Premium
Inference for general parametric functions in box‐Cox‐type transformation models
Author(s) -
Yang Zhenlin,
Wu Eden KaHo,
Desmond Anthony F.
Publication year - 2008
Publication title -
canadian journal of statistics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.804
H-Index - 51
eISSN - 1708-945X
pISSN - 0319-5724
DOI - 10.1002/cjs.5550360208
Subject(s) - power transform , inference , parametric statistics , mathematics , transformation (genetics) , normality , parametric model , type (biology) , function (biology) , statistics , algorithm , computer science , artificial intelligence , ecology , biochemistry , chemistry , consistency (knowledge bases) , biology , gene , geometry , evolutionary biology
The authors propose a simple but general method of inference for a parametric function of the Box‐Cox‐type transformation model. Their approach is built upon the classical normal theory but takes parameter estimation into account. It quickly leads to test statistics and confidence intervals for a linear combination of scaled or unsealed regression coefficients, as well as for the survivor function and marginal effects on the median or other quantité functions of an original response. The authors show through simulations that the finite‐sample performance of their method is often superior to the delta method, and that their approach is robust to mild departures from normality of error distributions. They illustrate their approach with a numerical example.