Chapter 5: The standard deviation as a

ruler and the Normal model p131

• Which is the better exam score?
• 67 on an exam with mean 50 and SD 10
• 62 on an exam with mean 40 and SD 12?

• Is it fair to say:
• 67 is better because 67 > 62?

• 62 is better because it is 22 marks above the mean and 67

is only 17 marks above the mean?

• Key: z-scores.
1
 Effect of a Linear Transformation on
summary statistics
• In a stat course class average is 65 and
standard deviation is 5
• If everyone gets 5 points added, what is the
new class average and the standard deviation?
• New mean = 65 + 5 = 70 = Old mean + 5
• New standard dev = 5 = Old Standard dev

2
 Effect of a Linear Transformation on
summary statistics
• In a stat course class average is 30 and
standard deviation is 5
• If everyone gets their grade multiplied by 2,
what is the new class average and the
standard deviation?
• New mean = 30 x 2 = 60 = Old mean x 2
• New standard dev = 5 x 2 = Old Std dev x 2

3
 Effect of a Linear Transformation on
summary statistics
• Multiplying each observation in a data set by a
number b multiplies the mean, median, by b
and the measures of spread (standard
deviation, IQR) by abs(b) .

• Adding the same number a to each

observation in a data set adds a to measures
of center, quartiles, percentiles but does not
change the measures of spread.

4
 Effect of a Linear Transformation on
summary statistics: Example
• The average salary for an employee of a company is $30,000
per year. This year, management awards the following
bonuses to every employee.
– Holiday bonus of $500.
– An incentive bonus equal to 10 percent of the employee's
salary.
• What is the mean bonus received by employees?
• Answer: Bonus = Salary 10 + 500
100
Ave Bonus = Ave Salary 10 + 500 = 30000 10 + 500 = $3500
100 100

5
 Effect of a Linear Transformation on
summary statistics: Example
• The average salary for an employee of a company is $30,000
per year, with a std dev of $2000. This year, management
awards the following bonuses to every employee.
– Holiday bonus of $500.
– An incentive bonus equal to 10 percent of the employee's
salary.
• What is the standard deviation of employee bonuses?
• Answer: Bonus = Salary 10 + 500
100
Std Dev Bonus = Std Dev Salary 10 = 2000 10 = $200
100 100

6
 Standardizing and z-scores p132
• If x is an observation from a distribution that has
mean  and std. dev.  , the standardized value of
x is x−
z= 
• A standardized value is often called a z-score.
• A z-score tells us how many standard deviations
the original observation falls away from the
mean.
• What are the mean and SD of z?

7
 Standardizing and z-scores p132
• What are the mean and SD of z?
• Answer: No matter what mean and SD x has, z
has mean 0, SD 1.

• Calculating a z-score sometimes called

“standardizing”. Above says why.
• Gives a basis for comparison for things with
different means and sds.
8
 Example on Exam Scores
• Which is the better exam score?
• 67 on an exam with mean 50 and SD 10
• 62 on an exam with mean 40 and SD 12?
• Answer
• Turn them into z-scores:
• For 67, z = x −  = 67 −50 =1.7
 10
• For 62,
−
z = x = 62− 40 =1.83
12
• so the 62 is a (slightly) better performance, relative to
the mean and SD.
9
Density Curve and the Normal Model
p138 , 140
• Density curve is a curve that
– 1) is always on or above the horizontal axis.
– 2) has area exactly 1 underneath it.

• A density curve describes the overall pattern

of a distribution.

10
 Density Curves
• Example: The curve
below shows the
density curve for scores
in an exam and the area
of the shaded region is
the proportion of
students who scored
between 60 and 80.

11
 Normal distributions p156

• An important class of density curves are the

symmetric unimodal bell-shaped
curves known as Normal curves.

• They describe Normal distributions.

• All normal distributions have the same overall

shape.

12
 Normal distributions

• The density curve for a particular normal

distribution is specified by giving the mean  and
the standard deviation .
• The mean is located at the center of the
symmetric curve and is the same as the median
(and mode).
• The standard deviation  controls the spread of a
normal curve.
• There are other symmetric bell- shaped density
curves that are not normal.
13
 Normal distributions

• The normal density curves are specified by a

particular equation. The height of a normal
density curve at any point x is given by

− 2
−1 
x
 
 
 
 

y= 1  

e 2






 2

14
 Example: Standard Normal Tables
• Roma tomatoes have weights that have a
normal distribution with mean 74 grams and
SD 2.5 grams. What proportion of these
tomatoes will weigh less than 70 grams?
• Answer
z = x −  = 70 −74 =−1.60
 2.5

• look up -1.60 in table Z to get 0.0548.

15
Standard Normal Table

16
Standard Normal Table

17
 Example: Standard Normal Tables
• Roma tomatoes have weights that have a
normal distribution with mean 74 grams and
SD 2.5 grams. What proportion of these
tomatoes will weigh less than 70 grams?
• Answer
z = x −  = 70 −74 =−1.60
 2.5

• look up -1.60 in table Z to get 0.0548.

18
 Roma tomatoes Example again
• Roma tomatoes again (mean 74, SD 2.5):
What proportion less than 77.4 grams?
• Answer
−
z = x = 77.4−74 =1.36
2.5

19
20
 Roma tomatoes Example again
• Roma tomatoes again (mean 74, SD 2.5):
What proportion less than 77.4 grams?
• Answer
−
z = x = 77.4−74 =1.36
2.5
• 0.9131

21
 Example Normal Tables
• What proportion of the Roma tomatoes in the
previous question will weigh more than 80
grams? (Mean 74, SD 2.5.)
• Answer
−  = 80−74 = 2.40
z = x
2.5
• Table value ?

22
23
 Example Normal Tables
• What proportion of the Roma tomatoes in the
previous question will weigh more than 80
grams? (Mean 74, SD 2.5.)
• Answer
−  = 80−74 = 2.40
z = x
2.5
• Table value 0.9918
• Proportion weighing more than 80 = 1- 0.9918
= 0.0082 (or 0.82%)
24
 Roma tomatoes Example again
• What proportion of the Roma tomatoes of the
previous two questions will weigh between 70 and
80 grams?
• Calculate the Z-value for both values (70 and 80)
• We calculated them in the previous parts
• For 70, z= -1.60, table gives 0.0548.
• For 80, z= 2.40, table gives 0.9918.
• Proportion between 70 and 80 is
• 0.9918-0.0548=0.9370 (or 93.70 percent)
25
Getting values from proportions

26
 Getting values from proportions
• Use Table Z backwards to get z that goes with
proportion less
• Turn z back into original scale by using
x =  +  z
• Why? Because z = x −
• and solve this for x

27
 Example: Getting values from
proportions
• Newborn babies in Canada have weights that
follow a normal distribution, with mean 3500
grams and SD 500 grams. (The mean is a little
less than 8 pounds.)
• A baby is defined as being “high birth weight”
if it is in the top 2% of birth weights. What
weight would make a baby “high birth
weight”?

28
Example: Getting values from
proportions

29
 Example: Getting values from
proportions
• Ans 2% more = 98% less = 0.9800 less
• z=2.05 (closest)
• Use the formula x =  +  z
𝑊𝑒𝑖𝑔ℎ𝑡 = 3500 + 500 × 2.05 = 4525g or more

30
 Example: Getting values from
proportions
• A baby is defined as being “very low birth
weight” if it is in the bottom 0.1% of birth
weights. What weight would make a baby
“very low birth weight”?

31
Example: Getting values from
proportions

32
 Example: Getting values from
proportions
• 0.1% less = 0.0010 less
• z = -3.09 (I picked the middle one)
• weight = 3500 + 500*(-3.09) = 1955 grams (or
less)

33
Normal quantile (or probability) plots
• A histogram can reveal distinctly nonnormal
features of a distribution.

• If the histogram appears roughly symmetric

and unimodal, we use another graph, the
normal quantile plot as a better way of
judging the adequacy of a normal model

34
 Use of normal quantile plots.
• If the points on a normal lie close to a straight
line, the plot indicates that the data are
normal.

• Outliers appear as points that are far away

from the overall pattern of the plot.

35
Histogram and the nscores plot for data generated from a
Normal distribution ( N(500, 20)) (for 1000 observations )

36
Histogram and the nscores plot for data generated from
a right skewed distribution

37
Histogram and the nscores plot for data generated from
a left skewed distribution

38
 R commands (Data on systolic blood of 500 individuals)
• X <- c(114.1, 121.9, 118.3, 117.3, 124.9, 118.8,…)
• hist(x)
• qqnorm(x)

39
The 68-95-99.7 rule For Normal
Models p142

40
 The 68-95-99.7 rule For Normal
Models p142
• In the Normal distribution with mean  and
std. deviation ,
• Approx. 68% of the observations fall within
one standard deviation of the mean.
• Approx. 95% of the observations fall within
two standard deviations of the mean .
• Approx. 99.7% of the observations fall within 3
standard deviations of the mean.

41
 The 68-95-99.7 rule For Normal
Models p142
• The grades in an exam has a Normal
distribution with mean 65 and std dev 5.

42