Articles • svyder

Getting Started with svyder: A hands-on introduction to Design Effect Ratio diagnostics with svyder: run the full pipeline in one call, interpret the three-tier classification, visualize DER profiles, extract corrected posterior draws, and build a pipe-friendly workflow — all using bundled datasets.

Understanding Design Effect Ratios: A deep dive into the Design Effect Ratio (DER) framework: definition and three regimes, why parameters differ in design sensitivity, the decomposition theorems linking DER to Kish DEFF and hierarchical shrinkage, the conservation law, three-tier classification, threshold sensitivity, and comparison between complex survey and equal-weight data.
Decomposition Theorems: Why DER Differs Across Parameters: A deep dive into the DER decomposition theorems: understand why different parameters have different design sensitivities, verify the closed-form approximations against empirical values, and explore the conservation law that links hierarchical shrinkage to survey design effects.

The Compute-Classify-Correct Pipeline: A comprehensive walkthrough of the three-step DER diagnostic pipeline: compute sandwich-based DER values, classify parameters into tiers with threshold-based flagging, and apply selective Cholesky correction to flagged parameters only. Includes sensitivity analysis, cross-clustering comparison, and practical guidance for working with corrected posterior draws.
Case Study: NSECE Survey Data Analysis: A comprehensive case study demonstrating the full DER diagnostic workflow on NSECE-like survey data: explore the survey design, run diagnostics, interpret fixed and random effect results, verify decomposition theorems, compare selective vs blanket correction, and report results for publication.