Section 2 Answer Key:

0) Find the median and quartiles of each of the following sets of numbers. These
represent the “four cases” that you should be able to compute using the rules in this
course.

(a) 23, 35, 28, 33, 5, 12, 40, 25, 20, 18, 1, 16

1, 5, 12, 16, 18, 20 23, 25, 28, 33, 35, 40

n = 12 M = (20 + 23)/ 2 = 21.5

Q1 = (12 + 16) / 2 = 14 Q3 = (28 + 33) /2 = 30.5

(b) 22, 33, 25, 28, 5, 12, 35, 23, 20, 18, 1, 40, 16

1, 5, 12, 16, 18, 20 | 22 | 23, 25, 28, 33, 35, 40

n = 13 M = 22 Q1 = 14 Q3 = 30.5

(c) 20, 28, 23, 25, 3, 5, 33, 22, 18, 16, 40, 1, 35, 12

1, 3, 5, 12, 16, 18, 20, 22, 23, 25, 28, 33, 35, 40

n = 14 M = (20 + 22)/2 = 21 Q1 = 12 Q3 = 28

(d) 20, 28, 23, 25, 3, 5, 30, 22, 18, 40, 16, 35, 1, 33, 12

1, 3, 5, 12, 16, 18, 20 | 22 | 23, 25, 28, 30, 33, 35, 40

n = 15 M = 22 Q1 = 12 Q3 = 30
1) A certain test is used to measure the reading ability of children. Here are the
scores of 44 third grade students.

40 26 39 14 42 18 25 43 56 27 19

47 19 26 35 34 15 44 40 38 31 46

52 25 35 35 33 28 34 41 49 28 52

47 35 48 22 33 41 51 27 14 54 45

(a) Find the mean, median, five number summary, IQR, and standard deviation of
the children's reading ability.

∑ 𝑥 = 𝑥1 + 𝑥2 + ⋯ + 𝑥44 = 40 + ⋯ + 45 = 1553

∑ 𝑥 2 = 𝑥1 2 + 𝑥2 2 + ⋯ + 𝑥44 2 = 402 + ⋯ + 452 = 60527

1553
𝑥̅ = = 35.30
44

2 ) − (Σ𝑥) (1553)2
2
√(Σ𝑥 60527 −
𝑠= 𝑛 =√ 44 = 11.53
𝑛−1 43
Ordering the numbers, the average of the middle two blanks is 35. 22 numbers on
each side means that the quartiles (26.5 and 34) are also averages of two numbers.
Five number summary is {14,26.5,35,44.5,56}. IQR = 44.5 – 26.5 = 18.

(b) The test scores are redesigned so that the previous scores are doubled, and
everyone gets an additional 10 points added in. How will the quantities in (a) be
affected, without recomputing?

The unit change is 𝑦 = 2𝑥 + 10. Mean, median, min, max, quartiles will be run
through this formula. Standard deviation and IQR will only be multiplied by 2.
2) Refer to problem 1.4:

Statistics Exam Scores:

60 47 82 95 88 72 67 66 68 98 90

77 86 58 64 95 74 72 88 74 77 39

90 63 68 97 70 64 70 70 58 78 89

44 55 85 82 83 72 77 72 86 50 94

92 80 91 75 76 78

(a) Find the mean, median, five number summary, IQR, and standard deviation of
the statistics exam scores. According to the rule of thumb we use, are there any
outliers?

∑ 𝑥 = 60 + ⋯ + 78 = 3746

∑ 𝑥 2 = 602 + ⋯ + 782 = 290524

4746
𝑥̅ = = 74.92
50

(3746)2
√290524 −
𝑠= 50 = 14.2
49

Remember that we have a stemplot, so we’ve already ordered them. Five number
summary is {39,67,75.5,86,98}, so IQR = 86 – 67 = 19.

Outliers are (1) more than 86 + 1.5*IQR = 114.5 or (2) less than 67 – 1.5*IQR =
38.5, so no outliers.

(b) Suppose the teacher adds five points to everyone's scores. What effect will this
have (without recomputing) on the quantities in (a)?

The unit change is 𝑦 = 𝑥 + 5. Mean, median, min, max, quartiles will be run
through this formula. Standard deviation and IQR will be unaffected.
3) Refer to problem 1.6:

(a) Find the mean, median, five number summary, IQR, and standard deviation of
the physical-fitness scores.

Third graders at Roth Elementary School were given a physical-fitness test. Their
scores were:

12 22 6 9 2 9 5 9 3 5 16 1 22 18

6 12 21 23 9 10 24 21 17 11 18 19 17 5

14 16 19 19 18 3 4 21 16 20 15 14 17 4

5 22 12 15 18 20 8 10 13 20 6 9 2 17

15 9 4 15 14 19 3 24

Remember that we have a stemplot, so we’ve already ordered them. Five number
summary is {1,7,14,18.5,24}, so IQR=18.5 – 7 = 11.5.

∑ 𝑥 = 12 + ⋯ + 24 = 832

∑ 𝑥 2 = 122 + ⋯ + 242 = 13572

4746
𝑥̅ = = 13
64

(832)2
√13572 −
𝑠= 64 = 6.6
63

(b) Are there any outliers according to our “rule of thumb”?

Outliers are (1) more than 18.5 + 1.5*11.5 = 35.75 or (2) less than 14 – 1.5*11.5R
= -3.25, so no outliers.
(c) Replacing 24 with a larger number will not affect either the third quartile or the
IQR, so what is the smallest whole number with which we could replace 24 and
have it be considered an outlier?

We have seen that any number larger than 35.75 would be considered an outlier.
36.8, for instance, would be considered an outlier. But the smallest whole number
which would be considered an outlier is 36.

4. The following are the golf scores of 12 members of a women's golf team.

89 90 87 95 86 81 102 105 83 88 91 79

(a) Compute the mean, median, five number summary, IQR, and standard
deviation of the scores. Are there any outliers, according to our rule of thumb?

Five number summary is {79,84.5,88.5,93,105}, so IQR = 93– 84.5 = 8.5.

Outliers are (1) more than 93 + 1.5*IQR = 105.75 or (2) less than 84.5 – 1.5*IQR
= 71.75, so no outliers.

∑ 𝑥 = 89 + ⋯ + 79 = 1076

∑ 𝑥 2 = 892 + ⋯ + 792 = 97156

4746
𝑥̅ = = 89.7
12

(1076)2
√97156 −
𝑠= 12 = 7.83
11

(b) If the 105 is replaced by 200, how will the mean, median, standard deviation,
and IQR be affected?

The mean and standard deviation will be affected by the replacement (mean will be
greater because the 200 will “pull” it, and standard deviation will be greater
because a more extreme replacement has been made; the “average distance from
the center” will increase. Replacing the largest of 12 numbers with a different one
which is still the largest of 12 numbers will not affect median or IQR.
5. Last year a small accounting firm paid each of its five clerks $25,000, two junior
accountants $60,000 each, and the firm's owner $255,000. Find the mean and
median salaries at this firm. How many employees earn less than the mean? Find
the standard deviation and IQR of the salaries. If next year everyone's salaries are
tripled, and they're given a $1000 bonus, how will this affect the mean, median,
IQR, and standard deviations of the salaries?

Data is

25000 25000 25000 25000 25000 60000 60000 255000

Five number summary is {25000, 25000, 25000, 60000, 255000}, so IQR =

60000– 25000 = 35000.

∑ 𝑥 = 25000 + ⋯ + 255000 = 500,000

∑ 𝑥 2 = 250002 + ⋯ + 2550002 = 75,350,000,000

500,000
𝑥̅ = = 62,500
8

(500,000)2
√75,350,000,000 − 8
𝑠= = 79,372.54
7

Seven out of eight employees earn less than the mean (the obvious outlier pulls the
mean up considerably).

The unit change is 𝑦 = 3𝑥 + 1000. Mean, median, min, max, quartiles will be run
through this formula. Standard deviation and IQR will only be multiplied by 3.

Notice that these calculations can be greatly simplified by doing everything in

“thousands of dollars” rather than “dollars”; i.e., a unit change. In this case, of
course, the final formula would be 𝑦 = 3𝑥 + 1, since the quantities on both sides
are “thousands of dollars”.
6. In each case, find a formula for the linear transformation from X to Y and
answer the questions.

(a) You're writing a report on the power of car engines. Your sources use
horsepower X. Your professor requires you to re-express power in watts Y. One
horsepower is 746 watts. What is the power in watts of a 140 horsepower engine?

746 watts
140 hp ∗ = 104,440 watts
1 hp

𝑌 = 746 ∗ 𝑋

(b) Convert speed X in miles/hour to into Y, kilometers/hour. (1 km = 0.62 mile)

What is 65 miles per hour in km/hr's? What is 82 km/hour in miles/hr?

In one hour, we travel:

1 km
65 miles ∗ = 104.65 km
0.62 mile
𝑋
𝑌= = 1.61 ∗ 𝑋, which looks nicer.
0.62
0.62 mile
82 km ∗ = 50.93 miles
1 km
(c) The recommended daily allowance (RDA) of vitamin C is 30 milligrams.
Convert milligrams of vitamin C (X) into Y, percent of RDA.

0 mg would be 0 RDA.

30 mg would be the total proportion of RDA (1).

15 mg would be half the RDA (0.5).

60 mg would be twice the RDA (2), etc.

There is a change of scale (*, ÷) but no shift (+, – ). To convert mg into RDA
we’re simply dividing by the total (30):

𝑝𝑟𝑜𝑝 𝑅𝐷𝐴 = 𝑋/30

To convert to percent, we’re multiplying by 100:

𝑋 10
𝑝𝑒𝑟𝑐𝑒𝑛𝑡 𝑅𝐷𝐴 = 100 ∗ 𝑝𝑟𝑜𝑝 𝑅𝐷 = 100 ∗ = 𝑋
30 3
Note that as usual, percents are easier to talk about informally, but more confusing
in formulas, so we tend to use percents only in casual conversation, not
calculations.

(d) High school students can take SAT's and ACT's, two different standardized
tests. The two exams use two very different grading scales, and yet we can
compare performance by a rule of thumb: multiply an ACT score by 40 and add
150 points to estimate the equivalent SAT score. If a large number of ACT scores
has five number summary 19, 24,28,30,36, mean 27, standard deviation 3, find the
corresponding quantities on the SAT exam. Find a formula for converting SAT
scores to ACT scores.

We are told that:

𝑋 = 𝐴𝐶𝑇, 𝑌 = 𝑆𝐴𝑇

𝑌 = 40𝑋 + 150

Mean, median, min, max, quartiles will be run through this formula. Standard
deviation and IQR will only be multiplied by 40. “”

For instance, if you want to verify for IQR:

𝐼𝑄𝑅𝑌 = (40 ∗ 30 + 150) − (40 ∗ 24 + 150) = 40(30 + 24) = 40 ∗ 𝐼𝑄𝑅𝑋

Solving for X, we get:

𝑌 − 150
𝑌 − 150 = 40𝑋 so =𝑋
40
In “𝑎 + 𝑏 ∗ 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒” form, if you like, this is:

𝑌 150 1
𝑋= − = 𝑌 − 3.75
40 40 40
7. Give an example of a set of twelve numbers (distribution) with IQR = 0, min =
20, max = 50.
We require 12 ordered blanks with:

20 50

The median will be the average of:

20 *** *** 50

The quartiles will be the average of:

20 *** *** *** *** 50

For the 𝐼𝑄𝑅 = 𝑄3 − 𝑄1 = 0, we need 𝑄3 = 𝑄1 .

This must mean that all four of these blanks are the same.

Since they’re in order, all the intervening blanks must be the same. This can be
any particular number between 20 and 50, inclusive. For instance,

20 32 32 32 32 32 32 32 32 50

The remaining blanks can be filled with anything as long as the numbers remain in
order. For instance,

20 25 32 32 32 32 32 32 32 32 47 50

In passing, notice that median and IQR give plenty of room for alternative
examples (compare #8, #11).

8. Give an example of a set of twelve numbers with standard deviation = 0, min =

28

We require 12 ordered blanks with:

28

Standard deviation (“average distance from the center”) being exactly 0 implies
that every distance from the center is 0; i.e., that all the numbers are the same.
(The moment two numbers are different, there is a strictly positive “average
distance from the center”). So the list must be:
28 28 28 28 28 28 28 28 28 28 28 28

Notice that specifying conditions on mean and/or standard deviation gives you a
lot less room to maneuver (compare #7).

9. Consider the coin distribution in 1.3 above. Which would be more useful to
summarize the results from many, many coins: the mean and standard deviation, or
the median and interquartile range? Which will be higher, the mean or median coin
year?

The mean/standard deviation “balance” analysis has the first claim. If there is
something about the list that will render it suspect (outliers, long tails/skewed,
generally asymmetric) then that’s the reason we would prefer the median, IQR,
Five Number Summary “order” analysis. Since this example was very left-
skewed, we’ll go with the latter. The long tail toward smaller numbers will pull
down the mean compared to the median. A coin from ancient Rome will
considerably pull down the average year; one more coin in the small side of the
median has very little or no effect.

10. The ages in years of everyone who visits a certain shopping mall on Saturday
afternoon is strongly right-skewed. (a) How will this characteristic be affected if
we measure in months? (b) Which will be higher, the mean or median age, or will
they be the same?

Unit changes don’t affect the generally shape of a distribution, only how
pronounced it physically looks. Since months result in bigger numbers than years
do, the long tail to the right will look longer. But bear in mind that this is very
cosmetic; how far apart the “tick marks” are placed when you’re drawing it is a
very arbitrary decision. The general shape (whether it possesses the various
characteristics from section 1, unimodal, bimodal, symmetric, right-skewed, left-
skewed, uniform) is not affected.

Since it is right-skewed, the mean will be pulled up by the tail, the median
relatively unaffected by data that is very far away. So the mean is higher.

11. Give an example of a distribution of numbers with mean 15 and standard

deviation 2

This is going to be very hard to do by trial and error. But we can start with any set
of numbers we like and think what unit change to run it through (shift, rescale) to
get what we want. The math is easiest if we start with the simplest mean (0) and
the simplest standard deviation (1). This is achieved by the numbers

mean st. dev.

X −1 0 1 0 1

This mean and standard deviation are easily checked.

Remember that shifts (+, – numbers) do not affect standard deviation. So once the
standard deviation is correct, we can re-center without changing it. So get the
standard deviation correct first. Multiplication has the obvious affect on standard
deviation, so we’ll double every number to get the correct standard deviation:

mean st. dev.

X −1 0 1 0 1
2X −2 0 2 2*0 = 0 2*1 = 2

Now we can re-center. Adding a number will have the obvious effect on mean but
no effect on standard deviation. So we’ll add 15 to change the mean to 15:

mean st. dev.

X −1 0 1 0 1
2X −2 0 2 2*0 = 0 2*1 = 2
2X+15 −13 15 17 2*0+15 = 15 2*1 = 2