True/False/Uncertain - Briefly explain. No credit without an explanation

1. At the 10th decile the (cumulative) lift would never be one.

2. With 4 categories of Y and 5 Xs, the multinomial logit has twice as many parameters as the cumulative logit.

3. If one uses US states data to model COVID-19 cases, one should use an offset.

4. Global homogeneity SSG is never zero when we start with each observation in its own group with the Ward clustering method.

5. Having more missing values in the variable education for higher levels of education is an example of MNAR