Poster #9 - Quantifying Differences Between Pairs: A Case Replacement Approach to Expressing Confidence in Statistical Inference

• In Event: 47.099 - MTCC Poster Session 11
  In Poster Session: 47.099-1 - Division D Section 2 Statistical Theory and Quantitative Methodologies

Abstract

We illustrate a case-replacement method for quantifying confidence in statistical estimates of the difference in means between two groups. This post hoc method occurs either after estimation or after a statistical test such as a t-test. We argue that the case replacement method is easier to understand for non-technical audiences than classical approaches such as standard errors and confidence intervals. We employ data from two large standardized assessment datasets to illustrate. We generate statements such as “XX% of average students from District A and XX% of average students from District B must be swapped to invalidate the inference that the District A is better than B”. A large percentage indicates more confidence in the inference.

Authors

• Qinyun Lin, Michigan State University

• Joseph A Marr, Michigan State University

• William Sullivan, Michigan State University