0
$\begingroup$

I have two logistic mixed-effects models, nested within each other and differing in only one fixed variable:

mod1<- glmer(y ~ x1 + x2 + (x1 + x2| subject_ID), data = dat, family = binomial)
mod2<- glmer(y ~ x1 + (x1 + x2| subject_ID), data = dat, family = binomial)

I use the R function anova to compare the two with test="Chisq", which gives me a likelihood ratio test of whether the fixed effect missing from mod2 significantly improves mod1. I would like to calculate the effect size (specifically, Cohen's w) for this test.

How can I accomplish this?

Relatedly, it seems that to calculate Cramer's V or Cohen's w I need to know the sample size, number of rows, and number of columns of my test. What do rows and columns refer to in this case?

$\endgroup$

1 Answer 1

0
$\begingroup$

I have found out here that Cohen's w is also known as phi. Rows and columns refer to contingency tables. Instead, here we care about (one-dimensional) goodness of fit. The correct way to get phi/w is:

w = sqrt(chisq/N)

Where chisq is the statistic obtained with the anova function and N is the sample size.

$\endgroup$

Your Answer

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

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