[R-sig-ME] very low conditiona; l/marginal r-squared for a mixed model

[Santosh Srinivas]

Dear List,

In a paper taking a hypothesis testing approach, we are reporting the results based on the following mixed model:

m = glmmTMB::glmmTMB(
      agentic_language ~ time + event + time_since_event + speaker_gender + (time|speaker_id), data = df_regress)

We reported the r-squared using the performance::r2(m) function.

A reviewer has critiqued that conditional and marginal r-squared values being so low (<  0.04) it is a very poor model.

We are not sure if the reviewer is making a valid critique based on r-squared value for a mixed model. Do you too think the mode has no merits?

Thanks for taking the time.

Kind regards
Santosh

	[[alternative HTML version deleted]]