to open 1-Var Stats.
 You need to make your input screen look like one of the screens below (depending
on which operating system your TI84 has). Push 2nd 1 to type L1.

 Then highlight “Calculate” and push ENTER

 You will need to scroll down to see the 5-number summary. These are the last 5
numbers displayed.
 To get the IQR you will need to calculate Q3  Q1

A box plot (or box-and-whisker plot) is a graphical display of the five-number

summary. It can be drawn vertically or horizontally. The basic format is a box from Q1
to Q3, a vertical line across the box for the median and horizontal lines as whiskers
extending out each end to the minimum and maximum. The minimum and maximum can
be represented with dots. Don’t forget to label the tick marks on the number line and
give the graph a title.

Figure #3.3.1: Typical Box Plot

62
 Chapter 3: Numerical Descriptions of Data

An alternate form of a box-and-whiskers plot, known as a modified box plot, only

extends the left line to the smallest value greater than the lower fence, and extends the left
line to the largest value less than the upper fence, and displays markers (dots, circles or
asterisks) for each outlier (values beyond the fences).

TECHNOLOGY: BOXPLOTS, MODIFIED BOXPLOTS

Using your TI84:
 First push STAT 1 and enter the data into L1 (or some other list)
 Push 2nd Y= to open the STAT PLOTS menu. Then push 1 to select Plot1
 You need to make your input screen look like the screen below. The 4th graph type
is the modified boxplot, the 5th is just a boxplot.

 Then push ZOOM 9 to see the boxplot or modified boxplot

If the data are symmetrical, then the box plot will be visibly symmetrical. If the data
distribution has a left skew or a right skew, the line on that side of the box plot will be
visibly longer. If the plot is symmetrical, and the four quartiles are all about the same
length, then the data are likely a near uniform distribution. If a box plot is symmetrical,
and both outside lines are noticeably longer than the Q1 to median and median to Q3
distance, the distribution is then probably bell-shaped.

Example #3.3.3: Five-number Summary for an Even Number of Data Points

The total assets in billions of Australian dollars (AUD) of Australian banks for the
year 2012 are given in table #3.3.1 ("Reserve bank of," 2013). Find the five-
number summary and the interquartile range (IQR), and draw a boxplot.

Table #3.3.1: Total Assets (in billions of AUD) of Australian Banks

2855 2862 2861 2884 3014 2965
2971 3002 3032 2950 2967 2964

Solution:
First identify the individual object, variable, and type of variable in the problem.
Individual object: a randomly selected Australian bank
Variable: total assets
Type of variable: quantitative-discrete
Table #3.3.2: Sorted Data for Total Assets (not needed when using tech)
2855 2861 2862 2884 2950 2964 2965 2967 2971 3002 3014 3032

For the 5-number summary, we need the min, Q1, Med, Q3 and max.

63
 Chapter 3: Numerical Descriptions of Data

From the sorted data above, we can see that the minimum is 2855 billion AUD
and the maximum is 3032 billion AUD. There are 12 data points so the median is
the average of the 6th and 7th numbers.

Table #3.3.3: Sorted Data for Total Assets with Median

2855 2861 2862 2884 2950 2964 2965 2967 2971 3002 3014 3032

2964  2965
Median =  2964.5 billion AUD
2

Table #3.3.4: To find Q1, find the median of the first half of the list.
2855 2861 2862 2884 2950 2964

Q1
2862  2884
Q1   2873 billion AUD
2

Table #3.3.5: To find Q3, find the median of the second half of the list.
2965 2967 2971 3002 3014 3032

Q3
2971  3002
Q3   2986.5 billion AUD
2

The 5-number summary is {2855, 2873, 2964.5, 2986.5, 3032} billion AUD.

To find the interquartile range, IQR, find Q3  Q1 .

IQR  2986.5  2873  113.5 billion AUD

This tells you the middle 50% of assets were within 113.5 billion AUD of each
other.

You can use the five-number summary to draw the boxplot:

64
 Chapter 3: Numerical Descriptions of Data

Graph #3.3.1: Box Plot of Total Assets of Australian Banks

The distribution is skewed right because the right tail is longer.

Using technology:
Using 1-Var Stats on the TI84 you would get the following:

The 5-number summary is {2855, 2873, 2964.5, 2986.5, 3032} billion AUD.

Using the TI84 graphing tool, you would get the following boxplot:

NOTE: On the TI84 boxplot you do not have a scale to refer to. By pressing the
“Trace” button, and then using the left and right cursors, you will see the 5-
number summary values listed at the bottom of the screen.

Example #3.3.4: Five-number Summary for an Odd Number of Data Points

The life expectancy for a person living in one of 11 countries in the region of
South East Asia in 2012 is given below ("Life expectancy in," 2013). Find the
five-number summary for the data and the IQR, then draw a box-and-whiskers
plot.

65
 Chapter 3: Numerical Descriptions of Data

Table #3.3.6: Life Expectancy of a Person Living in South-East Asia

70 67 69 65 69 77
65 68 75 74 64

Solution:
First identify the individual object, variable, and type of variable in the problem.
Individual object: a randomly selected person living in south-east Asia
Variable: life expectancy
Type of variable: quantitative-continuous

Table #3.3.7: Sorted Life Expectancies (not needed when using technology)
64 65 65 67 68 69 69 70 74 75 77
For the 5-number summary, we need the min, Q1, Med, Q3 and max.
From the sorted data above, we can see that the minimum is 64 years and the
maximum is 77 years. There are 11 data points so the median is the 6th number in
the ordered list.
Table #3.3.8: Finding the Median of Life Expectancies
64 65 65 67 68 69 69 70 74 75 77

Median = 69 years

Finding the Q1 and Q3 you need to find the median of the numbers below the
median and above the median. The median is not included in either calculation.

Table #3.3.9: Finding Q1

64 65 65 67 68

Q1
Q1 = 65 years

Table #3.3.10: Finding Q3

69 70 74 75 77

Q3
Q3 = 74 years

Now, the five-number summary is {64, 65, 69, 74, 77} (in years)

To find the interquartile range (IQR)

IQR  Q3  Q1  74  65  9 years

66
 Chapter 3: Numerical Descriptions of Data

The middle 50% of life expectancies are within a span of 9 years.

Graph #3.3.2: Box Plot of Life Expectancy

This distribution looks somewhat skewed right, since the whisker is longer on the
right. However, it could be considered almost symmetric too since the box looks
somewhat symmetric.

Using technology:

Using 1-Var Stats on the TI84 you would get the following:

The 5-number summary is {64, 65, 69, 74, 77} (in years).

Using the TI84, you would get the following boxplot:

The above data set does not have any outliers, so its modified boxplot would look the
same as its boxplot. Let’s take the original data and pretend that one of the countries had
a life expectancy of 32 years and make the modified boxplot. The steps are the same: just
be sure to select the 4h graph option for the modified boxplot.

70 67 69 65 69 77
65 68 75 74 64 32

Graph #3.3.4: Modified Boxplot for Life Expectancy with Outlier Added

67
 Chapter 3: Numerical Descriptions of Data

Notice the dot at 32 that is used to designate that data point as an outlier.

Comparing Data Sets with Boxplots

Many statistical studies are done to compare different groups.
o Compare SAT scores for students at different high schools
o Compare starting salaries for employees in different states
o Compare the number of defectives produced by different machines
o Compare SAT scores before and after taking a prep course

In a later chapter we will get methods for making these comparisons through formal
inferential procedures, but we always start first comparing with graphical and numerical
summaries. You can draw 2 boxplots side by side (or one above the other) to compare 2
samples. Since you want to compare the two data sets, make sure the box plots are on
the same axes.
o To compare measures of center, you can compare the medians
o To compare measures of variation you can compare ranges (distance between
whisker tips) or IQRs (length of boxes).
o To compare measures of position you can compare Min, Q1, Med, Q2, and
Max.

As an example, suppose you look at the boxplot for life expectancy for European
countries and Southeast Asian countries.

Graph #3.3.3: Box Plot of Life Expectancy of Two Regions

68
 Chapter 3: Numerical Descriptions of Data

o Looking at the boxplots above, you will notice that the three quartiles for life
expectancy are all higher for the European countries, yet the minimum life
expectancy for the European countries is less than that for the Southeast Asian
countries.
o The middle life expectancy for European countries is higher than that of the
Southeast Asian countries.
o The life expectancies of the European countries are more varied (spread out)
than those of the Southeast Asian countries.
o The middle 50% of the life expectancies of the Southeast Asian countries are
more varied (spread out) than those of the European countries.
o The life expectancy for the European countries appears to be skewed left,
while the life expectancies for the Southeast Asian countries appear to be
more symmetric.
Example #3.3.5: Putting it all together
A random sample was collected on the health expenditures (as a % of GDP) of countries
around the world. The data are in Table #3.3.11. Using graphical and numerical
descriptive statistics, analyze the data and use it to predict the health expenditures of all
countries in the world.

Table #3.3.11: Health Expenditures as a Percentage of GDP

3.35 5.94 10.64 5.24 3.79 5.65 7.66 7.38 5.87 11.15
5.96 4.78 7.75 2.72 9.50 7.69 10.05 11.96 8.18 6.74
5.89 6.20 5.98 8.83 6.78 6.66 9.45 5.41 5.16 8.55

Solution:
First, it might be useful to look at a visualization of the data, so create a histogram
using your TI84. Start the bins at 2 and use a bin width of 2. Remember, to
create a histogram where you (and not your calculator) determine the bin widths,
you will want to set your Window settings to look like the screen below:

Once you have set your window manually, press GRAPH (do NOT use ZOOM 9
since your calculator will override your window settings).

69
 Chapter 3: Numerical Descriptions of Data

Graph #3.3.5: Histogram of Health Expenditure

The graph above shows the results of pressing the “TRACE” button in order to
see the bin limits and frequency. Your can use the left and right arrows to read
other bins. If you were to create this histogram by hand or using other
technology, you would want to be sure to include vertical and horizontal scale as
well as a graph title.

From the graph, the data appears to be somewhat skewed right. So there are some
countries that spend more on health based on a percentage of GDP than other
countries, but the majority of countries appear to spend around 4 to 8% of their
GDP on health.

We can use our TI84 to get the numerical summaries.

Using 1-Var Stats on a list of data.

The sample mean is x  7.03%

The sample standard deviation is s = 2.27%,
The five-number summary is {2.72%, 5.71%, 6.70%, 8.46%, 11.96%}

To visualize the five-number summary and check for outliers, create a modified
boxplot.

70
 Chapter 3: Numerical Descriptions of Data

Graph #3.3.6: Modified Boxplot of Health Expenditure

So it appears that countries spend on average about 7% of their GPD on health.

The spread is somewhat low, since the standard deviation is fairly small, which
means that the data are fairly consistent. The five-number summary confirms that
the data are slightly skewed right. The modified box plot shows that there are no
outliers. So from all of this information, one could say that countries spend a
small percentage of their GDP on health and that most countries spend around the
same amount. There doesn’t appear to be any country that spends much more than
other countries or much less than other countries.

Another measure of position (or ranking) is the z-score. The z-score is a measure of how
many standard deviations an x value is away from the mean.
z-score
x
z

 Where  = mean of the population of all x values and  = standard deviation for the
population of all x values
 Z-scores are unitless statistics that are usually rounded to two decimal places.

Example #3.3.6: Z-Score

A student scores 70 points on an exam where the mean was 87 points and the standard
deviation was 5.7 points. On a later exam the student scores 75 points on an exam where
the mean was 85 points and the standard deviation was 8.9 points. On which exam did
the student score better relative to the group?

Solution:
To compare how the student did relative the rest of the people in the group we
need to compute a measure of ranking (position). Since we are given the means
and standard deviations, we are going to compute z-scores as the measures of
position.

71
 Chapter 3: Numerical Descriptions of Data

70 pts  87 pts 17 pts

z1    2.98
5.7 pts 5.7 pts

75 pts  85 pts 10 pts

z2    1.12
8.9 pts 8.9 pts

Interpret: The first exam score was 2.98 standard deviations below average. The
second exam score was only 1.12 standard deviations below average. The student
did better relative to the class on exam 2 (not as far below the average…higher z-
score).

NOTE: In Example #3.3.6 the larger z-score was the better score, but that is not always
the case. You always need to consider the context of the problem. If we were comparing
times to complete a task, we would want faster times (lower times) and so when
comparing z-scores in this case, lower would be better. When trying to create a product
with consistent measurements (like bolts), the better bolts are the ones with z-scores
closer to 0.

72
 Chapter 3: Numerical Descriptions of Data

Section 3.3: Homework

1.) Suppose you take a standardized test and you are in the 10th percentile. What
does this percentile mean? Can you say that you failed the test? Explain.

2.) Suppose your child takes a standardized test in mathematics and scores in the 96th
percentile. What does this percentile mean? Can you say your child passed the
test? Explain.

3.) Suppose your child is in the 83rd percentile in height and 24th percentile in weight.
Describe what this tells you about your child’s stature.

4.) Suppose your work evaluates the employees and places them on a percentile
ranking. If your evaluation is in the 65th percentile, do you think you are working
hard enough? Explain.

5.) Cholesterol levels (mg/dL) were collected from a random sample of patients two
days after they had a heart attack (Ryan, Joiner & Ryan, Jr, 1985) and are in table
#3.3.12.
Table #3.3.12: Cholesterol Levels
270 236 210 142 280 272 160
220 226 242 186 266 206 318
294 282 234 224 276 282 360
310 280 278 288 288 244 236
Find the five-number summary and interquartile range (IQR), and draw a
modified boxplot.

6.) The lengths (in kilometers) of a random sample of rivers on the South Island of
New Zealand that flow to the Pacific Ocean are listed in table #3.3.13 (Lee,
1994).
Table #3.3.13: Lengths of Rivers (km) Flowing to Pacific Ocean
River Length River Length
(km) (km)
Clarence 209 Clutha 322
Conway 48 Taieri 288
Waiau 169 Shag 72
Hurunui 138 Kakanui 64
Waipara 64 Waitaki 209
Ashley 97 Waihao 64
Waimakariri 161 Pareora 56
Selwyn 95 Rangitata 121
Rakaia 145 Ophi 80
Ashburton 90
Find the five-number summary and interquartile range (IQR), and draw a
modified boxplot.

73
 Chapter 3: Numerical Descriptions of Data

7.) The lengths (in kilometers) of a sample of rivers on the South Island of New
Zealand that flow to the Tasman Sea are listed in table #3.3.14 (Lee, 1994).

Table #3.3.14: Lengths of Rivers (km) Flowing to Tasman Sea

River Length River Length
(km) (km)
Hollyford 76 Waimea 48
Cascade 64 Motueka 108
Arawhata 68 Takaka 72
Haast 64 Aorere 72
Karangarua 37 Heaphy 35
Cook 32 Karamea 80
Waiho 32 Mokihinui 56
Whataroa 51 Buller 177
Wanganui 56 Grey 121
Waitaha 40 Taramakau 80
Hokitika 64 Arahura 56

Find the five-number summary and interquartile range (IQR), and draw a
modified boxplot.

8.) The FDA regulates that fish that are consumed should contain no more than 1.0
mg/kg of mercury. In Florida, bass fish were collected in 53 different lakes to
measure the amount of mercury in the fish. The data for the average amount of
mercury in the fish from each lake is in table table #3.3.15 ("Multi-disciplinary
niser activity," 2013).

Table #3.3.15: Average Mercury Levels (mg/kg) in Fish

1.23 1.33 0.04 0.44 1.20 0.27

0.48 0.19 0.83 0.81 0.71 0.5
0.49 1.16 0.05 0.15 0.19 0.77
1.08 0.98 0.63 0.56 0.41 0.73
0.34 0.59 0.34 0.84 0.50 0.34
0.28 0.34 0.87 0.56 0.17 0.18
0.19 0.04 0.49 1.10 0.16 0.10
0.48 0.21 0.86 0.52 0.65 0.27
0.94 0.40 0.43 0.25 0.27

Find the five-number summary and interquartile range (IQR), and draw a
modified boxplot.

74
 Chapter 3: Numerical Descriptions of Data

9.) A study was conducted to see the effect of exercise on pulse rate (in bpm). A
random sample of male subjects were taken who do not smoke, but do drink.
Their pulse rates were measured ("Pulse rates before," 2013). Then they ran in
place for one minute and then measured their pulse rate again. Graph #3.3.7 is of
boxplots that were created of the before and after pulse rates. Discuss any
conclusions you can make from the graphs.

Graph #3.3.7: Boxplots of Pulse Rates for Males

10.) A study was conducted to see the effect of exercise on pulse rate (in bpm). A
random sample of female subjects were taken who do not smoke, but do drink.
Their pulse rates were measured ("Pulse rates before," 2013). Then they ran in
place for one minute, and after measured their pulse rate again. Graph #3.3.8 is of
boxplots that were created of the before and after pulse rates. Discuss any
conclusions you can make from the graphs.

Graph #3.3.8: Boxplots of Pulse Rates for Females

75
 Chapter 3: Numerical Descriptions of Data

11.) The number of deaths attributed to UV radiation in African countries and Middle
Eastern countries in the year 2002 were collected by the World Health
Organization ("UV radiation: Burden," 2013). Graph #3.3.10 is of boxplots that
were created of the deaths in African countries and deaths in Middle Eastern
countries. Discuss any conclusions you can make from the graphs.

Graph #3.3.10: Boxplots of UV Radiation Deaths in Different Regions

12.) A child is 37 inches tall. The population of children of the same age and gender
have a mean height of 39.2 inches with a standard deviation of 6.3 inches.
Another child is 39 inches tall. The population of children of the same age and
gender as this child have a mean height of 40.5 inches with a standard deviation
of 3.9 inches.
a) Which child is taller? (this is not a trick question…no statistics involved here)
b) Which child is taller relative to their peers? (Now you have to use statistics)

13.) A runner for team 1 can run a race in 57 seconds. Team 1 has running times with
a mean of 64.2 seconds and a standard deviation of 4.2 seconds. A runner for
team 2 can run a race in 58 seconds. Team 2 has running times with a mean of
62.1 seconds and a standard deviation of 1.2 seconds.
a) Which runner is faster? (This is not a trick question: no statistics involved
here)
b) Which runner is faster relative to their team? (Now you have to use statistics)

76
 Chapter 3: Numerical Descriptions of Data

Data Sources:
Annual maximums of daily rainfall in Sydney. (2013, September 25). Retrieved from
[Link]

Lee, A. (1994). Data analysis: An introduction based on r. Auckland. Retrieved from

[Link]

Life expectancy in southeast Asia. (2013, September 23). Retrieved from

[Link]

Olson, K., & Hanson, J. (1997). Using reiki to manage pain: a preliminary report. Cancer
Prev Control, 1(2), 108-13. Retrieved from
[Link]

Pulse rates before and after exercise. (2013, September 25). Retrieved from
[Link]

Reserve bank of Australia. (2013, September 23). Retrieved from

[Link]

Ryan, B. F., Joiner, B. L., & Ryan, Jr, T. A. (1985). Cholesterol levels after heart attack.
Retrieved from [Link]

Time between nerve pulses. (2013, September 25). Retrieved from

[Link]

Time of passages of play in rugby. (2013, September 25). Retrieved from

[Link]

U.S. tornado climatology. (17, May 2013). Retrieved from

[Link]

UV radiation: Burden of disease by country. (2013, September 4). Retrieved from

[Link]

CO2 emissions. (2013, November 19). Retrieved from

[Link]

Multi-disciplinary niser activity - mercury in bass. (2013, November 16). Retrieved from
[Link] - [Link]

77