1
$\begingroup$

I have run a t-test contrast by dummy coding three groups. Let's call them A, B, and C. The contrast is A and B against C. Now, I want to calculate an effect size for the resulting t-test. I am not sure of the best method. The formula that I have for Cohen's D indicates that the numerator is the difference between two means. I have the mean of C, but not sure what I should use for the mean of A and B. Similarly, the denominator is the pooled standard deviation, not sure what I should be pooling in this situation...Should I be calculating the mean and standard deviation of the combined observations in A and B?

$\endgroup$

1 Answer 1

1
$\begingroup$

I think you can use Cohen's $d'$ defined as $$d'= L_1/\sqrt{MS_{Error}},$$ where $L_1$ is the estimate of the contrast. In your case, this is simply $L_1=\bar{y}_A-\bar{y}_C$ (and $L_2=\bar{y}_B-\bar{y}_C$). This can be found, for example, in Equation 10.11 in Myers et al. (2010, p. 246).

There is probably also an $\eta^2$ for this, but I don't have a formula at hand right now.

Myers, J. L., Well, A. D., Lorch, R. F., (2010). Research design and statistical analysis (3rd ed.).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.