This is a question for Cochrane. You could ask either a tech service there. They are quick most of the time. You could also ask it in a forum at:

https://training.cochrane.org/trainers-hub

Hello Laura,

You didn't indicate what your preferred ES metric was. However, the conversion of an observed one-way F (especially with just two levels for the effect/comparison) is pretty easy.

F(1, 2345) = 16.2 equates to:

1) partial eta-squared (or R-squared) = .0069;

2) Mahalanobis D-squared = .0276 (take square root for Cohen's d = 0.1661)

3) Cohen's f = .0831

Good luck with your work.

Hi there, thanks for providing quick answer. I have similar set of data such as F stat within the group and between the group e.g (ANOVA: F5,35 = 1.48, P = 0.22), treatment arm sample :10, control arm sample 9. Could you explain the whole process? David Morse

Hello Ramsagar,

F(5, 35) = 1.48, converts to:

1. Wilks' lambda = .8255;

2. Partial eta-squared (or R-squared) = .1745;

3. Cohen's f = 0.4598

Mahalanobis D-squared isn't appropriate, since it is a measure of distance between two centroids/points in space, and your question is asking about a comparison among six batches.

Attached is a brief explanation of how to elicit some common ES values from the results of an anova or manova F-ratio. I generated this a long time ago and used it as a handout when I taught a multivariate statistics course.

Good luck with your work.