Generalized Likelihood Ratio Statistic Research Articles

Generalized Likelihood Ratio Statistics and Uncertainty Adjustments in Efficient Adaptive Design of Clinical Trials

A new approach to adaptive design of clinical trials is proposed in a general multiparameter exponential family setting, based on generalized likelihood ratio statistics and optimal sequential testing theory. These designs are easy to implement, maintain the prescribed Type I error probability, and are asymptotically efficient. Practical issues involved in clinical trials allowing mid-course adaptation and the large literature on this subject are discussed, and comparisons between the proposed and existing designs are presented in extensive simulation studies of their finite-sample performance, measured in terms of the expected sample size and power functions.

Wavelet-domain test for long-range dependence in the presence of a trend

We propose a test to distinguish a weakly-dependent time series with a trend component, from a long-memory process, possibly with a trend. The test uses a generalized likelihood ratio statistic based on wavelet domain likelihoods. The trend is assumed to be a polynomial whose order does not exceed a known value. The test is robust to trends which are piecewise polynomials. We study the empirical size and power by means of simulations and find that they are good and do not depend on specific choices of wavelet functions and models for the wavelet coefficients. The test is applied to annual minima of the Nile River and confirms the presence of long-range dependence in this time series.

Nonparametric inference with generalized likelihood ratio tests

The advance of technology facilitates the collection of statistical data. Flexible and refined statistical models are widely sought in a large array of statistical problems. The question arises frequently whether or not a family of parametric or nonparametric models fit adequately the given data. In this paper we give a selective overview on nonparametric inferences using generalized likelihood ratio (GLR) statistics. We introduce generalized likelihood ratio statistics to test various null hypotheses against nonparametric alternatives. The trade-off between the flexibility of alternative models and the power of the statistical tests is emphasized. Well-established Wilks’ phenomena are discussed for a variety of semi- and non-parametric models, which sheds light on other research using GLR tests. A number of open topics worthy of further study are given in a discussion section.

Nonparametric Inferences on Conditional Quantile Processes

This paper is concerned with tests of restrictions on the sample path of conditional quantile processes. These tests are tantamount to assessments of lack of fit for models of conditional quantile functions or more generally as tests of how certain covariates affect the distribution of an outcome variable of interest. This paper extends tests of the generalized likelihood ratio (GLR) type as introduced by Fan, Zhang and Zhang (2001) to nonparametric inference problems regarding conditional quantile processes. As such, the tests proposed here present viable alternatives to existing methods based on the Khmaladze (1981, 1988) martingale transformation. The range of inference problems that may be addressed by the methods proposed here is wide, and includes tests of nonparametric nulls against nonparametric alternatives as well as tests of parametric specifications against nonparametric alternatives. In particular, it is shown that a class of GLR statistics based on nonparametric additive quantile regressions have pivotal asymptotic distributions given by the suprema of squares of Bessel processes, as in Hawkins (1987) and Andrews (1993). The tests proposed here are also shown to be asymptotically rate-optimal for nonparametric hypothesis testing according to the formulations of Ingster (1993) and of Spokoiny (1996).

Nonparametric Inferences for Additive Models

Additive models with backfitting algorithms are popular multivariate nonparametric fitting techniques. However, the inferences of the models have not been very well developed, due partially to the complexity of the backfitting estimators. There are few tools available to answer some important and frequently asked questions, such as whether a specific additive component is significant or admits a certain parametric form. In an attempt to address these issues, we extend the generalized likelihood ratio (GLR) tests to additive models, using the backfitting estimator. We demonstrate that under the null models, the newly proposed GLR statistics follow asymptotically rescaled chi-squared distributions, with the scaling constants and the degrees of freedom independent of the nuisance parameters. This demonstrates that the Wilks phenomenon continues to hold under a variety of smoothing techniques and more relaxed models with unspecified error distributions. We further prove that the GLR tests are asymptotically optimal in terms of rates of convergence for nonparametric hypothesis testing. In addition, for testing a parametric additive model, we propose a bias corrected method to improve the performance of the GLR. The bias-corrected test is shown to share the Wilks type of property. Simulations are conducted to demonstrate the Wilks phenomenon and the power of the proposed tests. A real example is used to illustrate the performance of the testing approach.

Characteristics of Detection of Spatial Signals for Generalized Likelihood Ratio Statistics in the Case of Short Samples

In this paper, we study characteristics of detection of spatial signals from an antenna array in the case of short samples. All the studied decision statistics are obtained on the basis of the generalized likelihood ratio for arbitrary-size samples. As an example, we develop the working characteristics for detecting a useful signal representing one or two coherent plane waves against the background of spatially uniform and nonuniform noise. The efficiency of the used statistics is compared with the χ2 criterion for detecting noise-like signals.

Sieve empirical likelihood ratio tests for nonparametric functions

Generalized likelihood ratio statistics have been proposed in Fan, Zhang and Zhang [Ann. Statist. 29 (2001) 153-193] as a generally applicable method for testing nonparametric hypotheses about nonparametric functions. The likelihood ratio statistics are constructed based on the assumption that the distributions of stochastic errors are in a certain parametric family. We extend their work to the case where the error distribution is completely unspecified via newly proposed sieve empirical likelihood ratio (SELR) tests. The approach is also applied to test conditional estimating equations on the distributions of stochastic errors. It is shown that the proposed SELR statistics follow asymptotically rescaled \chi^2-distributions, with the scale constants and the degrees of freedom being independent of the nuisance parameters. This demonstrates that the Wilks phenomenon observed in Fan, Zhang and Zhang [Ann. Statist. 29 (2001) 153-193] continues to hold under more relaxed models and a larger class of techniques. The asymptotic power of the proposed test is also derived, which achieves the optimal rate for nonparametric hypothesis testing. The proposed approach has two advantages over the generalized likelihood ratio method: it requires one only to specify some conditional estimating equations rather than the entire distribution of the stochastic error, and the procedure adapts automatically to the unknown error distribution including heteroscedasticity. A simulation study is conducted to evaluate our proposed procedure empirically.

Stepwise Likelihood Ratio Statistics in Sequential Studies

SummaryIt is well known that in a sequential study the probability that the likelihood ratio for a simple alternative hypothesis H1versus a simple null hypothesis H0 will ever be greater than a positive constant c will not exceed 1/c under H0. However, for a composite alternative hypothesis, this bound of 1/c will no longer hold when a generalized likelihood ratio statistic is used. We consider a stepwise likelihood ratio statistic which, for each new observation, is updated by cumulatively multiplying the ratio of the conditional likelihoods for the composite alternative hypothesis evaluated at an estimate of the parameter obtained from the preceding observations versus the simple null hypothesis. We show that, under the null hypothesis, the probability that this stepwise likelihood ratio will ever be greater than c will not exceed 1/c. In contrast, under the composite alternative hypothesis, this ratio will generally converge in probability to ∞. These results suggest that a stepwise likelihood ratio statistic can be useful in a sequential study for testing a composite alternative versus a simple null hypothesis. For illustration, we conduct two simulation studies, one for a normal response and one for an exponential response, to compare the performance of a sequential test based on a stepwise likelihood ratio statistic with a constant boundary versus some existing approaches.

Adaptive server staffing in the presence of time-varying arrivals: a feed-forward control approach

We study the short-term staffing problem of systems that experience random, non-stationary demand. The typical method to accommodate changes in arrival rate is to use historical data to identify peak periods and associated forecasting for upcoming time windows. In this paper, we develop a method that instead detects change as it happens. Motivated by an automatic call distributor system in a call centre with time-varying arrivals, we propose a change detection algorithm based upon non-homogeneous Poisson processes. The proposed method is general and may be thought of as a feed-forward strategy, in which we detect a change in the arrival process, estimate the new magnitude of the arrival rate, and assign an appropriate number of servers to the tasks. The generalized likelihood ratio statistic is used and a recommendation for its decision limit is developed. Simulation results are used to evaluate the performance of the detector in the context of a telephone call centre.

Calibrating the Degrees of Freedom for Automatic Data Smoothing and Effective Curve Checking

Curve fitting and curve checking based on the local polynomial regression technique are commonly used data-analytic methods in statistics. This article examines, in nonparametric settings, both the asymptotic expressions and empirical formulas for degrees of freedom (DF), a notion introduced by Hastie and Tibshirani, of linear smoothers. The asymptotic results give useful insights into the nonparametric modeling complexity. Meanwhile, by substituting the exact DFs by the empirical formula, an empirical version of the generalized cross-validation (EGCV) is obtained. An automatic bandwidth selection method based on minimizing EGCV is proposed for conducting local smoothing. This procedure preserves full benefits of the ordinary and generalized cross-validation, but offers a substantial reduction in computational burden. Furthermore, the EGCV-minimizing bandwidth can be extended in a very simple manner to fit multivariate models, such as the varying-coefficient models. Applications of calibrating DFs to important inferential issues, such as assessing the validity of useful model assumptions and measuring the significance of predictor variables based on the generalized likelihood ratio statistics are also discussed. Simulation studies are presented to illustrate the performance of the proposed procedures in a range of statistical problems.

Correction: Generalized likelihood ratio statistics and Wilks phenomenon

The authors of the paper would like to correct the following typographical errors: 1. On page 175, line 4 from the bottom, replace ∑ j2k(1 + jξn,c) by ∑ j2k(1 + j2kξn,c)−2. 2. On page 176, line 2 from the top, replace n−1 ∑ j2k(1 + jξn,c) by n−1 ∑ j2k(1 + j2kξn,c)−2. 3. On page 180, line 1 from the top, replace |θ |ξcnn−2k/(4k+1) by and ‖θ‖ ≥ cnn −2k/(4k+1). 4. On page 184, line 4 from the bottom, replace Op(h−1) by Op{(nh)−1}.

Generalized Likelihood Ratio Statistics and Wilks Phenomenon

Likelihood ratio theory has had tremendous success in parametric inference, due to the fundamental theory of Wilks. Yet, there is no general applicable approach for nonparametric inferences based on function estimation. Maximum likelihood ratio test statistics in general may not exist in nonparametric function estimation setting. Even if they exist, they are hard to find and can not; be optimal as shown in this paper. We introduce the generalized likelihood statistics to overcome the drawbacks of nonparametric maximum likelihood ratio statistics. A new S Wilks phenomenon is unveiled. We demonstrate that a class of the generalized likelihood statistics based on some appropriate nonparametric estimators are asymptotically distribution free and follow chi (2)-distributions under null hypotheses for a number of useful hypotheses and a variety of useful models including Gaussian white noise models, nonparametric regression models, varying coefficient models and generalized varying coefficient models. We further demonstrate that generalized likelihood ratio statistics are asymptotically optimal in the sense that they achieve optimal rates of convergence given by Ingster. They can even be adaptively optimal in the sense of Spokoiny by using a simple choice of adaptive smoothing parameter. Our work indicates that the generalized likelihood ratio statistics are indeed general and powerful for nonparametric testing problems based on function estimation.

Asymptotic approximations for error probabilities of sequential or fixed sample size tests in exponential families

Asymptotic approximations for the error probabilities of sequential tests of composite hypotheses in multiparameter exponential families are developed herein for a general class of test statistics, including generalized likelihood ratio statistics and other functions of the sufficient statistics. These results not only generalize previous approximations for Type I error probabilities of sequential generalized likelihood ratio tests, but also pro- vide a unified treatment of both sequential and fixed sample size tests and of Type I and Type II error probabilities. Geometric arguments involving integration over tubes play an important role in this unified theory.

Extension of the Wald statistic to models with dependent observations

A generalization of the Wald statistic for testing composite hypotheses is suggested for dependent data from exponential models which include Levy processes and diffusion fields. The generalized statistic is proved to be asymptotically chi-squared distributed under regular composite hypotheses. It is simpler and more easily available than the generalized likelihood ratio statistic. Simulations in an example where the latter statistic is available show that the generalized Wald test achieves higher average power than the generalized likelihood ratio test.

Case-control isotonic regression for investigation of elevation in risk around a point source.

Stone's isotonic regression method for analysing count data to estimate disease risk in relation to a point source of environmental pollution is now routinely used. This paper develops the corresponding procedure for case-control data consisting of the locations of individual cases with controls with associated covariate information. In this setting, the generalized likelihood ratio statistic to test the null hypothesis of constant risk against the alternative that risk is a monotone non-increasing function of distance from the point source is intractable. An approximate Monte Carlo test is described, extending an exact test proposed by Bithell for the situation in which there are no covariates. Interval estimates of risk as a function of distance from the point source are constructed by simulation of the sampling distribution of the isotonic regression estimator. The methodology is illustrated by two applications: one to the relative risk of larynx cancers and lung cancers near a now-disused industrial incinerator; the other to the risk of asthma in children in relation to distance of residence from the nearest main road.

Some asymptotic results for hypothesis testing in multivariate failure time data

Abstract This paper presents a class of generalized Wald, generalized score and generalized likelihood ratio statistics for hypothesis testing and model selection for multivariate failure time data. These statistics are based on a marginal hazard model with a common baseline hazard function. The large sample distributions of these statistics are examined. It is shown that the proposed test statistics follow asymptotically a weighted sum of independent χ12 distributions.

Optimum array detector for a weak signal in unknown noise

We study the design of constant false-alarm rate (CFAR) tests for detecting a rank-one signal in the presence of background Gaussian noise with unknown spatial covariance. We look at invariant tests, i.e., those tests whose performance is independent of the nuisance parameters, like the background noise covariance. Such tests are shown to have the desirable CFAR property. We characterize the class of all such tests by showing that any invariant decision statistic can be written as a function of two basic statistics which are in fact the adaptive matched filter (AMF) statistic and Kelly's generalized likelihood ratio statistic. Further, we establish an optimum test in the limit of low signal-to-noise ratio (SNR), the locally most powerful invariant (LMPI) test. We also derive the bound for the probability of detection of any invariant detector, at a fixed false-alarm rate, and compare the LMPI and the published detectors (Kelly and AMF) to it.

Using the Generalized Likelihood Ratio Statistic for Sequential Detection of a Change-Point

We study sequential detection of a change-point using the generalized likelihood ratio statistic. For the special case of detecting a change in a normal mean with known variance, we give approximations to the average run lengths and compare our procedure to standard CUSUM tests and combined CUSUM-Shewhart tests. Several examples indicating extensions to problems involving multiple parameters are discussed.

A Modification of schwarz's sequential likelihood ratio tests in multivariate sequential analysis

By developing and applying certain results on boundary crossing probabilities of generalized likelihood ratio statistics in multiparameter exponential families, we show that Lai's (1988a) modification of Schwarz's (1962) sequential likelihood ratio tests of one-sided hypotheses in the one-parameter case can be extended to general hypotheses in the multiparameter case. Such tests are shown to be asymptotically optimal, from both frequentist and Bayesian points of view, and numerical results are also given to demonstrate their performance.

Estimating Hot Numbers and Testing Uniformity for the Lottery

Abstract We consider the problem of estimating probabilities for the different numbers in a lottery when the winning numbers are selected sequentially without replacement. Numerical methods are developed to compute maximum likelihood estimates of the probabilities and their variance-covariance matrix based on observed outcomes. The generalized likelihood ratio statistic –2 ln(λ) for testing equiprobable numbers is also given. The methods are applied to a data set from the Lotto America lottery. The properties of the procedure are studied by simulation.