Local power of some tests in exponential family nonlinear models

In this paper we obtain asymptotic expansions up to order n(-1/2) for the nonnull distribution functions of the likelihood ratio, Wald, score and gradient test statistics in exponential family nonlinear models (Cordeiro and Paula, 1989), under a sequence of Pitman alternatives. The asymptotic distributions of all four statistics are obtained for testing a subset of regression parameters and for testing the dispersion parameter, thus generalising the results given in Cordeiro et al. (1994) and Ferrari et al. (1997). We also present Monte Carlo simulations in order to compare the finite-sample performance of these tests. (C) 2010 Elsevier B.V. All rights reserved.

Wald tests for IV regression with weak instruments

This dissertation deals with the problem of making inference when there is weak identification in models of instrumental variables regression. More specifically we are interested in one-sided hypothesis testing for the coefficient of the endogenous variable when the instruments are weak. The focus is on the conditional tests based on likelihood ratio, score and Wald statistics. Theoretical and numerical work shows that the conditional t-test based on the two-stage least square (2SLS) estimator performs well even when instruments are weakly correlated with the endogenous variable. The conditional approach correct uniformly its size and when the population F-statistic is as small as two, its power is near the power envelopes for similar and non-similar tests. This finding is surprising considering the bad performance of the two-sided conditional t-tests found in Andrews, Moreira and Stock (2007). Given this counter intuitive result, we propose novel two-sided t-tests which are approximately unbiased and can perform as well as the conditional likelihood ratio (CLR) test of Moreira (2003).

Testes de hipóteses em modelos de sobrevivência com Fração de cura

Survival models deals with the modeling of time to event data. However in some situations part of the population may be no longer subject to the event. Models that take this fact into account are called cure rate models. There are few studies about hypothesis tests in cure rate models. Recently a new test statistic, the gradient statistic, has been proposed. It shares the same asymptotic properties with the classic large sample tests, the likelihood ratio, score and Wald tests. Some simulation studies have been carried out to explore the behavior of the gradient statistic in fi nite samples and compare it with the classic statistics in diff erent models. The main objective of this work is to study and compare the performance of gradient test and likelihood ratio test in cure rate models. We first describe the models and present the main asymptotic properties of the tests. We perform a simulation study based on the promotion time model with Weibull distribution to assess the performance of the tests in finite samples. An application is presented to illustrate the studied concepts

A single statistic for monitoring the covariance matrix of bivariate processes

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Evaluating Measurement Invariance with Censored Ordinal Data: A Monte Carlo Comparison of Alternative Model Estimators and Scales of Measurement

Evaluations of measurement invariance provide essential construct validity evidence. However, the quality of such evidence is partly dependent upon the validity of the resulting statistical conclusions. The presence of Type I or Type II errors can render measurement invariance conclusions meaningless. The purpose of this study was to determine the effects of categorization and censoring on the behavior of the chi-square/likelihood ratio test statistic and two alternative fit indices (CFI and RMSEA) under the context of evaluating measurement invariance. Monte Carlo simulation was used to examine Type I error and power rates for the (a) overall test statistic/fit indices, and (b) change in test statistic/fit indices. Data were generated according to a multiple-group single-factor CFA model across 40 conditions that varied by sample size, strength of item factor loadings, and categorization thresholds. Seven different combinations of model estimators (ML, Yuan-Bentler scaled ML, and WLSMV) and specified measurement scales (continuous, censored, and categorical) were used to analyze each of the simulation conditions. As hypothesized, non-normality increased Type I error rates for the continuous scale of measurement and did not affect error rates for the categorical scale of measurement. Maximum likelihood estimation combined with a categorical scale of measurement resulted in more correct statistical conclusions than the other analysis combinations. For the continuous and censored scales of measurement, the Yuan-Bentler scaled ML resulted in more correct conclusions than normal-theory ML. The censored measurement scale did not offer any advantages over the continuous measurement scale. Comparing across fit statistics and indices, the chi-square-based test statistics were preferred over the alternative fit indices, and ΔRMSEA was preferred over ΔCFI. Results from this study should be used to inform the modeling decisions of applied researchers. However, no single analysis combination can be recommended for all situations. Therefore, it is essential that researchers consider the context and purpose of their analyses.

A note on the local power of the LR, Wald, score and gradient tests

This paper examines the local power of the likelihood ratio, Wald, score and gradient tests under the presence of a scalar parameter, phi say, that is orthogonal to the remaining parameters. We show that some of the coefficients that define the local powers remain unchanged regardless of whether phi is known or needs to be estimated, where as the others can be written as the sum of two terms, the first of which being the corresponding term obtained as if phi were known, and the second, an additional term yielded by the fact that phi is unknown. The contribution of each set of parameters on the local powers of the tests can then be examined. Various implications of our main result are stated and discussed. Several examples are presented for illustrative purposes

Local power properties of some asymptotic tests in symmetric linear regression models

In this paper we obtain asymptotic expansions, up to order n(-1/2) and under a sequence of Pitman alternatives, for the nonnull distribution functions of the likelihood ratio, Wald, score and gradient test statistics in the class of symmetric linear regression models. This is a wide class of models which encompasses the t model and several other symmetric distributions with longer-than normal tails. The asymptotic distributions of all four statistics are obtained for testing a subset of regression parameters. Furthermore, in order to compare the finite-sample performance of these tests in this class of models, Monte Carlo simulations are presented. An empirical application to a real data set is considered for illustrative purposes. (C) 2011 Elsevier B.V. All rights reserved.

The new class of Kummer beta generalized distributions

Ng and Kotz (1995) introduced a distribution that provides greater flexibility to extremes. We define and study a new class of distributions called the Kummer beta generalized family to extend the normal, Weibull, gamma and Gumbel distributions, among several other well-known distributions. Some special models are discussed. The ordinary moments of any distribution in the new family can be expressed as linear functions of probability weighted moments of the baseline distribution. We examine the asymptotic distributions of the extreme values. We derive the density function of the order statistics, mean absolute deviations and entropies. We use maximum likelihood estimation to fit the distributions in the new class and illustrate its potentiality with an application to a real data set.

Assessment of variance components in nonlinear mixed-effects elliptical models

The issue of assessing variance components is essential in deciding on the inclusion of random effects in the context of mixed models. In this work we discuss this problem by supposing nonlinear elliptical models for correlated data by using the score-type test proposed in Silvapulle and Silvapulle (1995). Being asymptotically equivalent to the likelihood ratio test and only requiring the estimation under the null hypothesis, this test provides a fairly easy computable alternative for assessing one-sided hypotheses in the context of the marginal model. Taking into account the possible non-normal distribution, we assume that the joint distribution of the response variable and the random effects lies in the elliptical class, which includes light-tailed and heavy-tailed distributions such as Student-t, power exponential, logistic, generalized Student-t, generalized logistic, contaminated normal, and the normal itself, among others. We compare the sensitivity of the score-type test under normal, Student-t and power exponential models for the kinetics data set discussed in Vonesh and Carter (1992) and fitted using the model presented in Russo et al. (2009). Also, a simulation study is performed to analyze the consequences of the kurtosis misspecification.

Bartlett corrections in Birnbaum-Saunders nonlinear regression models

Lemonte and Cordeiro [Birnbaum-Saunders nonlinear regression models, Comput. Stat. Data Anal. 53 (2009), pp. 4441-4452] introduced a class of Birnbaum-Saunders (BS) nonlinear regression models potentially useful in lifetime data analysis. We give a general matrix Bartlett correction formula to improve the likelihood ratio (LR) tests in these models. The formula is simple enough to be used analytically to obtain several closed-form expressions in special cases. Our results generalize those in Lemonte et al. [Improved likelihood inference in Birnbaum-Saunders regressions, Comput. Stat. DataAnal. 54 (2010), pp. 1307-1316], which hold only for the BS linear regression models. We consider Monte Carlo simulations to show that the corrected tests work better than the usual LR tests.

A Single Statistic for Monitoring the Covariance Matrix of Bivariate Processes

In this article, we present a new control chart for monitoring the covariance matrix in a bivariate process. In this method, n observations of the two variables were considered as if they came from a single variable (as a sample of 2n observations), and a sample variance was calculated. This statistic was used to build a new control chart specifically as a VMIX chart. The performance of the new control chart was compared with its main competitors: the generalized sampled variance chart, the likelihood ratio test, Nagao's test, probability integral transformation (v(t)), and the recently proposed VMAX chart. Among these statistics, only the VMAX chart was competitive with the VMIX chart. For shifts in both variances, the VMIX chart outperformed VMAX; however, VMAX showed better performance for large shifts (higher than 10%) in one variance.

The local power of the gradient test

The asymptotic expansion of the distribution of the gradient test statistic is derived for a composite hypothesis under a sequence of Pitman alternative hypotheses converging to the null hypothesis at rate n(-1/2), n being the sample size. Comparisons of the local powers of the gradient, likelihood ratio, Wald and score tests reveal no uniform superiority property. The power performance of all four criteria in one-parameter exponential family is examined.

Local power and size properties of the LR, Wald, score and gradient tests in dispersion models

We derive asymptotic expansions for the nonnull distribution functions of the likelihood ratio, Wald, score and gradient test statistics in the class of dispersion models, under a sequence of Pitman alternatives. The asymptotic distributions of these statistics are obtained for testing a subset of regression parameters and for testing the precision parameter. Based on these nonnull asymptotic expansions, the power of all four tests, which are equivalent to first order, are compared. Furthermore, in order to compare the finite-sample performance of these tests in this class of models, Monte Carlo simulations are presented. An empirical application to a real data set is considered for illustrative purposes. (C) 2012 Elsevier B.V. All rights reserved.

Essays in Macroecometrics: methodological aspects and empirical applications

In the first chapter, I develop a panel no-cointegration test which extends Pesaran, Shin and Smith (2001)'s bounds test to the panel framework by considering the individual regressions in a Seemingly Unrelated Regression (SUR) system. This allows to take into account unobserved common factors that contemporaneously affect all the units of the panel and provides, at the same time, unit-specific test statistics. Moreover, the approach is particularly suited when the number of individuals of the panel is small relatively to the number of time series observations. I develop the algorithm to implement the test and I use Monte Carlo simulation to analyze the properties of the test. The small sample properties of the test are remarkable, compared to its single equation counterpart. I illustrate the use of the test through a test of Purchasing Power Parity in a panel of EU15 countries. In the second chapter of my PhD thesis, I verify the Expectation Hypothesis of the Term Structure in the repurchasing agreements (repo) market with a new testing approach. I consider an "inexact" formulation of the EHTS, which models a time-varying component in the risk premia and I treat the interest rates as a non-stationary cointegrated system. The effect of the heteroskedasticity is controlled by means of testing procedures (bootstrap and heteroskedasticity correction) which are robust to variance and covariance shifts over time. I fi#nd that the long-run implications of EHTS are verified. A rolling window analysis clarifies that the EHTS is only rejected in periods of turbulence of #financial markets. The third chapter introduces the Stata command "bootrank" which implements the bootstrap likelihood ratio rank test algorithm developed by Cavaliere et al. (2012). The command is illustrated through an empirical application on the term structure of interest rates in the US.

Study of the sensitivity of the XENON1T experiment with the profile likelihood method

Oggi sappiamo che la materia ordinaria rappresenta solo una piccola parte dell'intero contenuto in massa dell'Universo. L'ipotesi dell'esistenza della Materia Oscura, un nuovo tipo di materia che interagisce solo gravitazionalmente e, forse, tramite la forza debole, è stata avvalorata da numerose evidenze su scala sia galattica che cosmologica. Gli sforzi rivolti alla ricerca delle cosiddette WIMPs (Weakly Interacting Massive Particles), il generico nome dato alle particelle di Materia Oscura, si sono moltiplicati nel corso degli ultimi anni. L'esperimento XENON1T, attualmente in costruzione presso i Laboratori Nazionali del Gran Sasso (LNGS) e che sarà in presa dati entro la fine del 2015, segnerà un significativo passo in avanti nella ricerca diretta di Materia Oscura, che si basa sulla rivelazione di collisioni elastiche su nuclei bersaglio. XENON1T rappresenta la fase attuale del progetto XENON, che ha già realizzato gli esperimenti XENON10 (2005) e XENON100 (2008 e tuttora in funzione) e che prevede anche un ulteriore sviluppo, chiamato XENONnT. Il rivelatore XENON1T sfrutta circa 3 tonnellate di xeno liquido (LXe) e si basa su una Time Projection Chamber (TPC) a doppia fase. Dettagliate simulazioni Monte Carlo della geometria del rivelatore, assieme a specifiche misure della radioattività dei materiali e stime della purezza dello xeno utilizzato, hanno permesso di predire con accuratezza il fondo atteso. In questo lavoro di tesi, presentiamo lo studio della sensibilità attesa per XENON1T effettuato tramite il metodo statistico chiamato Profile Likelihood (PL) Ratio, il quale nell'ambito di un approccio frequentista permette un'appropriata trattazione delle incertezze sistematiche. In un primo momento è stata stimata la sensibilità usando il metodo semplificato Likelihood Ratio che non tiene conto di alcuna sistematica. In questo modo si è potuto valutare l'impatto della principale incertezza sistematica per XENON1T, ovvero quella sulla emissione di luce di scintillazione dello xeno per rinculi nucleari di bassa energia. I risultati conclusivi ottenuti con il metodo PL indicano che XENON1T sarà in grado di migliorare significativamente gli attuali limiti di esclusione di WIMPs; la massima sensibilità raggiunge una sezione d'urto σ=1.2∙10-47 cm2 per una massa di WIMP di 50 GeV/c2 e per una esposizione nominale di 2 tonnellate∙anno. I risultati ottenuti sono in linea con l'ambizioso obiettivo di XENON1T di abbassare gli attuali limiti sulla sezione d'urto, σ, delle WIMPs di due ordini di grandezza. Con tali prestazioni, e considerando 1 tonnellata di LXe come massa fiduciale, XENON1T sarà in grado di superare gli attuali limiti (esperimento LUX, 2013) dopo soli 5 giorni di acquisizione dati.