Archives of Hellenic Medicine

Last update:

05-Jul-2026

Arch Hellen Med, 43(5), September-October 2026, 682-690

SPECIAL ARTICLE

Ordinal logistic regression as a tool to estimate the risk of ranked outcomes

C. Gnardellis,¹ V. Notara,² G. Tzamalouka,³ M. Papadakaki,³ J. Chliaoutakis³
¹Department of Fisheries and Aquaculture, School of Agricultural Sciences, University of Patras, Messolonghi
²Department of Public and Community Health, School of Public Health, University of West Attica, Athens
³Department of Social Work, School of Health Sciences, Hellenic Mediterranean University, Heraklion, Crete, Greece

The use of logistic regression models in data analysis and machine learning has expanded in recent years and has become the preferred choice of researchers in risk assessment studies across various scientific fields. These models are widely applied, from identifying the etiological factors of chronic diseases to assessing the risk of traffic accidents or evaluating credit risk in financial processes. All logistic regression models are natural extensions of the simple binary model, and their interpretation is based on it. Using data from a cross-sectional study on drowsiness while driving and its association with traffic accident risk, this study presents a detailed examination of the two main extensions of logistic regression techniques: multinomial and ordinal logistic regression. Special emphasis is placed on ordinal regression, as the outcome variable in the collision data is measured on an ordinal scale, reflecting an underlying continuous latent variable.

Key words: Car accident data, Collision risk, Multinomial logistic regression, Ordinal logistic regression, Partial proportional odds model (PPOM), Proportional odds model (POM).