easystats / easystats/performance
Interpretation of negative ICC (from variance decomposition) for cumulative (logit) model
Nobody has claimed this yet.
- Dominant language
- R
- Stars
- 1.2k
- Forks
- 109
- Avg merge
- 6h 34m
- Merged PRs (30d)
- 8
Description
I am not sure this is an issue, rather than a difficulty of understanding.
I fitted a cumulative ordered logit model with brms, on data from subjects that have repetitions. I included the subject as a group-level/random effect. I now wanted to quantify how much the group-level effect contributes to the variance in the data and used the performance::icc function. The function returned NA, so I used performance::variance_decomposition instead:
performance::variance_decomposition(fit, robust = TRUE)
The result is:
## Variance Ratio (comparable to ICC)
Ratio: -0.40 CI 95%: [-2.31 0.51]
How can the negative ratio be interpreted? Shouldn't an ICC lie between 0 and 1?
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reviewing performance::icc and performance::variance_decomposition for cumulative ordered logit models fitted with brms, focusing on why icc returns NA and variance_decomposition returns a negative ratio. Reproduce the reported result if possible, then document whether the negative ratio is expected and how it should be interpreted.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- r
- Domain
- data
- Issue type
- Documentation
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 30/100