Master of Science / Department of Statistics / Christopher I. Vahl / In the classic split-plot design where whole plots have a completely randomized design, the conventional analysis approach assumes a compound symmetry (CS) covariance structure for the errors of observation. However, often this assumption may not be true. In this report, we examine using different covariance models in PROC MIXED in the SAS system, which are widely used in the repeated measures analysis, to model the covariance structure in the split-plot data in which the simple compound symmetry assumption does not hold. The comparison of the covariance structure models in PROC MIXED and the conventional split-plot model is illustrated through a simulation study. In the example analyzed, the heterogeneous compound symmetry (CSH) covariance model has the smallest values for the Akaike and Schwarz’s Bayesian information criteria fit statistics and is therefore the best model to fit our example data.

This dissertation focuses on issues related to fitting an optimal variance-covariance structure in multilevel linear modeling framework with two Monte Carlo simulation studies. In the first study, the author evaluated the performance of common fit statistics such as Likelihood Ratio Test (LRT), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC) and a new proposed method, standardized root mean square residual (SRMR), for selecting the correct within-subject covariance structure. Results from the simulated data suggested SRMR had the best performance in selecting the optimal covariance structure. A pharmaceutical example was also used to evaluate the performance of these fit statistics empirically. The LRT failed to decide which is a better model because LRT can only be used for nested models. SRMR, on the other hand, had congruent result as AIC and BIC and chose ARMA(1,1) as the optimal variance-covariance structure. In the second study, the author adopted a first-order autoregressive structure as the true within-subject V-C structure with variability in the intercept and slope (estimating [tau]00 and [tau]11 only) and investigated the consequence of misspecifying different levels/types of the V-C matrices simultaneously on the estimation and test of significance for the growth/fixed-effect and random-effect parameters, considering the size of the autoregressive parameter, magnitude of the fixed effect parameters, number of cases, and number of waves. The result of the simulation study showed that the commonly-used identity within-subject structure with unstructured between-subject matrix performed equally well as the true model in the evaluation of the criterion variables. On the other hand, other misspecified conditions, such as Under G & Over R conditions and Generally misspecified G & R conditions had biased standard error estimates for the fixed effect and lead to inflated Type I error rate or lowered statistical power. The two studies bridged the gap between the theory and practical application in the current literature. More research can be done to test the effectiveness of proposed SRMR in searching for the optimal V-C structure under different conditions and evaluate the impact of different types/levels of misspecification with various specifications of the within- and between- level V-C structures simultaneously.

In this paper we consider conditional maximum likelihood (cml) estimates for item parameters in the Rasch model under random subject parameters. We give a simple approximation for the asymptotic covariance matrix of the cml-estimates. The approximation is stated as a limit theorem when the number of item parameters goes to infinity. The results contain precise mathematical information on the order of approximation. The results enable the analysis of the covariance structure of cml-estimates when the number of items is large. Let us give a rough picture. The covariance matrix has a dominating main diagonal containing the asymptotic variances of the estimators. These variances are almost equal to the efficient variances under ml-estimation when the distribution of the subject parameter is known. Apart from very small numbers n of item parameters the variances are almost not affected by the number n. The covariances are more or less negligible when the number of item parameters is large. Although this picture intuitively is not surprising it has to be established in precise mathematical terms. This has been done in the present paper. The paper is based on previous results [5] of the author concerning conditional distributions of non-identical replications of Bernoulli trials. The mathematical background are Edgeworth expansions for the central limit theorem. These previous results are the basis of approximations for the Fisher information matrices of cmlestimates. The main results of the present paper are concerned with the approximation of the covariance matrices. Numerical illustrations of the results and numerical experiments based on the results are presented in Strasser, [6].

FMRI datasets allow scientists to assess functionality of the brain by measuring the response of blood flow to a stimulus. Since the responses from neighboring locations within the brain are correlated, simple linear models that assume independence of measurements across locations are inadequate. Mixed models can be used to model the spatial correlation between observations, however selecting the correct covariance structure is difficult. Information criteria, such as AIC are often used to choose among covariance structures. Once the covariance structure is selected, significance tests can be used to determine if a region of interest within the brain is significantly active. Through the use of simulations, this project explores the performance of AIC in selecting the covariance structure. Type I error rates are presented for the fixed effects using the the AIC chosen covariance structure. Power of the fixed effects are also discussed.

No description available.

The dissertation focuses on developing statistical models in environmental and life sciences. The Generalized Extreme Value distribution is used to model annual monthly maximum rainfall data from 44 locations in Florida. Time dependence of the rainfall data is incorporated into the model by assuming the location parameter to be a function of time, both linear and quadratic. Estimates and confidence intervals are obtained for return levels of return periods of 10, 20, 50, and 100 years. Locations are grouped into statistical profiles based on their similarities in return level graphs for all locations, and locations within each climatic zone. A family of extreme values distributions is applied to model simulated maximum drug concentration (Cmax) data of an anticoagulant drug. For small samples (n <̲ 100) data exhibited bimodality. The results of investigating a mixture of two extreme value distributions to model such bimodal data using two-parameter Gumbel, Pareto and Weibu ll concluded that a mixture of two Weibull distributions is the only suitable FTSel.For large samples , Cmax data are modeled using the Generalized Extreme Value, Gumbel, Weibull, and Pareto distributions. These results concluded that the Generalized Extreme Value distribution is the only suitable model. A system of random differential equations is used to investigate the drug concentration behavior in a three-compartment pharmacokinetic model which describes coumermycin's disposition. The rate constants used in the differential equations are assumed to have a trivariate distribution, and hence, simulated from the trivariate truncated normal probability distribution. Numerical solutions are developed under different combinations of the covariance structure and the nonrandom initial conditions. We study the dependence effect that such a pharmacokinetic system has among the three compartments as well as the effect of variance in identifying the concentration behavior in each compartment. We identify the time delays in each compartment. We extend these models to incorporate the identified time delays. We provide the graphical display of the time delay effects on the drug concentration behavior as well as the comparison of the deterministic behavior with and without the time delay, and effect of different sets of time delay on deterministic and stochastic behaviors.

Today's high-throughput data collection devices, e.g. spectrometers and gene chips, create information in abundance. However, this poses serious statistical challenges, as the number of features is usually much larger than the number of observed units.  Further, in this high-dimensional setting, only a small fraction of the features are likely to be informative for any specific project. In this thesis, three different approaches to the two-class supervised classification in this high-dimensional, low sample setting are considered. There are classifiers that are known to mitigate the issues of high-dimensionality, e.g. distance-based classifiers such as Naive Bayes. However, these classifiers are often computationally intensive and therefore less time-consuming for discrete data. Hence, continuous features are often transformed into discrete features. In the first paper, a discretization algorithm suitable for high-dimensional data is suggested and compared with other discretization approaches. Further, the effect of discretization on misclassification probability in high-dimensional setting is evaluated.   Linear classifiers are more stable which motivate adjusting the linear discriminant procedure to high-dimensional setting. In the second paper, a two-stage estimation procedure of the inverse covariance matrix, applying Lasso-based regularization and Cuthill-McKee ordering is suggested. The estimation gives a block-diagonal approximation of the covariance matrix which in turn leads to an additive classifier. In the third paper, an asymptotic framework that represents sparse and weak block models is derived and a technique for block-wise feature selection is proposed.      Probabilistic classifiers have the advantage of providing the probability of membership in each class for new observations rather than simply assigning to a class. In the fourth paper, a method is developed for constructing a Bayesian predictive classifier. Given the block-diagonal covariance matrix, the resulting Bayesian predictive and marginal classifier provides an efficient solution to the high-dimensional problem by splitting it into smaller tractable problems. The relevance and benefits of the proposed methods are illustrated using both simulated and real data. / Med dagens teknik, till exempel spektrometer och genchips, alstras data i stora mängder. Detta överflöd av data är inte bara till fördel utan orsakar även vissa problem, vanligtvis är antalet variabler (p) betydligt fler än antalet observation (n). Detta ger så kallat högdimensionella data vilket kräver nya statistiska metoder, då de traditionella metoderna är utvecklade för den omvända situationen (p&lt;n).  Dessutom är det vanligtvis väldigt få av alla dessa variabler som är relevanta för något givet projekt och styrkan på informationen hos de relevanta variablerna är ofta svag. Därav brukar denna typ av data benämnas som gles och svag (sparse and weak). Vanligtvis brukar identifiering av de relevanta variablerna liknas vid att hitta en nål i en höstack. Denna avhandling tar upp tre olika sätt att klassificera i denna typ av högdimensionella data.  Där klassificera innebär, att genom ha tillgång till ett dataset med både förklaringsvariabler och en utfallsvariabel, lära en funktion eller algoritm hur den skall kunna förutspå utfallsvariabeln baserat på endast förklaringsvariablerna. Den typ av riktiga data som används i avhandlingen är microarrays, det är cellprov som visar aktivitet hos generna i cellen. Målet med klassificeringen är att med hjälp av variationen i aktivitet hos de tusentals gener (förklaringsvariablerna) avgöra huruvida cellprovet kommer från cancervävnad eller normalvävnad (utfallsvariabeln). Det finns klassificeringsmetoder som kan hantera högdimensionella data men dessa är ofta beräkningsintensiva, därav fungera de ofta bättre för diskreta data. Genom att transformera kontinuerliga variabler till diskreta (diskretisera) kan beräkningstiden reduceras och göra klassificeringen mer effektiv. I avhandlingen studeras huruvida av diskretisering påverkar klassificeringens prediceringsnoggrannhet och en mycket effektiv diskretiseringsmetod för högdimensionella data föreslås. Linjära klassificeringsmetoder har fördelen att vara stabila. Nackdelen är att de kräver en inverterbar kovariansmatris och vilket kovariansmatrisen inte är för högdimensionella data. I avhandlingen föreslås ett sätt att skatta inversen för glesa kovariansmatriser med blockdiagonalmatris. Denna matris har dessutom fördelen att det leder till additiv klassificering vilket möjliggör att välja hela block av relevanta variabler. I avhandlingen presenteras även en metod för att identifiera och välja ut blocken. Det finns också probabilistiska klassificeringsmetoder som har fördelen att ge sannolikheten att tillhöra vardera av de möjliga utfallen för en observation, inte som de flesta andra klassificeringsmetoder som bara predicerar utfallet. I avhandlingen förslås en sådan Bayesiansk metod, givet den blockdiagonala matrisen och normalfördelade utfallsklasser. De i avhandlingen förslagna metodernas relevans och fördelar är visade genom att tillämpa dem på simulerade och riktiga högdimensionella data.

国立情報学研究所で電子化したコンテンツを使用している。

Because the way mothers play with their children may have significant impacts on children's social, cognitive, and linguistic development, researchers have become interested in potential predictors of maternal play. In the present study, 40 mother–infant dyads were followed from child age 5–20 months. Five-month habituation rate and 13 and 20 month temperamental difficulty were found to be predictive of maternal play quality at 20 months. The most parsimonious theoretical model was one in which habituation was mediated by temperamental difficulty in predicting mother play. Consistent with prior speculation in the literature, these data support the possibility that mothers adjust some aspects of their play behaviors to fit their children's cognitive and temperamental capabilities.

Therapeutic Assessment (TA) is a hybrid of assessment and therapy techniques in which assessors actively collaborate with clients during an individualized assessment. TA is centered around client assessment questions and provides a safe environment where clients can create shifts in their ‘story’ of self. More specifically, TA with children and their parents has demonstrated more confident parenting and parents’ better understanding of their child’s difficulties, while children have shown decreased problem behaviors and improved social/emotional functioning. The theoretical framework behind TA emphasizes the importance of the interpersonal interactions between the assessor and client, such as the development of a strong assessor client relationship and collaboration. These interpersonal processes are conceptualized as catalysts for greater depth of parent investment in the assessment and deeper levels of feedback results. The need for greater parent involvement and partnership in child mental health services is increasingly recognized in the client/parent satisfaction literature. Parent feedback to child mental health services is most often acquired through satisfaction questionnaires. However, the satisfaction literature has well known limitations, specifically a lack of unifying theory and methodological issues in scale development. Parent satisfaction research indicates that interpersonal experiences are more related to satisfaction than outcomes or client characteristics, and that more psychometrically sound measures are needed. Currently, satisfaction surveys do not provide a detailed understanding of parents’ experiences to inform practice and research. The current study outlines the development of the Parent Experience of Assessment Scale (PEAS). The PEAS is anchored in the theoretical orientation of TA and provides a more quantifiable measure of hypothesized underlying TA constructs. Moreover, the development of the PEAS uses advanced statistical techniques, such as Confirmatory Factor Analysis and invariance testing, to provide a higher level of psychometric rigor. The resulting scale consists of 24 items divided among 5 subscales with demonstrated relationships to general satisfaction. Structural equation modeling provides insight via direct and indirect effects among the PEAS subscales and their relationship to general satisfaction. Through the development of the PEAS, this study provides empirical evidence and support for TA theory and a more nuanced understanding of parent experiences related to satisfaction. / text