Source
Adapted from my own handwritten and hand-typed notes from Year 1 to Year 3 of my EEE degree
The sample variance estimates the population variance . When the population is normal, a scaled version of follows a chi-square distribution.
Sample variance
We divide by , not , so that (an unbiased estimator). is the sample standard deviation.
Key theorem
If , then
has a chi-square distribution with degrees of freedom (d.f.).
Chi-square distribution
Main features:
- continuous, only takes values
- not symmetric (skewed to the right), and the shape depends on the degrees of freedom
- table notation: is the value with area to its right, (Table A2)
Because it is not symmetric, lower tail values must be read separately, for example and are not negatives of each other.
Worked example
A sample of 10 from a normal population with . Find roughly.
Solution:
- with 9 degrees of freedom
- from the table, for 9 d.f., so the probability is about 0.05
This theorem is the basis of the Confidence Interval for the Variance.