|
This formula for a confidence interval for a variance assumes values are normally distributed, and
compares the Sum-of-Squares (the numerator values) to values from a Chi-square table with n - 1 degrees of freedom.
Bootstrap PLUS .... an alternative approach is to use the computer to take repeated random
samples from sample data (with replacement), calculate a variance for each sample, and do this many times (e.g. 1000 times). Finally, sort the variances, and use the middle 95% as the boundaries for a bootstrap 95% confidence interval. |