Title: A goodness-of-fit test for variable-adjusted models
Authors: Chuanlong Xie - Jinan University, Guangzhou (China) [presenting]
Abstract: A goodness-of-fit test is provided to checking parametric single-index regression structure when both the response and covariates are measured with distortion errors. Under the null hypothesis, the proposed test statistic asymptotically behaves like a test with univariate covariate and thus, can work better on the significance level maintenance and power performance than existing tests in multivariate covariate cases. With properly selected bandwidths, the proposed test is not seriously affected by distortion measurement errors in the sense that the limiting null distributions in the cases with and without distortion measurement errors can be identical. Numerical studies are conducted to examine the performance of the test in finite sample scenarios.