Are Robust Poisson Models Less Affected by Model Misspecification Compared to the Log-Bionomial Models When Estimating Relative Risks for Common Binary Outcomes?

• Activity Number: 42
• Type: Contributed
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Wansu Chen*+ and Lei Qian and Jiaxiao Shi and Stanley Azen
• Companies: Kaiser Permanente and Kaiser Permanente and Kaiser Permanente and University of Southern California
• Keywords: relative risk ; log-binomial regression ; robust Poisson regression ; model misspecification ; common binary outcomes

Abstract

To estimate relative risks or risk ratios for common binary outcomes, the most popular model-based methods are the robust (also known as modified) Poisson (RP) and the log-binomial (LB) regression. Of the two methods, it is believed that the LB method yields more efficient estimators because they are maximum likelihood based while the RP model may be less affected by misspecification. Evidence to support the robustness of the RP models when they are compared with LB models under model misspecification is very limited. In this study, simulation was conducted to evaluate the performance of the two methods in several scenarios when the models were mis-specified. Various types of model misspecification were simulated. Our findings suggest that when the link functions were mis-specified or the distrutions of the outcome variable were altered, the RP models significantly outperformed the LB models in various scenarios. Users should be aware of the limitations when choosing the most appropriate models to estimate relative risks.