Changes in version 2.0.0 (2026-02-04)                  

Major Changes

  - NEW: Added Graded Response Model (GRM) for ordinal/Likert-scale data
    
      - lsirmgrm(): 1PL GRM with latent space
      - lsirmgrm2pl(): 2PL GRM with item discrimination parameters
      - Supports complete data, MAR, and MCAR missing data mechanisms
      - Implements threshold ordering constraint (β₁ > β₂ > ... > βₖ₋₁)
        for identifiability
      - Reference: De Carolis, Kang, & Jeon (2025),
        DOI: 10.1080/00273171.2025.2605678

  - NEW: Adaptive MCMC algorithm
    
      - Automatic tuning of proposal standard deviations during burn-in
      - Parameter-specific target acceptance rates
      - Improved mixing and convergence for complex models
      - Available for all model types (1PL, 2PL, GRM)