Basic Concepts
As in Levene’s test, for O’Brien’s test of the homogeneity of group variances, we perform ANOVA on a transformation of the group data. For this test, we use the following transformed data uij where xij is the score of the ith element in the jth group,
SSj is the sum of the squared deviations from the jth group mean, and
nj is the number of elements in the jth group.
Example
Example 1: Repeat Example 1 of Levene’s Test (the data is repeated on the left side of Figure 1).
The transformation of the data is shown on the right side of Figure 1.
Figure 1 – O’Brien’s transformation
For example, cell G6 contains the formula
=IF(ISNUMBER(B6),B$14*(B$14-1.5)*(B6-B$15)^2/((B$14-1)*(B$14-2))-B$16/(2*(B$14-2)),””)
We can now perform ANOVA on the transformed data using the Real Statistics One Factor Anova data analysis tool (see Confidence Interval for ANOVA) as shown in Figure 2.
Figure 2 – O’Brien’s test
Since p-value = .971311 (cell Q14), we can’t reject the null hypothesis that the group variances are equal. In fact, we conclude that the homogeneity of variances assumption is quite likely to hold.
Worksheet Function
Real Statistics Function: The Real Statistics Resource Pack provides the following worksheet function:
OBrienTest(R1) = p-value of O’Brien’s test for the data in R1
This function ignores blank entries in R1.
For Example 1, the output from the formula =OBrienTest(B6:E13) is .971311.
Ramsey’s Conditional Test
Ramsey noted that when the group kurtosis < 0 then O’Brien’s test tends to be best, while if the group kurtosis ≥ 0 then the Brown-Forsythe test (i.e. Levene’s test using deviations from the median) tends to be best.
Examples Workbook
Click here to download the Excel workbook with the examples described on this webpage.
References
Abdi, H. (2007) O’Brien test for homogeneity of variance
https://www.utdallas.edu/~herve/Abdi-OBrien2007-pretty.pdf
Ramsey, P. H. (1994) Testing variances in psychological and educational research
https://psycnet.apa.org/record/1994-28101-001
Wang, Y. et al. (2016) Comparing the performance of approaches for testing the homogeneity of variance assumption in one-factor ANOVA models
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5965542/#bibr32-0013164416645162