Skip to content

Where to start: running multinomial analyses with LME4

5 messages · João Veríssimo, Karl Ove Hufthammer, Phillip Alday +1 more

#
Dear all,


I am now for the first time needing to run ME analyses with
likert-scale-type data obtained from acceptability judgments in a
linguistic experiment. The responses range between 1 and 5, are ordered,
and will involve continuous predictors as well as dichotomous factors.
Gelman and Hill 2007's section on ordered logistic regression (p.119-124)
does not involve MEs nor are there any worked examples with R coding as far
as I can see in the book.

Thus, I would be very grateful to anyone who could point me to some
literature with worked examples involving coding for ordered logistic
regression, preferably in linguistics. Obviously, if anyone has alternative
ways of approaching this type of data, I am open to suggestions.

Thanks in advance,


Francesco Romano PhD
#
Have a look at the ordinal package and its vignettes. You can fit fit 
mixed-effects ordinal models with the clmm() and clmm2() functions.

For Bayesian mixed-effects models (in brms), these papers will be helpful:
https://doi.org/10.1017/S1366728921000316
https://doi.org/10.1177/2515245918823199

Jo?o
On 18/03/2023 14:44, Francesco Romano wrote:
#
Francesco Romano skreiv 18.03.2023 15:44:
You might find the vignette on the clmm2() function in the ?ordinal? 
package useful:
https://cran.r-project.org/web/packages/ordinal/vignettes/clmm2_tutorial.pdf

(But you should probably use the newer, though very similar, clmm() 
function in the same package instead.)
#
See also 
https://doingbayesiandataanalysis.blogspot.com/2018/09/analyzing-ordinal-data-with-metric.html
On 3/19/23 06:40, Karl Ove Hufthammer wrote:
#
Dear Joao and Phillip,

Thank you so much for your suggestions.
Please keep them coming everyone ?

Francesco
On Sun, 19 Mar 2023 at 16:17, Phillip Alday <me at phillipalday.com> wrote: