Fisher's Exact Test

Posted in Statistics, Total Reads: 1408
Advertisements

Definition: Fisher's Exact Test

Fisher’s exact test is a test for finding out the statistical significance, generally employed for small sample sizes. It is used mostly in analyzing the contingency tables. The major advantage of this test is that the amount of deviation from null hypothesis can be obtained without any approximation. When we classify objects in two different ways, Fisher’s test uses that data to find out dependency or contingency of one classification on the other.

E.g.: Analyzing a group of college students classified as male and female and also as those who use their own vehicle to come to college and those who don’t. The data may be:

 

Boys

Girls

Total

Not own vehicle

1        (a)

2        9 (b)

10

Own vehicle

11    c)

12    3 (d)

14

Total

12

12

24 (a +b +c +d=n)

 

According to Fisher’s formula, to find out the association between these classifications, we can use the probability of getting such a set of values using the hyper geometric distribution:

p   =           (a+b)! * (c+d)! * (a+c)! * (b+d)!        / a! * b! * c! * d! * n!

! stands for factorial sign.

Hence in this example, p = (10! * 14! * 12! * 12! ) / (1! * 9! * 11! * 3! * 24! ) = 0.001346076

Search & Explore : Management Dictionary

Browse the definition and meaning of more terms similar to Fisher's Exact Test. The Management Dictionary covers over 7000 business concepts from 6 categories.

Advertisements



Share this Page on: