Abstract:
This article discusses the regression models with ordinal respon that are adjacent
categories (AC) model and proportional odds (PO) model. The estimated regression
coefficient is obtained through maximum likelihood which is used to calculate the
relative risk for each model. Futhermore, the likelihood ratio test is used to check
the significance of explanatory variable in the model.
Data analysis and simulation have been conducted using R version 3.4.3. We use
a small cell lung cancer data to check the relative risk of tumour respon to patients
receiving one of two therapy strategies and sex of patients. Simulation shows that
AC model is slightly bettter to reject null hipothesis when p-value < 0.05 for larger
sample size than PO model while in contrary for a smaller sample size.
