Variation F-test

To calculate if there is a significant difference between variation of the two data-sets. It calculates the F-test between this column and the comparing column.

Interpretation

• The smaller the p value is the more likely there is a significant difference in the variation of the data-sets.
• Develve uses the commonly accepted value of p < 0.05 for significance.
• For a good power (0.8 in Develve) the sample size for both data sets must be bigger than the minimum sample size calculated.
• For a good F-test the data-sets must be normally distributed see Anderson Darling normality test.
  • If the data set is not normally distributed see "What to do with not normally distributed data". The Levene test is the not normally distributed counter part of the F-test.

Options

• Bigger/Smaller (one sided test) or not equal test (two sided)

Colors of the cells

• Red
  Data-set is not normally distributed
• Green
  No significant difference
• Yellow
  Significant difference
• Orange
  Sample size to small

Formula Calculating F value

Where s1 is the smallest of the comparing STDEV values.

With the F value and the degrees of freedom can the program calculate the p value.

Sample size

not equal

This results in the degrees of freedom out of the table, the Minimum sample size is

bigger smaller

This results in the degrees of freedom out of the table, the Minimum sample size is

Legend

n = n
= STDEV smallest variation
= STDEV biggest variation
Degrees of freedom

Example

Select Variation test. To use the F-test test first unselect "non normal distributed" when the box is selected the Levene test is calculated. Then select Diff variation.

• The difference in variance between data set A and B is not significant (Row F test p >0.05) and the sample size is to small (Row min Samples 240).
• The difference in variance between data set A and C is significant (Row F test p <0.05) and the sample size is big enough.
• The difference in variance between data set A and D is not significant (Row F test p >0.05) and the sample size is big enough.

Data file

External links

• http://en.wikipedia.org/wiki/F-test_of_equality_of_variances
• http://en.wikipedia.org/wiki/F-test
• http://www.itl.nist.gov/div898/handbook/prc/section3/prc32.htm
• http://www.itl.nist.gov/div898/handbook/eda/section3/eda359.htm