MEASURES OF CENTRAL

TENDENCY &
VARIABILITY
Agenda

• Measures of
central
tendency
• Mode,
Descripti median, mean
ve • Measures of
Statistics variability
• Range,
variance,
standard
deviation
Descriptive
Statistics
Statistical procedures that
summarize and describe
data
1. Central tendency –
identifies typical score;
or
2. Variability – identifies
how much data varies
around typical score
CENTRAL TENDENCY
Measures of Central
Tendency

Tell us the value...

That is the most common/typical

Around which other values cluster

Why??

Because measures of central

tendency...

Help researchers efficiently

describe data

Help us understand the meaning of

individual scores within a dataset

 Central
Tendency

Example – how many siblings do you

have?

0, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 3, 3, 3, 3,

3, 4, 5, 6
Central
Tendency
1) The mode

Value with the greatest frequency

 Central
Tendency

Example – how many siblings do you

have?

0, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 3, 3, 3, 3,

3, 4, 5, 6
Central
Tendency
1) The mode

Value with the greatest frequency

Mode for number of siblings = 1

Central
Tendency
1) The mode

Distributions can have 1 or more

mode
Unimodal  1

Bimodal  2

Multimodal  3 or more
Central Tendency

2) The median (Mdn)

The middle value

Value at 50th percentile

If even # of values, median = sum of

2 middle values/2
 Central
Tendency

Example – how many siblings do you

have?

0, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 3, 3, 3, 3,

3, 4, 5, 6
Central
Tendency
2) The median (Mdn)

The middle value

Value at 50th percentile

If even # of values, median = sum of

2 middle values/2

Median for number of siblings = 2

Central
Tendency
2) The mean ()

The average of all values

Mean for number of siblings =

 Central
Tendency

Example – how many siblings do you

have?

0, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 3, 3, 3, 3,

3, 4, 5, 6
Central
Tendency

Calculating
mean # of
siblings
Central
Tendency
2) The mean ()

Journal articles

APA style (according to the

Publication Manual of the American

Psychological Association)

Mean represented as M instead of

SING
IT
WITH
ME…
Which measure to use?

Depends on…

1. The measurement level used

2. The shape of the frequency

distribution of the scores

Which measure to
use?
Measur Pros Cons
e
Mode  Easy to use  Jumps around in small
samples
 Straightforward

Median

Mean
 Example
Our values = 1, 1, 5, 8, 12

What is the mode?

Our values = 1, 5, 8, 12, 12

What is the mode?

Which measure to
use?
Measur Pros Cons
e
Mode  Easy to use  Jumps around in small
samples
 Straightforward  Possibility of no mode
 Possibility of multiple
modes
Median  Easy to use

 Not affected by outliers

or skew
Mean
 Example

Our values = 1, 3, 4, 5, 6, 7, 9

What is the median?

What is the mean?

 Example

Our values = 1, 3, 4, 5, 6, 7, 9, 62

What is the median?

What is the mean?

What is Skew?
Characteristic of distribution  describes

shape

Negatively skewed

Many higher values

Positively skewed

Many lower values

Exampl
es of
Skew
Distributio
ns
Normal
Distribution
Median =
mean = mode
Distributio
ns

Positively
Skewed
Median lower
than mean
Distributio
ns

Negatively
Skewed
Median
higher than
mean
Which measure to
use?
Measur Pros Cons
e
Mode  Easy to use  Jumps around in small
samples
 Straightforward  Possibility of no mode
 Can be used at all  Possibility of multiple
measurement levels modes
Median  Easy to use
 Not affected by outliers
or skew
Mean  Incorporates all values;  Affected by outliers and
most sensitive to change skew
 Most useful as an
inferential statistic
 Level of Measurement Considerations

Nominal variable = favourite movie

genre
Value Frequency (f)
Can you calculate the
“Rom com” mean? The
10
Comedy 9
median? The mode?
Mystery 3
Action 7
Horror 5
Other 3
Which measure to use?

Recommended:
Scale of Measure of
Measurement Central
Tendency
Nominal Mode
Ordinal Median
Interval/Ratio* Mean
* If skewed distribution, use median
 CLASS ACTIVITY

1. Get into groups

2. Choose either an interval/ratio variable that is
relevant to you (e.g., number of pets)
3. Have everyone in your group provide a value
(e.g., answer question)
4. As a group, identify the (1) scale points, (2)
mode, (3) median, and (4) mean for your
variable
5. Present to class
Variability

Degree to which values are clustered

around a central value (i.e., measure of

central tendency).

The more scores differ  the more they differ from

measure of central tendency  more variability

Variabilit
y
Variabilit
y
Range
Interval that contains 100% of

observed scores

Presented as:

Xhighest – Xlowest
Range
Drawbacks

Only dependent on 2 values

Relatively unstable measure of variability

Sensitive to outliers

Increased risk with increasing sample size

MEASURES OF
VARIABILITY
Standard Deviation
More stable and most popular measure

of variability

Square root of the variance of a

distribution

Variance - the average of the squared

differences from the mean

Population Variance
 Sum of Squares
Population Variance (SS)
 Population Standard
Deviation

Where X = the individual scores of members in a population

μ = the mean of the scores in the population
Npopulation = the total number of scores in a population
 Population Course
Standard Grades
Deviation
100%
95% Mark
90%
85%
Nadi
80%
75%
a
70% Lynn
65%
Raymond
60%
55%
50% Marilyn
45%
40%
Series1
Population Standard
Deviation
Grades = 95, 85, 73, 65, 53

1. Find the mean

Population Standard
Deviation
Grades = 95, 85, 73, 65, 53

1. Find the mean  74.2

 Population Course
Standard Grades
Deviation
100%
95% Mark
90%
85%
Nadi
80%
75%
a
70% Lynn
65%
Raymond
60%
55%
50% Marilyn
45%
40%
Series1
Population Standard
Deviation
Grades = 95, 85, 73, 65, 53

1. Find the mean  74.2

2. Subtract each value from mean

Population Standard
Deviation

Where X = the individual scores of members in a population

μ = the mean of the scores in the population
Npopulation = the total number of scores in a population
 Population
Standard Course
Deviation Grades
100%
95% Mark
90%
85%
Nadi
80%
75%
a
70% Lynn
65%
Raymon
60%
55% d
50% Marilyn
45%
40%
Series1
Population Standard
Deviation
Grades = 95, 85, 73, 65, 53

1. Find the mean  74.2

2. Subtract each value from mean

3. Square each value and sum them together

4. Divide by sample size

5. Take square root of this number

Population Standard
Deviation

Where X = the individual scores of members in a population

μ = the mean of the scores in the population
Npopulation = the total number of scores in a population
Population Standard
Deviation

 Population Course
Standard Grades
Deviation
100%
95% Mark
90%
85% 1 SD Nadi
80%
14.73 a
75%
70% Lynn
1 SD
14.73 Raymond
65%
60%
55%
50% Marilyn
45%
40%
Series1
Population Standard
Deviation

Why square differences?

If we add up deviations from , sum

= 0.
 Population Course
Standard Grades
Deviation
100%
95% Mark
90%
85%
Nadi
80%
75%
a
70% Lynn
65%
Raymon
60%
55% d
50% Marily
45% n
40%
Series1
Problem

• What if we do not know population mean (μ)?

Solution
• We can use sample statistics to estimate

population variance and standard deviation

Estimated Population
Variance
Estimated Population
Standard Deviation
 Why N – 1?

• Scores in a sample  usually less

variable than scores in population

• Thus, estimations based on sample

underestimate σ2 and σ
• N – 1 helps account for this
 Standard Deviation -
Interpretation
Average of how much scores differ from mean

Using original units

In majority of cases, most scores fall within

1 s above or below mean…

Thus, can calculate range within most scores lie
 Population Course
Standard
Grades
Deviation
100%
95% Mark
90%
85% 1 SD Nadi
80%
14.73 a
75%
70% Lynn
1 SD
14.73 Raymon
65%
60%
55% d
50% Marily
45% n
40%
Series1
More Practice…

Find the s for each distribution of scores…

Distribution
#1 #2 #3
5 4 1
5 6 4
5 4 7
5 10 10
5 8 15
Standard Deviation in
Research Papers
APA style (according to the Publication

Manual of the American Psychological

Association

Standard deviation represented as

SD instead of s

E.g., Mage = 20 years (SD = 1.5)

Which Measure of Variability to Use?

Scale of Measure of Measure of

Measureme Central Variability
nt Tendency
Nominal Mode None
Ordinal Median Range
Interval/Ratio* Mean Standard
deviation
* If skewed distribution, use median and range