An Empirical Study of Accuracy and Precision in the Estimation of Difference Between Correlated Proportions

• In Event: Poster Session 7
  In Poster Session: Division D Section 2 Poster Session

Abstract

A Monte Carlo study investigated three methods for computing confidence intervals (CI) around differences in correlated proportions (Wald CI, adjusted Wald CI, and a likelihood-based CI method proposed by Tango, (1998)) to determine which one produces the most accurate and precise CI estimates. The manipulated factors included overall sample size (10, 20, 30, 40, 50, 100, 500, 1000), direction and strength of the relationship between two proportions (±.40, ±.30, ±.20, ±10, 0), and the population difference in marginal proportions (±.3, ±.25, ±.10, ±.05, 0). The adjusted Wald CI provided the best coverage across the conditions investigated. Both the original Wald CI and the Tango CI produced substantial under-coverage in some small sample conditions.

Authors

• Thanh Vinh Pham, University of South Florida

• Jeanine L. Romano, American University of Sharjah

• Eun Sook Kim, University of South Florida

• Patricia Rodriguez de Gil, University of South Florida

• Diep Thi Nguyen, University of South Florida

• Pei-Chen Wu, University of South Florida

• Jeffrey D. Kromrey, University of South Florida