Tootfinder

Sensitivity analysis method in the presence of a missing not at random ordinal independent variable
Abdoulaye Dioni, Alexandre Bureau, Lynne Moore, Aida Eslami
https://arxiv.org/abs/2506.24025

Sensitivity analysis method in the presence of a missing not at random ordinal independent variable
Data analysis often encounters missing data, which can result in inaccurate conclusions, especially when it comes to ordinal variables. In trauma data, the Glasgow Coma Scale is useful for assessing the level of consciousness. This score is often missing in patients who are intubated or under sedation upon arrival at the hospital, and those with normal reactivity without head injury, suggesting a Missing Not At Random (MNAR) mechanism. The problem with MNAR is the absence of a definitive analys…