Scribd
Upload
30 views

F Distribution and F-Statistic

The document discusses Fisher's F distribution and its application in testing whether two sample variances are equal. Specifically: - Fisher's F distribution is used to draw statistical inferences about the ratio of two sample variances from normal populations. - The F distribution table provides critical values that allow one to calculate the probability that two variances are equal, based on the F-statistic and the degrees of freedom. - An example demonstrates calculating the F-statistic from sample data and comparing it to the critical value to determine if there is a statistically significant difference between two variances.

Uploaded by

Learn First
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
30 views

F Distribution and F-Statistic

The document discusses Fisher's F distribution and its application in testing whether two sample variances are equal. Specifically: - Fisher's F distribution is used to draw statistical inferences about the ratio of two sample variances from normal populations. - The F distribution table provides critical values that allow one to calculate the probability that two variances are equal, based on the F-statistic and the degrees of freedom. - An example demonstrates calculating the F-statistic from sample data and comparing it to the critical value to determine if there is a statistically significant difference between two variances.

Uploaded by

Learn First
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

• F Distribution (Fisher’s F Distribution)

✓ Another distribution which plays an important role in connection with sampling


from normal populations.
✓ Fisher’s F distribution is used to draw statistical inferences about the ratio of two
sample variances.
• F Distribution Continue…
✓ The probabilities for F distribution can be used directly from the table which
contains values of type 𝑓𝛼,𝜈1 ,𝜈2 where 𝑓𝛼,𝜈1 ,𝜈2 is such that the area to its right under
the curve of the F distribution with ν1 and ν2 degrees of freedom (see figure) is
equal to α.
✓ That means 𝑓𝛼,𝜈1 ,𝜈2 is such that 𝑃(𝐹 ≥ 𝑓𝛼,𝜈1 ,𝜈2 ) = 𝛼

𝑓𝛼,𝜈1 ,𝜈2 for F distribution


F Distribution Table
• Application of F Distribution:
✓ Check whether two sample variances are equal or not (Ratio of sample variances)

Example 3: Suppose you randomly select 7 women from a normal population of


women, and 12 men from a normal population of men. The table below shows the
standard deviation in each sample and in each population.
Compute f statistic and find out the cumulative probability of women’s to men’s
sample variances.
Solution:
Given, 𝑠1 = 35, 𝑠2 = 45, 𝜎1 = 30, 𝜎2 = 50, 𝑛1 = 7, 𝑛2 = 12
Then, f statistic is:
𝑠12 Τ𝜎12 𝜎22 𝑠12 (50)2 (35)2 1.361
𝑓= 2 2= 2 2= 2 2
= = 1.68
𝑠2 Τ𝜎2 𝜎1 𝑠2 (30) (45) 0.81
According to theorem, f will follow a F distribution with 𝑛1 − 1 i.e. 6 and 𝑛2 − 1 i.e. 11
degrees of freedom.
Now, according to F table, 𝑃(𝐹 ≥ 𝑓𝛼,𝜈1 ,𝜈2 ) = 𝑃(𝐹 ≥ 1.68 ) = 0.22
Therefore , cumulative probability of women’s to men’s sample variances
𝑃(𝐹 ≤ 𝑓𝛼,𝜈1 ,𝜈2 ) = 1 − 𝑃(𝐹 ≥ 𝑓𝛼,𝜈1 ,𝜈2 ) = 1 − 0.22 = 0.78
• F-test (Equality of two variances):
✓ Two random samples having variances 𝑠12 and 𝑠22 , respectively, and size 𝑛1 and 𝑛2 ,
respectively, have been drawn from normal populations (meaning population has
approximately a normal distribution).
𝑠12
✓ Then 𝐹 = will have a F distribution having 𝑛1 − 1 and 𝑛2 − 1 degrees of
𝑠22
freedom. (Remember larger variance should be put in the numerator.)
✓ F-test considers two hypotheses:
1) Null hypothesis: Two sample variances are equal i.e. 𝑠12 = 𝑠22
2) Alternate hypothesis: Two sample variances are not equal i.e. 𝑠12 ≠ 𝑠22
✓ If calculated f-statistic value is greater than or equal to the actual value in F table
with 𝑛1 − 1 and 𝑛2 − 1 degrees of freedom and at significance level 𝛼 (generally
0.05), then we reject the null hypothesis and accept the alternative hypothesis
which concludes that two sample variances are not equal or they differ
significantly.
Example 4: Two samples of sizes 7 and 8, having variances 205 and 121,
respectively, were drawn from a normal population. Check the claim that variances
of the samples do not differ significantly.
Solution:
𝑠12 205
Calculate f statistic value: 𝑓 = = = 1.694
𝑠22 121
Check f value in F table with degree of freedom 𝜈1 = 7 − 1 = 6 and 𝜈2 = 8 − 1 = 7 and
at significance level 𝛼 = 0.05 which gives 3.87.
Clearly, 1.694 < 3.87, which implies we cannot reject the null hypothesis. Therefore, f-
test suggest that variances of given samples are equal or they do not differ significantly.
So what is claimed in the question is right.

You might also like