A new test for heteroscedasticity in single-index models
Introduction
In this article, we consider the SIM where are covariates, is an unknown smooth link function, is an independent random error with . are unknown with and for identification.
In regression analysis, we often assume that error terms have common variance. Under such an assumption, many authors investigate the inference for SIM. See [1], [2], [3], [4], [5], [6] for more details. Without this assumption, we need more complicated methods to estimate the unknown parameter and regression function. Therefore, detection of heteroscedasticity in SIM is an important issue. Our objective is to check variance heterogeneity in (1) by testing the following hypothesis Obviously, the heteroscedasticity test in (2) is equivalent to determining whether is equal to .
A number of authors have investigated heteroscedasticity tests for regression models, such as [7], [8], [9], [10], [11], [12]. For testing heteroscedasticity in SIM, however, the literature is insufficient. [13] proposed a kernel smoothing-based statistic for checking the heteroscedasticity in SIM, which has a drawback of the dimensionality problem because of the estimation inefficiency for the multivariate nonparametric function. As a result, the significance level cannot be well maintained when we apply the limiting null distribution in moderate sample size, and the test in [13] can only check alternative hypothesis that is different from the null hypothesis at the rate of . Therefore, the test statistic in [13] is less powerful.
In this paper, we develop a new statistic for detecting heteroscedasticity in SIM. This statistic is based on the weighted integral of the residual marked characteristic function. The weight function plays an important role in the proposed test statistic, in which the density function of a spherical stable law is used as the weight function. Under this case, the weighted integral can be transformed into a simple unconditional expectation. This yields the statistic is constructed only by pairwise distances between points in a sample, which avoids the dimension problem of the kernel smoothing-based statistic in [13]. To the best of our knowledge, this work is the first applying characteristic function to check heteroscedasticity in SIM.
The rest of this article is organized as follows. In Section 2, the test statistic is developed and its asymptotic properties are established. In Section 3, we propose a simple bootstrap algorithm to detect heteroscedasticity for the SIM. In Section 4, numerical studies are conducted to evaluate the performance of the test. In Section 5, a real data is analyzed for illustrating the proposed methodology. Conclusions are given in Section 6. Technical assumptions and proofs are provided in Appendix.
Section snippets
Construction of test statistic
Let with . We have , under . According to the uniqueness of the Fourier transform of a function, we can do the following equivalent substitution for Note that is not a statistic by itself, then we have where is a suitable weight function. In terms of the definition of complex modulus, we have where is an independent copy of . The characteristic function of a spherical
Estimation of SIM
We first estimate and using the MAVE, and see [5], [6] for more details. The main implementation steps are as follows:
- (1)
For the given initial estimate , calculate and where , is a kernel function, and are estimators of and , respectively.
- (2)
Calculate where
Numerical studies
In this section, we investigate the performance of the proposed test statistic in a finite sample size by numerical studies. To assess the performance of power, the following two examples are designed, in which 1000 replications of the experiment are taken to calculate empirical sizes and powers at the significance level . The sample sizes are , 200 and 400, and the number of bootstrap sample is set to be . For comparison, we consider other two competitors, i.e., (a) Zheng’s
Delft data
In this section, the Delft dataset is analyzed for illustration. This dataset was performed at the Delft Ship Hydromechanics Laboratory, comprising 308 full-scale experiments, which can be obtained from the website https://archive.ics.uci.edu/ml/machine-learning-databases/00243/. These experiments derived from a parent form closely related to the ‘Standfast 43’ designed by Frans Maas. The covariates are —longitudinal position of the center of buoyancy, —prismatic coefficient,
Conclusions
In this article, a new test is proposed to check the heteroscedasticity in SIM. The test statistic is based on the pairwise distance between sample points. The results show that the statistic shares asymptotic normal distributions under the null, local alternative, and fixed alternative hypotheses, but the asymptotic means and variances are different. The proposed method is easy to implement and the convergence rate of the statistic does not depend on the dimensions of the covariates, which
Acknowledgments
We are very grateful to the Associate Editor and reviewers for their helpful comments and suggestions which largely improved our work. This research was supported by Science Foundation of Xuzhou University of Technology of China No. XKY2018120.
References (19)
- et al.
Heteroscedasticity checks for single index models
J. Multivariate Anal.
(2015) Integral transformations of distributions and estimates of parameters of multidimensional spherically symmetric stable laws
- et al.
On a dimension reduction regression with covariate adjustment
J. Multivariate Anal.
(2012) - et al.
Model checking for parametric single-index models: a dimension reduction model-adaptive approach
J. R. Stat. Soc. Ser. B Stat. Methodol.
(2016) - et al.
Direct estimation of the index coefficient in a single-index model
Ann. Statist.
(2001) Semiparametric least squares (SLS) and weighted SLS estimation of singleindex models
J. Econometrics
(1993)- et al.
Penalized least squares for single index models
J. Statist. Plann. Inference
(2011) An adaptive estimation of dimension reduction space
J. R. Stat. Soc. Ser. B Stat. Methodol.
(2002)Asymptotic distributions for two estimators of the single-index model
Econometric Theory
(2006)
Cited by (0)
- 1
Yan-Yong Zhao and Jian-Qiang Zhao are co-first authors of the paper.