I am new at using generalized liner mixed models (GLMM) and not sure which distribution I should specify for my dependent variable (auditory perception). It's an integer: positive numbers from 1 to 10. The distribution is unimodal and has an obvious negative skew. Any ideas?
Thank you!
I'm not entirely sure but I would try the same as for ordinal logistic regression. Perhaps these notes will help. Best, David Booth
An ordinal logistic model, as David suggested, is presumably to "most correct" model. However, with 10 levels it will be difficult to interpret. I thus think taking the ordinal response as "quasi-continuous" (and pretending that the scale has some evaluated quantitative meaning), you could opt for a beta-regression (after rescaling the response from (1...10) to (0...1)).
You can use Logistic regrssion and you can Download atrical from me in researchgate in my count
With Best Wishes
Could you describe your response variable in more detail? Is it a rating scale? It seems, your response has two bounds (1 and 10), resembling an experimental outcome with number of successes in ten trials, except that you don't have zero.
That means you could try a beta-binomial regression.
As Jochen Wilhelm said, another possibility is to ignore that responses are discrete and go for a beta regression. Note that you'll have to rescale to ]0;1[, excluding the endpoints.
See chapter 7 of my book draft, where I describe these models:
https://www.researchgate.net/project/Book-New-statistics-for-the-design-researcher
Also see the documentation on response distributions of the R package brms:
https://cran.r-project.org/web/packages/brms/vignettes/brms_families.html
It is a rating scale from 1 to 10. Participants had to evaluate a distance between two sounds with 0 indicating no distance (similar sounds) and 10 — a large distance (different sounds). The testing was done in one sitting.
I'd give Beta regression a try. See https://bookdown.org/schmettow/NewStats/GLM.html#beta-regression
Dear Vita Kogan ,
Define clearly the independent variable and the dependent variable under your study. Also, mention whether the variables under study are discrete or continuous.
From the histogram, the distribution is seemed to be approximately exponential. However, it depends upon the type of variable(s) under study.
- Badges
- Science topic
- Similar topics
- Agriculture
- Soil Science