F TEST

Dr. Rachita Jayswal

Introduction and Formula
•The F-test is named in the honor of the great statistician R.A. Fisher.
Objective of F-Test
•To find out whether the two independent estimates of population variance differ significantly OR
•To find out whether the two samples may be regarded as drawn from the normal populations having the same
variance.
So, to carry out the test of significance, we have to calculate the ratio of F:
Hence,
F test is based on the ratio of two variances, it is also known as the "Variance Ratio Test."
 Properties of F-distribution
•f-distribution curve is skewed towards the right with range 0 to ∞ and having the
roughly median value 1.
•Value of F will always be more than 0.
•Shape of f-distribution curve is dependent on:
• d.f. of Numerator
• d.f. of Denominator.
•f-distribution curve is never symmetrical, but if d.f. will be increased, then it will be
more similar to the symmetrical shape.
Properties of F-distribution
• Degree of skewness decreases with
increase in degree of freedom (𝑣)
for Numerator (𝑣1 ) &
Denominator (𝑣2).
• Shape of the curve will be more
symmetrical by increase in degree
of freedom.
• In F-test, variance will be compared
from randomly drawn samples and
the observations are independent.
Question 1
Two random samples were drawn from two normal populations and
their values are:
A: 16, 17, 25, 26, 32, 34, 38, 40, 42
B: 14, 16, 24, 28, 32, 35, 37, 42, 43, 45, 47
Test whether the two populations have the same variance at 5% level
of significance.
 Given data:
A: 16, 17, 25, 26, 32, 34, 38, 40, 42 270 363
𝑥ҧ𝑎 = = 30 𝑥ҧ𝑏 = = 33
n=9 9 11
B: 14, 16, 24, 28, 32, 35, 37, 42, 43, 45, 47
n = 11
Let’s take the Hypothesis:
H₀: σ₁² = σ₂² Calculation Table:
Hₐ: σ₁² > σ₂²
A (x - x̄ₐ) (x - x̄ₐ)² B (x - x̄B) (x - x̄B)²
s² = (Σ(x - x̄)²) / (n - 1) 16 -14 196 14 -19 361
For A: 17 -13 169 16 -17 289
s₁² = 734 / (9 - 1) = 734 / 8 = 91.75 25 -5 25 24 -9 81
For B:
26 -4 16 28 -5 25
s₂² = 1298 / (11 - 1) = 1298 / 10 = 129.8
32 +2 4 32 -1 1
v₁ = 11 - 1 = 10
F = s₂² / s₁² = 129.8 / 91.75 34 +4 16 35 +2 4
v₂ = 9 - 1 = 8
F = 1.4147 (Calculated F value) 38 +8 64 37 +4 16
f₀.05 = 3.35 (Tabulated Value)
40 +10 100 42 +9 81
Conclusion:
42 +12 144 43 +10 100
Calculated F value is smaller than tabulated F value, so Null
Hypothesis is accepted. 45 +12 144
Hence, we can say that two populations have the same 47 +14 196
variance. 270 Σ(x - x̄ₐ)² = 734 363 Σ(x - x̄B)² = 1298
Question No. 2
In a sample of 9 observations, the sum of squared deviation of
items from the mean was 64.
In another sample of 11 observations, the value was found to be
88.
Test whether the difference is significant at 5% level.
Given data:- Let’s take the hypothesis that variance of the two sample are not significant:
n = 9, Σ (x - x̄)² = 64 H₀: σ₁² = σ₂²
(sum of the squared deviation) Hₐ: σ₁² > σ₂²
n = 11, Σ (x - x̄)² = 88

F = S₁² / S₂² , S₁² > S₂²

S² = Σ(x - x̄)² / (n - 1)
S₂² = 64 / (9 - 1) = 64 / 8 = 8
S₁² = 88 / (11 - 1) = 88 / 10 = 8.8
f = 8.8 / 8
f = 1.1 (Calculated F value)

v₁ = n₁ - 1 = 11 - 1 = 10
v₂ = n₂ - 1 = 9 - 1 = 8
f₀.₀₅ = 3.35 (Tabulated F value)

Calculated F value is smaller than Tabulated F value, so H₀ is accepted.

Hence, the difference in the variance of two samples are not significant at 5% level of significance.
Question No. 3
A research was conducted to understand whether women have a
greater variation in attitude on political issue than men.
Two independent samples of 31 men and 61 women were used
for the study.
The sample variances calculated were 130 for women and 70 for
men.
Test whether the difference in attitude towards political issues is
significant at 5% level of significance.
Given data:-
S₁² = 130
n₁ = 61 (no. of women)
S₂² = 70 Let’s take the hypothesis that there is no significant difference between the
n₂ = 31 (no. of men) variance of women and men.
H₀: σ² (women) = σ² (men)
Hₐ: σ² (women) > σ² (men)

f = S₁² / S₂²
F = 130 / 70
f = 1.857 (Calculated f value)
v₁ = n₁ - 1 = 61 - 1 = 60
v₂ = n₂ - 1 = 31 - 1 = 30
f₀.₀₅ = 1.74 (Tabulated f value)

Calculated f value is greater than tabulated f value, so H₀ is rejected, and we have to accept our alternative
hypothesis (Hₐ).
Hence, the result of the research supports the belief that women have a greater variation in attitude on
political issues than men.