Book Chapter

Inference for Response-Adaptive Randomization

Rosenberger, William F, Lachin, John M

2015,2016

Overview

Inference for response‐adaptive randomization is very complicated because both the treatment assignments and responses are correlated. This leads to nonstandard problems and new insights into conditioning. This chapter first examines likelihood‐based inference and then randomization‐based inference. More details on the theory of likelihood‐based inference for response‐adaptive randomization can be found in Hu and Rosenberger (2006). Response‐adaptive randomization induces additional correlation among the responses, and this leads to an increase in the variance of the test statistic. This increased variance contributes to a decrease in power for standard tests based on a population model. The chapter explores the power of response‐adaptive randomization procedures. As with restricted randomization procedures, randomization‐based inference can be performed following a response‐adaptive randomization procedure using the family of linear rank tests.

• Publisher: John Wiley & Sons, Incorporated,Wiley,John Wiley & Sons, Inc
• Subject: linear rank tests / maximum likelihood estimators / population‐based inference / randomization‐based inference / response‐adaptive randomization
• ISBN: 1118742249, 9781118742242, 1118742117, 9781118742112