Q1. Show that a Negative Binomial loglinear model is a GLM by specifying the three components of the model. Assume the number of success for the Negative binomial distribution is known.
Q2. In an I x J x K table, show that homogeneous XY association implies XZ association and homogeneous YZ association.
Q3. Consider the Alcohol Sales survey and read the data set in SAS and answer the following questions.(provide the SAS Code)
(a) Find the typo errors, outliers and the missing values in the data set and take care of them for the next questions.
(b) Which variables in this study can be considered as a response variable?
(c) Construct the largest possible contingency table using the categorical variables. Is this a useful table? Why or why not?
(d) Test for the independence of "able-to-buy" and "gender-clerk".
(e) Draw a graph, representing the relationship between "being able to buy" and "Time". Using a logistic regression, test the hypothesis that the probability of being able to buy is the same for all times.
(f) Draw a graph, representing the relationship between "being able to buy" and "confederate age". Then, using a probit regression, test the hypothesis that the probability of being able to buy is the same for all confederate ages.
Q4. Simulation Study: Consider the Wald confidence interval for, the proportion of a Bernoulli distribution. Using simulation in SAS, study the coverage probability of this confidence interval for different values of π;α and n. Write down your conclusion and using literature, list some alternative solutions for the problem. (provide the SAS Code).
Attachment:- Assignment Files.rar