10.12 Assignment 3

Q1. If the classification accuracy for an attribute is estimated to be .8, What does it mean? Can we say a model with higher classification accuracy is better than a model producing lower classifications accuracy?

Q2. Fit the G-DINA model to data2 and Q2, both of which can be downloaded from Blackboard, and answer the following questions.

  1. How many item and joint attribute distribution parameters do we have?
  2. What is the estimated size of latent class with attribute profile 100?
  3. What are the estimated delta parameters of Item 5 and associated standard errors?
  4. What is the EAP estimate of attribute profile of student 10?
  5. What is the MLE estimate of attribute profile of student 15?

Q3. Fit the DINA, LLM, ACDM, and G-DINA model to data2 and Q2.

  1. Check which model has the highest test level classification accuracy?
  2. If we are considering only attribute class 000, which model has the best classification accuracy?
  3. Please discuss the attribute level classification accuracy for this models.

Q4. Fit G-DINA model to data2 and Q2. Please obtain item-level absolute fit measures. Discuss your results.

Q5. What are the major differences between MLE, MAP and EAP estimation procedures for estimating individual attributes? In which situation, you should consider MAP estimation process compared to other methods?