Name of Quantlet: Marginal Effect for ordinal data

Description: The R code is designed to compute and visualize effect measures for covariates in ordinal data models, with a focus on interpreting effects in extreme response categories, evaluating response styles, and informing the collapsing of categories.

Key functionalities include: 
- Fitting standard cumulative link models (CLMs) under the proportional odds assumption, using functions from packages like ordinal.
- Implementing mixture models—specifically the Combination of Uncertainty and Preference (CUP) framework—to account for uncertainty in respondents’ ratings.
- Computing marginal effect measures that quantify the influence of covariates on each response category, particularly highlighting the underestimation of effects when uncertainty is ignored.
- Visualizing covariate effects through custom plots, including relative effect sizes, partial effects, and response probability profiles across categories.
- Evaluating rates of change to provide interpretable summaries of covariate impacts using real data examples. The code is structured to facilitate comparisons between models with and without uncertainty components and supports inference and interpretation in applied ordinal data analysis. 'It is referred to Iannario, M., & Tarantola, C. (2021). How to Interpret the Effect of Covariates on the Extreme Categories in Ordinal Data Models. Sociological Methods & Research, 52(1), 231-267. https://doi.org/10.1177/0049124120986179 (Original work published 2023)'

Submitted: 20 May 2025

Keywords: 
- ordinal data
- cumulative link models
- cumulative logits
- CUP models
- extreme categories
- marginal effects
- mixture models
- proportional odds
- rating data
- uncertainty

Author: 
- Maria Iannario, Claudia Tarantola