This paper investigates the power properties of the Sargan test in the presence of measurement errors in dynamic panel data models. The conclusion from Monte Carlo simulations, and an application on the data used by Arellano and Bond (1991), is that in the very likely case of measurement errors in either the dependent or any of the independent variables, we will, if we rely on the Sargan test, quite likely accept a mis-specified model and end up with biased results.

When scientists collect data to be analyzed, regardless of what quantities are being measured, there are inevitably errors in the measurements. In cases where two independent variables are measured with errors, many existing techniques can produce an estimated least-squares linear fit to the data, taking into consideration the size of the errors in both variables. Yet some experiments yield data that do not only contain errors in both variables, but also a non-zero covariance between the errors. In such situations, the experiment results in measurements with error ellipses with tilts specified by the covariance terms. Following an approach suggested by Dr. Edward Robinson, Professor of Astronomy at the University of Texas at Austin, this report describes a methodology that finds the estimates of linear regression parameters, as well as an estimated covariance matrix, for a dataset with tilted error ellipses. Contained in an appendix is the R code for a program that produces these estimates according to the methodology. This report describes the results of the program run on a dataset of measurements of the surface brightness and Sérsic index of galaxies in the Virgo cluster. / text

>Magister Scientiae - MSc / There exists an extensive statistics literature dealing with non-parametric deconvolution, the estimation of the underlying population probability density when sample values are subject to measurement errors. In parametric deconvolution, on the other hand, the data are known to be from a specific distribution. In this case the parameters of the distribution can be estimated by e.g. maximum likelihood. In realistic cases the measurement errors may be heteroscedastic and there may be unknown parameters associated with the distribution. The specific realistic case is investigated in which the measurement error standard deviation is proportional to the true sample values. In this case it is shown that the method of moment’s estimation is particularly simple. Estimation by maximum likelihood is computationally very expensive, since numerical integration needs to be performed for each data point, for each evaluation of the likelihood function. Method of moment’s estimation sometimes fails to give physically meaningful estimates. The origin of this problem lies in the large sampling variations of the third moment. Possible remedies are considered. Due to the fact that a convolution integral needed to be calculated for each data point, and that this has to be repeated for each iteration towards the solution, maximum likelihood computing cost is very high. New preliminary work suggests that saddle point approximations could sometimes be used for the convolution integrals. This allows much larger datasets to be dealt with. Application of the theory is illustrated with simulation and real data.

Consideramos a predição ótima de valores latentes com base em dados sujeitos a erros de medida endógenos e exógenos, obtidos a partir de uma amostra aleatória de uma população finita. Consideramos o modelo misto para populações finitas (MMPF) com erros de medida exógenos e endógenos usando o enfoque proposto por Stanek et al. (2004) e Stanek & Singer (2004), e calculamos o melhor preditor linear não enviesado (BLUP) do valor latente da i-ésima unidade selecionada na amostra. Quando as variâncias endógenas são heterocedásticas, o preditor obtido sob o MMPF é diferente do preditor obtido sob o modelo misto usual, pois a constante de encolhimento depende da média das variâncias individuais. Utilizamos simulação para comparar o preditor obtido sob o modelo misto usual (utilizado conforme a interpretação usual) com o preditor obtido sob o MMPF, mostrando que apesar do primeiro ser enviesado, ele geralmente apresenta erro quadrático médio (EQM) menor (ou ligeiramente maior) do que aquele obtido sob o MMPF. Adicionalmente, mostramos como utilizar dois pacotes de \\emph estatístico (Proc MIXED do SAS e lme(nlme) do R), construídos sob o modelo misto usual, para ajustar corretamente modelos em situações com erros exógenos e endógenos, heterocedásticos ou homocedásticos. / We consider optimal estimation and prediction of latent values based on data subject to endogenous and exogenous measurement errors, obtained via simple random sample from a finite population. We consider a finite population mixed model (FPMM) with endogenous and exogenous measurement errors proposed by Stanek III et al. (2004) and Stanek III & Singer (2004) and obtained the best linear unbiased predictor (BLUP) of the latent value of the i-th unit selected in the sample. When the endogenous variances are heteroscedastic, the predictor obtained under the FPMM is different than the predictor obtained with the usual mixed model, because the shrinkage constant depends on the average of the individual variances. We consider simulation studies to compare the predictor obtained under the usual mixed model (used according to the usual interpretation) with the predictor obtained under the FPMM, and show that the former is biased, but usually presents smaller (or slightly larger) mean squared error (MSE) than the predictor obtained under the FPMM. Additionally, we indicate how two commonly used statistical software packages (SAS\'s Proc MIXED and R\'s lme(nlme) ) may be employed to fit mixed models in situations with heteroscedastic or homoscedastic exogenous and endogenous errors.

Consideramos a predição ótima de valores latentes com base em dados sujeitos a erros de medida endógenos e exógenos, obtidos a partir de uma amostra aleatória de uma população finita. Consideramos o modelo misto para populações finitas (MMPF) com erros de medida exógenos e endógenos usando o enfoque proposto por Stanek et al. (2004) e Stanek & Singer (2004), e calculamos o melhor preditor linear não enviesado (BLUP) do valor latente da i-ésima unidade selecionada na amostra. Quando as variâncias endógenas são heterocedásticas, o preditor obtido sob o MMPF é diferente do preditor obtido sob o modelo misto usual, pois a constante de encolhimento depende da média das variâncias individuais. Utilizamos simulação para comparar o preditor obtido sob o modelo misto usual (utilizado conforme a interpretação usual) com o preditor obtido sob o MMPF, mostrando que apesar do primeiro ser enviesado, ele geralmente apresenta erro quadrático médio (EQM) menor (ou ligeiramente maior) do que aquele obtido sob o MMPF. Adicionalmente, mostramos como utilizar dois pacotes de \\emph estatístico (Proc MIXED do SAS e lme(nlme) do R), construídos sob o modelo misto usual, para ajustar corretamente modelos em situações com erros exógenos e endógenos, heterocedásticos ou homocedásticos. / We consider optimal estimation and prediction of latent values based on data subject to endogenous and exogenous measurement errors, obtained via simple random sample from a finite population. We consider a finite population mixed model (FPMM) with endogenous and exogenous measurement errors proposed by Stanek III et al. (2004) and Stanek III & Singer (2004) and obtained the best linear unbiased predictor (BLUP) of the latent value of the i-th unit selected in the sample. When the endogenous variances are heteroscedastic, the predictor obtained under the FPMM is different than the predictor obtained with the usual mixed model, because the shrinkage constant depends on the average of the individual variances. We consider simulation studies to compare the predictor obtained under the usual mixed model (used according to the usual interpretation) with the predictor obtained under the FPMM, and show that the former is biased, but usually presents smaller (or slightly larger) mean squared error (MSE) than the predictor obtained under the FPMM. Additionally, we indicate how two commonly used statistical software packages (SAS\'s Proc MIXED and R\'s lme(nlme) ) may be employed to fit mixed models in situations with heteroscedastic or homoscedastic exogenous and endogenous errors.

Measures of data quality are important for the evaluation and improvement of survey design and procedures. A detailed investigation of the sources, magnitude and impact of errors is necessary to identify how survey design and procedures may be improved and how resources allocated more efficiently among various aspects of the survey operation. A major part of this thesis is devoted to the overview of statistical theory and methods for measuring the contribution of response variability to the overall error of a survey. A very common practice in surveys is to select groups (clusters) of elements together instead of independent selection of elements. In practice cluster samples tend to produce higher sampling variance for statistics than element samples of the same size. Their frequent use stems from the desirable cost features that they have. Most data collection and sample designs involve some overlapping between interviewer workload and the sampling units (clusters). For those cases, a proportion of the measurement variance, which is due to interviewers, is reflected to some degree in the sampling variance calculations. The prime purpose in this thesis is to determine a variance formula that decomposes the total variance into sampling and measurement variance components for two commonly used data collection and sample designs. Once such a decomposition is obtained, determining an optimum allocation in existence of measurement errors would be possible.

Aquesta tesi s'emmarca en l'anàlisi de la supervivència. A la primera part es proposa una metodologia per estimar els coeficients d'un model lineal amb censura a la resposta i error de mesura a les covariants. Es proposa un estimador consistent que combina les metodologies per estimar models lineals de resposta censurada i, les utilitzades per als models amb variables explicatives mesurades amb error. Els errors estàndard de l'estimador es calculen numèricament mitjançant bootstrap. Algunes simulacions mostren les propietats de l'estimador.A la segona part s'analitzen històries laborals espanyoles. Hi ha tres tipus d'episodis: de treball autònom, per compte d'altri i sense treballar. Es duen a terme els següents estudis: estimació no paramètrica de la funció de supervivència pels diferents episodis; anàlisi de la durada del primer episodi de la història laboral; models de competing risk per a cada tipus d'episodis; i una anàlisi longitudinal dels cinc primers episodis tractats conjuntament. / This thesis is about survival analysis. The first part is focused on linear regression models with a censored dependent variable and, the explanatory variable contaminated with measurement error. A methodology which produces consistent estimates of the regression coefficients is proposed. It combines the procedures for fitting linear models with censoring and, the estimation methods for linear models with measurement error. Standard errors of the estimator are computed using the bootstrap method. The performance of the estimator is evaluated by Monte Carlo studies.The second part analyzes data about Spanish labor market histories. We study the duration of three types of labor spells (self-employment, wage-earner and non-working). We carry out several analyses: non-parametric approaches of the survival functions for several sets of spells a standard duration analysis for the first spells of the labor history a competing risk analysis for each type of spell; and a longitudinal analysis for the five early spells of the labor histories.

The thesis focuses on measurements of oscillators’ frequency stability. The theoretical part describes the basic concepts of measurement of frequency, measuring systems and measurement errors. The second part of thesis describes the design of measurement system. The design contains all data for build prototype and software, including all source codes.

The diploma thesis deals with the detection of measurement error based on the question order using the reaction time. The major part of the theoretical chapter deals with the respondent as the source of the error. A respondent needs to proceed through a cognitive process to answer a question. Various influences affect the process and they might have an effect on the accuracy of the answer. The frequently mentioned effects are age, education and motivation of respondent. The absence of respondents' motivation might lead to a reduction of their effort to properly proceed with the cognitive process. In that case, the respondent does not dedicate the appropriate cognitive effort and time to a question in the questionnaire. That length of time, which the respondent takes to answer, is called the reaction time and it is suggested that it might be used as an indicator of the amount of cognitive effort. Based on previously executed studies, the author designed hypotheses about the change in length of the reaction time with a change of question position in a questionnaire, about the effect of respondents' characteristics on the reaction time and its shortening. The significant shortening of the reaction time on the question located closer to the end of the experimental questionnaire is used as an indicator...

This diploma thesis is dealing with pressure measurement. It describes instruments dedicated to purpose - pressure gauges. Basic of correct pressure measurement is knowledge of behaviour of used pressure gauge. This knowledge is acquired by calibration. With calibration is associated also analysis of errors and uncertainty of measurement and creation of calibration list. Calibration takes place on model workspace pressure measuring in laboratory.