VARIANCE ( 2) AND STANDARD DEVIATION () OF A DISCRETE PROBABILITY
DISTRIBUTION:
The expected value [E(x) or gives one an idea of the central
tendency or “average” of a probability distribution. However, two
distributions may have the same numerical expected (average) value
but be entirely different in spread. To measure dispersion of a random
variable’s probability distribution, we use a measure called variance
(denoted by 2).
Variance 2 = E(x-)2 = (x )
all x
2
f ( x)
Computational formula: 2 E(x 2 ) 2
Where E(x2)= x
all x
2
f ( x)
The standard deviation (denoted by ) is the square root of
variance:
Computational formula: E ( x 2 ) 2
Let’s have an example whre three probability distributions that
have the same mean but different standard deviations:
Random variable 1: the result of a fair die.
Probability distribution 1:
X 1 2 3 4 5 6
P(x 1/ 1/ 1/ 1/ 1/ 1/
) 6 6 6 6 6 6
Expected value E(x) = (1+2+3+4+5+6)x 1/6 Variance 2 =
E(x2)-2
=3.5 = (12+22+32+42+52+62)x1/6
-3.52
= 2.916667
Standard deviation =
sqrt(2.916667) =
2.3868.
Random variable 2: the result of a die that is not fair, as in the
table below.
Probability distribution 2:
X 1 2 3 4 5 6
P(x 0.0 0.1 0.3 0.3 0.1 0.0
) 5 0 0 0 0 5
Expected value E(x) = (1+6)x0.05 Variance 2 = E(x2)-2
+(2+5)x0.10 = (12+62)x0.05
+(22+52)x0.10
+(3+4)x0.30 +(32+42)x0.30 – 3.52
=3.5 = 1.25
Standard deviation = sqrt(1.25)
= 1.118.
Random variable 3: the result of a die that is not fair, as in the
table below.
Probability distribution 3:
X 1 2 3 4 5 6
P(x 0.3 0.1 0.0 0.0 0.1 0.3
) 0 0 5 5 0 0
Expected value E(x) = (1+6)x0.30 Variance 2 = E(x2)-2
+(2+5)x0.10 = (12+62)x0.30
+(22+52)x0.10
+(3+4)x0.05 +(32+42)x0.05 – 3.52
=3.5 = 4.85
Standard deviation = sqrt(4.85)
= 2.2023.
Summary Statistics
Distri Avera Varian StdD
b ge ce ev
1 3.5 2.92 1.707
8
2 3.5 1.25 1.118
0
3 3.5 4.85 2.202
3
From the summary statistics table: one can see that all three
distributions has the same expected value E(x) = 3.5. But their
variances and standard deviations are different.
Notice that for distribution 1, the probabilities are all uniform.
For this reason, the midpoint 3.5 is the expected value or average. Its
standard deviation is 1.7078.
On the other hand, distribution 2 has its probabilities clustered
around the midpoint 3.5. In fact, the symmetry in the distribution
exists such that the distribution is not spread out evenly across the 6
possible values of X. The standard deviation is consequently lower at
1.118.
Distribution 3 has the highest dispersion from the average .
This is because the larger probabilities are placed farther from =3.5,
in contrast to those of distribution 2. Distribution 3 consequently has
the largest standard deviation among the three distributions.
Visual graphs of the probability distributions can be created :
Graph for probability distribution 1
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
1 2 3 4 5 6
P(x) 0.1667 0.1667 0.1667 0.1667 0.1667 0.1667
Graph for probability distribution 2
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
1 2 3 4 5 6
P(x) 0.05 0.1 0.3 0.3 0.1 0.05
Su
Graph for probability distribution 3 S
Summary Statistics
0.4
0.35 Distri Avera Varian StdD
0.3
b ge ce ev
0.25
0.2 1 3.5 2.92 1.707
0.15 8
0.1 2 3.5 1.25 1.118
0.05 0
0 3 3.5 4.85 2.202
1 2 3 4 5 6
3
P(x) 0.3 0.1 0.05 0.05 0.1 0.3
Distributions with lower standard deviations (and variances)
indicate that the probability values are higher for random variables
close to the expected value or average. This also means that the
probabilities are less spread out from the average. Higher standard
deviations (and variances) indicate higher dispersion of probabilities
and higher probability values for random variables away from the
mean.
As a practical measure, larger standard deviations mean that a
random variable’s probability distribution cannot be a reliable indicator
of the value per observation. Conversely, lower standard deviations
show high reliability of the expected value’s predictability per
observation. From the three distributions shown, the die that is
modeled by distribution 2 is more likely to have a 3.5 value per throw
than the die modeled by distribution 1 and 3. Distibution 3 is less
likely to be 3.5 per throw when enough observations (or throws) are
made.
Put another way, if you use one of the die in a game of
accumulated values like in Monopoly™, you would be more likely to
advance consistently with a pace of 3.5 steps per throw when you use
the second die. When the fair die (die no. 1) is used, you could find
your advances to have some inconsistency (not always 3 or 4), like
normal die throws. When you use the 3 rd die with the highest standard
deviation, you should expect extreme values (i.e. 6 and 1) to come out
more often than the middle values. Extremes of fortunes are
expected with die no. 3. Higher predictability should be experienced
with die no. 2.