MEASURES OF VARIABILITY OR DISPERSION

Measures of Variability or Dispersion are measures of the average distance of each observation from the
center of the distribution. They measure the homogeneity or heterogeneity of a particular group.

A small measure of variability would indicate that the data are

1. Clustered closely around the mean;
2. More homogeneous;
3. Less variable;
4. More consistent and;
5. More uniformly distributed.
Consider the following sets of grades in Mathematics of two groups of 5 students each:
Male Group Female Group
Vincent Troy:70 Marianne Joy: 82
Kensly Kyle:95 Kwenn Anne: 80
Patrick:60 Patricia: 83
Gabriel: 80 Olivia: 81
Jerod: 100 Lurice: 79
Mean: 81 Mean: 81
The mean grade of both groups is 81. By just looking at their mean grade, we can only conclude that both
groups performed equally well in the said test, but this does not explain how far apart the grades are from
one another. Let us picture the position of each grade in a number line.

MALES

! ! ! ! ! !
60 70 80 90 95 100
FEMALES

! ! ! ! !
79 8081 82 83
Notice that the grades of the males are far apart from each other, while the grades of the female are more
compressed or clustered together. Thus the measure of the center of the distribution is of little help in
describing and comparing these two sets of data. By getting the average distance of each item from the
center of the distribution, the group can be described more completely and, likewise, similarities and
differences can be easily identified.

DIFFERENT MEASURES OF VARIABILITY

1. RANGE
2. AVERAGE DEVIATION or MEAN ABSOLUTE DEVIATION
3. STANDARD DEVIATION
4. VARIANCE
Other Measures of Variability

1. Interquartile Range
2. QUARTILE DEVIATION or Semi-Interquartile Range
3. The 10-90 Percentile Range

THE RANGE
The range refers to the difference between the highest and the lowest score. If the highest score is 25
and the lowest score in a distribution is 10, then the range is equal to 25-10 = 15. This range is specifically
called as the exclusive range. If the difference between the exact lower limit of the lowest score and the
exact upper limit of the highest score is solved, the result is called the inclusive range. Using the same
example, we get 16 from 25.5-9.5. The inclusive range is simply determined by adding 1 to the exclusive
range. Generally, the exclusive range is used for ungrouped data while the inclusive range is used for grouped
data.
The range is the easiest and simplest to determine among the measures of variability because it
depends only on the pair of extreme values. However, it is also the most unstable because its value easily
fluctuates with the change in either of the highest or lowest scores. It is also considered the most unreliable
because it does not give the dispersion or spread of the scores in between the two extreme values. A more
reliable measure should involve all the values in a distribution to provide us an adequate spread of all the
scores from the average.
Example 1. Find the range of the grades in Math of three two groups of students in the preceding example.
Male: 100-60 = 40
Female: 83-79=4
The range of grades of the male group is 40 while that of female group is 4. This shows that the grades of the
males are scattered while the grades of the female group are close to each other. It shows further that
females are more homogeneous than the males in their math ability.
Uses of the Range
The range may be computed and used when
1. The spread or scatter of the scores is all that is wanted;
2. The scores are very few;
3. There are extremely low and extremely high scores that there are big gaps between the scores; and
4. A rough comparison of the variability of two or more groups is desired. A group whose range is
smaller is more homogeneous than a group with a bigger range.
THE INTERQUARTILE RANGE AND QUARTILE DEVIATION/SEMI-INTERQUARTILE RANGE
The interquartile range is the distance between the quartiles Q1 and Q3. It includes the middle 50% of
the scores. But it is the semi-interquartile range called quartile deviation which is more commonly used as a
measure of dispersion or variability.
Q3−Q 1
IQR = Q3- Q1 SIQR =
2
COMPUTATION OF THE IQR & SIQR (AVERAGE DEVIATION)
The class frequency distribution of test scores in Math 2.
Class Interval f <cf
95 – 99 2 50
90 - 94 4 48
85 - 89 7 44
80 - 84 9 37
75 - 79 10 28
70 - 74 8 18
65 - 69 6 10
60 – 64 3 4
55 - 59 1 1
n=50
Solve for Q1: Solve for Q3:
xn/4 = 1(50) / 4 = 12.5 xn/4 = 3(50) / 4 = 37.5
<cf = 10 <cf = 37
f =8 f =7
i =5 i =5
lb = 69.5 lb = 84.5

Substitute the values in the formula: Substitute the values in the formula:

( ) ( )
xn xn
Q1 = lb + 4
−¿ cf
f
i (
= 69.5 +
12.5−10
8 ) 5 Q3 = lb + 4
−¿ cf
f
i
= 84.5 +( 37.5−37
7 ) 5

= 69.5 + 1.56 = 71.06 = 84.5 + 0.36 = 84.86

Solve for IQR:
IQR = Q3 – Q1 = 84.86 – 71.06 = 13.8
Solve for SIQR:
Q3−Q 1 84 . 86−71. 06
SIQR = = = 6.9
2 2
USES OF THE QUARTILE DEVIATION
The quartile deviation/semi-interquartile range may be computed and used under the following
conditions:
1. When the measure of central tendency is the median. The quartile deviation is used with the median
when classifying students.
 2. When the homogeneity or heterogeneity of a group is to be determined. When quartile deviation is
small the group is more or less homogeneous but when the quartile deviation is large the group is
more or less heterogeneous.
3. When there are extremely high and low scores especially when there are big gaps between scores.
This is because other measures of variability, especially the standard deviation, are seriously
disproportionate under this condition.
4. When the main concern is the concentration of the middle 50% of the scores around the median.
THE AVERAGE OR MEAN ABSOLUTE DEVIATION (MAD)
Mean Absolute Deviation is the average of the summation of the absolute deviation of each
observation from the mean.
x is a value or score from the raw data
x /¿
MAD = Σ/ x− ¿, where x is the mean
N
N is the total number of scores

Example 1. A. Find the mean absolute deviation of the male group in example 1.
Solution: The mean of the male group is 81.
Score Mean /x- x /
x x
70 81 11
95 81 14
60 81 21
80 81 1
100 81 19
Σ/x- /=66
x

x /¿ 66
MAD = Σ/ x− = ¿= 13.2
n 5

B. Find the mean absolute deviation of the female group.

Solution: The mean of the female group is 81.

Score Mean /x- x /
x x
82 81 1
80 81 1
83 81 2
81 81 0
79 81 2
Σ/x- x / =6

x /¿ 6
MAD = Σ/ x− = ¿= 1.2
n 5

The male group has a MAD of 13.2 while the female group has 1.2. These results confirm our earlier findings
using the range that the female group is more homogeneous than the male group.

VARIANCE
Variance is the average of the squared deviation from the mean. Formulas for finding the variance for
ungrouped data are shown below
2
Σ /x −x /¿
Population Variance : Vp= ¿
n
2
Σ /x −x /¿
Sample Variance: VS= ¿
n−1
STANDARD DEVIATION
Standard Deviation is the square root of the average deviation from the mean, or simply the square
root of the variance.

√
2
Population Standard Deviation: SDp= Σ ¿ x−x /¿ ¿
N
 √
2
Sample Standard Deviation: SDs = Σ ¿ x−x /¿ ¿
n−1

Example 4. Find the variance and the standard deviation of male groups in Example 1.
Male group:
x x x- x (x- x )2
70 81 -11 121
95 81 14 196
60 81 -21 441
80 81 -1 1
100 81 19 361
Σ (x- )2= 1120
x

a. Treating the data as population, the variance and the standard deviation is
2
Σ /x −x /¿ 1120
Vp = = ¿ = = 224 square units
n 5

SDp = √ 224 = 14.97 units

b. Treating the data as sample, the variance and the standard deviation is:
square units

2
Σ /x −x /¿ 1120 1120
Vs = ¿ = = = 280 square units
n−1 5−1 4
SDs = √ 280 = 16.73 units
 Find the variance and the standard deviation of the female group.
An equivalent formula gives another way of computing the standard deviation. It does away with
computing for the mean and the deviation of the scores. We can use the equivalent formula
below.

√
2
2 (Σ x)
SDs = Σx −
n
n –1
Male Group:
x X2
70 4900
95 9025
60 3600
80 6400
100 10,000
Σx=405 Σx2= 33925

√ √ √
2 2 ❑
2 (Σx) (405) 164025
SDs = Σx − = 33925− = 33925−
n 5 5
n –1 5–1 4
=
√ 33925−32805
4
=
√ 1120
4
= √ 280 = 16.73 units

variance = 280, standard deviation = 16.73

STANDARD DEVIATION AND VARIANCE OF GROUPED DATA

 √ ( )
' 2 ' 2
Σf (d ) Σf d
SD = i –
n n
Example: The grouped data which gives the wages per day of the laborers in a certain construction.
Wages No. of Laborers (f) d’ fd’ f(d’)2 or d’(fd’)
80-84 8 2 16 32
75-79 12 1 12 12
70-74 16 0 0 0
65-69 13 -1 -13 13
60-64 9 -2 -18 36
’
Σf= 58 Σfd = -3 Σf(d’)2= 93

√ ( )
' 2 ' 2
Σf (d ) Σf d
SD = i –
n n

√√ ( )
2
93 −3
SD = 5 −
58 58
= 5 1.603−(−0.052 )2
= 5√ 1.603−0.003
= 5 √ 1.6
= 5 (1.265)
SD= 6.325

Another formula:
Σ f (x−x )2
√
2
Vp = and SDp = Σ f x−x
N N

Σ f (x−x )2
√
2
Vs= = and SDs = Σf x− x
n−1 n−1
Example: The grouped data which gives the wages per day of the laborers in a certain construction.
Wages No. of Laborers Mean x- x f(x- x )2
(f) X fX x (x- x )2
80-84 8 82 656 71.74 10.26 105.2676 842.1408
75-79 12 77 924 71.74 5.26 27.6676 332.0112
70-74 16 72 1152 71.74 0.26 0.0676 1.0816
65-69 13 67 871 71.74 -4.74 22.4676 292.0788
60-64 9 62 558 71.74 -9.74 94.8676 853.8084
Σf= 58 fx=4161 2321.1208
Mean = Σfx/ N =4161/58=71.74

Population Variance Population Standard Deviation

√
2
Vp = Σf (x- x )2 / N and SDp = Σf x− x =√ 40.02=6.326
N
= 2321.1208/ 58
= 40.02

Sample Variance Sample Standard Deviation

√
2
Vs= Σf (x- x )2 / n-1 and SDs = Σ f x−x = √ 40.72=6 .38
n−1
= 2321.1208 / 57
= 40.72

Exercises:
1. The IQ’s of 5 members of the family are 108, 112, 127, 118 and 113.
Find the
a. Range b.. Variance c. Standard Deviation

2. Below is a frequency distribution showing the result of an IQ test of a group of students in a certain
college.
Classes f
80-85 2
 86-91 8
92-97 19
98-103 21
104-109 25
110-115 52
116-121 12
122-127 11
Find: a. Range b .Mean Absolute Deviation
c. Population Variance d. Population Standard Deviation

3. For 108 randomly selected high school students, the following IQ frequency distributions were
obtained.
Class limits frequency
90-98 8
99-107 22
108-116 43
117-125 28
126-134 9
N=
Find the population variance and population standard deviation.

COEFFICIENT OF VARIATION
To relate the measure of dispersion to its average and to convert it to percent form, the
standard deviation is divided by the arithmetic mean. Stating this measure in percentage
form solves the problem presented by differing units. The resulting measure developed by
Pearson is known as the Coefficient of variation.
CV= (SD/mean)100%

Other comparative coefficient of dispersion may be computed when using the other measures
of dispersion:

Quartile Coefficient of Dispersion

CVq= (Q3- Q1) / (Q1+Q3) x 100%