Statistics and Probability Commons^™

Joint Mixed-Effects Models For Longitudinal Data Analysis: An Application For The Metabolic Syndrome, John Thorp Iii

Theses and Dissertations

Mixed-effects models are commonly used to model longitudinal data as they can appropriately account for within and between subject sources of variability. Univariate mixed effect modeling strategies are well developed for a single outcome (response) variable that may be continuous (e.g. Gaussian) or categorical (e.g. binary, Poisson) in nature. Only recently have extensions been discussed for jointly modeling multiple outcome variables measures longitudinally. Many diseases processes are a function of several factors that are correlated. For example, the metabolic syndrome, a constellation of cardiovascular risk factors associated with an increased risk of cardiovascular disease and type 2 diabetes, is often …

A Comparison Of Frequentist And Bayesian Approaches To The Estimation Of Long-Stay Per-Diems, Jeff Hatcher, Jason M. Sutherland

Dartmouth Scholarship

Within many diagnosis related group (DRG) systems, there is recognition that a single cost weight per DRG is not suitable, and that cost weights should take into account extremely lengthy hospital stays. Long lengths of stay are considered to be due to factors largely beyond the control of the hospital, and a single weight per DRG would potentially place hospitals under financial risk.

Within Canada's acute-care, inpatient grouping methodology - Case Mix Groups (CMG+) - long-stay episodes represent approximately 4.5% of all discharges. Within a CMG (analogous to DRG), the cost weight assigned to long-stay cases consists of the typical …

Lot Quality Assurance Sampling (Lqas) And The Mozambique Malaria Indicator Surveys, Caitlin Biedron, Marcello Pagano, Bethany L. Hedt, Albert Kilian, Amy Ratcliffe, Samuel Mabunda, Joseph J. Valadez

Harvard University Biostatistics Working Paper Series

No abstract provided.

Modeling Multilevel Sleep Transitional Data Via Poisson Log-Linear Multilevel Models, Bruce J. Swihart, Brian Caffo, Ciprian Crainiceanu, Naresh M. Punjabi

Johns Hopkins University, Dept. of Biostatistics Working Papers

This paper proposes Poisson log-linear multilevel models to investigate population variability in sleep state transition rates. We specifically propose a Bayesian Poisson regression model that is more flexible, scalable to larger studies, and easily fit than other attempts in the literature. We further use hierarchical random effects to account for pairings of individuals and repeated measures within those individuals, as comparing diseased to non-diseased subjects while minimizing bias is of epidemiologic importance. We estimate essentially non-parametric piecewise constant hazards and smooth them, and allow for time varying covariates and segment of the night comparisons. The Bayesian Poisson regression is justified …

Bayesian Functional Data Analysis Using Winbugs, Ciprian M. Crainiceanu, A. Jeffrey Goldsmith

Johns Hopkins University, Dept. of Biostatistics Working Papers

We provide user friendly software for Bayesian analysis of Functional Data Models using WinBUGS 1.4. The excellent properties of Bayesian analysis in this context are due to: 1) dimensionality reduction, which leads to low dimensional projection bases; 2)the mixed model representation of functional models, which provides a modular approach to model extension; and 3) the orthogonality of the principal component bases, which contributes to excellent chain convergence and mixing properties. Our paper provides one more, essential, reason for using Bayesian analysis for Functional models: the existence of software.

Application Of The Truncated Skew Laplace Probability Distribution In Maintenance System, Gokarna R. Aryal, Chris P. Tsokos

Journal of Modern Applied Statistical Methods

A random variable X is said to have the skew-Laplace probability distribution if its pdf is given by f(x) = 2g(x)G(λx), where g (.) and G (.), respectively, denote the pdf and the cdf of the Laplace distribution. When the skew Laplace distribution is truncated on the left at 0 it is called it the truncated skew Laplace (TSL) distribution. This article provides a comparison of TSL distribution with twoparameter gamma model and the hypoexponential model, and an application of the subject model in maintenance system is studied.

Examples Of Computing Power For Zero-Inflated And Overdispersed Count Data, Suzanne R. Doyle

Journal of Modern Applied Statistical Methods

Examples of zero-inflated Poisson and negative binomial regression models were used to demonstrate conditional power estimation, utilizing the method of an expanded data set derived from probability weights based on assumed regression parameter values. SAS code is provided to calculate power for models with a binary or continuous covariate associated with zero-inflation.

An Inductive Approach To Calculate The Mle For The Double Exponential Distribution, W. J. Hurley

Journal of Modern Applied Statistical Methods

Norton (1984) presented a calculation of the MLE for the parameter of the double exponential distribution based on the calculus. An inductive approach is presented here.

New Effect Size Rules Of Thumb, Shlomo S. Sawilowsky

Journal of Modern Applied Statistical Methods

Recommendations to expand Cohen’s (1988) rules of thumb for interpreting effect sizes are given to include very small, very large, and huge effect sizes. The reasons for the expansion, and implications for designing Monte Carlo studies, are discussed.

Generating And Comparing Aggregate Variables For Use Across Datasets In Multilevel Analysis, James Chowhan, Laura Duncan

Journal of Modern Applied Statistical Methods

This article examines the creation of contextual aggregate variables from one dataset for use with another dataset in multilevel analysis. The process of generating aggregate variables and methods of assessing the validity of the constructed aggregates are presented, together with the difficulties that this approach presents.

Detecting Lag-One Autocorrelation In Interrupted Time Series Experiments With Small Datasets, Clare Riviello, S. Natasha Beretvas

Journal of Modern Applied Statistical Methods

The power and type I error rates of eight indices for lag-one autocorrelation detection were assessed for interrupted time series experiments (ITSEs) with small numbers of data points. Performance of Huitema and McKean’s (2000) z_HM statistic was modified and compared with the z_HM, five information criteria and the Durbin-Watson statistic.

Relationship Between Internal Consistency And Goodness Of Fit Maximum Likelihood Factor Analysis With Varimax Rotation, Gibbs Y. Kanyongo, James B. Schreiber

Journal of Modern Applied Statistical Methods

This study investigates how reliability (internal consistency) affects model-fitting in maximum likelihood exploratory factor analysis (EFA). This was accomplished through an examination of goodness of fit indices between the population and the sample matrices. Monte Carlo simulations were performed to create pseudo-populations with known parameters. Results indicated that the higher the internal consistency the worse the fit. It is postulated that the observations are similar to those from structural equation modeling where a good fit with low correlations can be observed and also the reverse with higher item correlations.

Estimating Model Complexity Of Feed-Forward Neural Networks, Douglas Landsittel

Journal of Modern Applied Statistical Methods

In a previous simulation study, the complexity of neural networks for limited cases of binary and normally-distributed variables based the null distribution of the likelihood ratio statistic and the corresponding chi-square distribution was characterized. This study expands on those results and presents a more general formulation for calculating degrees of freedom.

Level Robust Methods Based On The Least Squares Regression Estimator, Marie Ng, Rand R. Wilcox

Journal of Modern Applied Statistical Methods

Heteroscedastic consistent covariance matrix (HCCM) estimators provide ways for testing hypotheses about regression coefficients under heteroscedasticity. Recent studies have found that methods combining the HCCM-based test statistic with the wild bootstrap consistently perform better than non-bootstrap HCCM-based methods (Davidson & Flachaire, 2008; Flachaire, 2005; Godfrey, 2006). This finding is more closely examined by considering a broader range of situations which were not included in any of the previous studies. In addition, the latest version of HCCM, HC5 (Cribari-Neto, et al., 2007), is evaluated.

Least Error Sample Distribution Function, Vassili F. Pastushenko

Journal of Modern Applied Statistical Methods

Email: The empirical distribution function (ecdf) is unbiased in the usual sense, but shows certain order bias. Pyke suggested discrete ecdf using expectations of order statistics. Piecewise constant optimal ecdf saves 200%/N of sample size N. Results are compared with linear interpolation for U(0, 1), which require up to sixfold shorter samples at the same accuracy.

Confidence Interval Estimation For Intraclass Correlation Coefficient Under Unequal Family Sizes, Madhusudan Bhandary, Koji Fujiwara

Journal of Modern Applied Statistical Methods

Confidence intervals (based on the χ² -distribution and (Z) standard normal distribution) for the intraclass correlation coefficient under unequal family sizes based on a single multinormal sample have been proposed. It has been found that the confidence interval based on the χ² -distribution consistently and reliably produces better results in terms of shorter average interval length than the confidence interval based on the standard normal distribution: especially for larger sample sizes for various intraclass correlation coefficient values. The coverage probability of the interval based on the χ² -distribution is competitive with the coverage probability of the interval …

On Some Discrete Distributions And Their Applications With Real Life Data, Shipra Banik, B. M. Golam Kibria

Journal of Modern Applied Statistical Methods

This article reviews some useful discrete models and compares their performance in terms of the high frequency of zeroes, which is observed in many discrete data (e.g., motor crash, earthquake, strike data, etc.). A simulation study is conducted to determine how commonly used discrete models (such as the binomial, Poisson, negative binomial, zero-inflated and zero-truncated models) behave if excess zeroes are present in the data. Results indicate that the negative binomial model and the ZIP model are better able to capture the effect of excess zeroes. Some real-life environmental data are used to illustrate the performance of the proposed models.

Closed Form Confidence Intervals For Small Sample Matched Proportions, James F. Reed Iii

Journal of Modern Applied Statistical Methods

The behavior of the Wald-z, Wald-c, Quesenberry-Hurst, Wald-m and Agresti-Min methods was investigated for matched proportions confidence intervals. It was concluded that given the widespread use of the repeated-measure design, pretest-posttest design, matched-pairs design, and cross-over design, the textbook Wald-z method should be abandoned in favor of the Agresti-Min alternative.

On Type-Ii Progressively Hybrid Censoring, Debasis Kundu, Avijit Joarder, Hare Krishna

Journal of Modern Applied Statistical Methods

The progressive Type-II censoring scheme has become quite popular. A drawback of a progressive censoring scheme is that the length of the experiment can be very large if the items are highly reliable. Recently, Kundu and Joarder (2006) introduced the Type-II progressively hybrid censored scheme and analyzed the data assuming that the lifetimes of the items are exponentially distributed. This article presents the analysis of Type-II progressively hybrid censored data when the lifetime distributions of the items follow Weibull distributions. Maximum likelihood estimators and approximate maximum likelihood estimators are developed for estimating the unknown parameters. Asymptotic confidence intervals based on …

Multiple Search Paths And The General-To-Specific Methodology, Paul Turner

Journal of Modern Applied Statistical Methods

Increased interest in computer automation of the general-to-specific methodology has resulted from research by Hoover and Perez (1999) and Krolzig and Hendry (2001). This article presents simulation results for a multiple search path algorithm that has better properties than those generated by a single search path. The most noticeable improvements occur when the data contain unit roots.

Ordinal Regression Analysis: Fitting The Proportional Odds Model Using Stata, Sas And Spss, Xing Liu

Journal of Modern Applied Statistical Methods

Researchers have a variety of options when choosing statistical software packages that can perform ordinal logistic regression analyses. However, statistical software, such as Stata, SAS, and SPSS, may use different techniques to estimate the parameters. The purpose of this article is to (1) illustrate the use of Stata, SAS and SPSS to fit proportional odds models using educational data; and (2) compare the features and results for fitting the proportional odds model using Stata OLOGIT, SAS PROC LOGISTIC (ascending and descending), and SPSS PLUM. The assumption of the proportional odds was tested, and the results of the fitted models were …

Estimation Of The Standardized Mean Difference For Repeated Measures Designs, Lindsey J. Wolff Smith, S. Natasha Beretvas

Journal of Modern Applied Statistical Methods

This simulation study modified the repeated measures mean difference effect size, d⁼_RM , for scenarios with unequal pre- and post-test score variances. Relative parameter and SE bias were calculated for dRM ≠ versus dRM = . Results consistently favored d^≠_RM over d⁼_RM with worse positive parameter and negative SE bias identified for d⁼_RM for increasingly heterogeneous variance conditions.

Estimating The Parameters Of Rayleigh Cumulative Exposure Model In Simple Step-Stress Testing. Natasha Beretvas Is An, Mohammed Al-Haj Ebrahem, Abedel-Qader Al-Masri

Journal of Modern Applied Statistical Methods

Assumes the life distribution of a test unit for any stress follows a Rayleigh distribution with scale parameterθ , and that Ln(θ ) is a linear function of the stress level. Maximum likelihood estimators of the parameters under a cumulative exposure model are obtained. The approximate variance estimates obtained from the asymptotic normal distribution of the maximum likelihood estimators are used to construct confidence intervals for the model parameters. A simulation study was conducted to study the performance of the estimators. Simulation results showed that in terms of bias, mean squared error, attainment of the nominal confidence level, symmetry …

Test For The Equality Of The Number Of Signals, Madhusudan Bhandary, Debasis Kundu

Journal of Modern Applied Statistical Methods

A likelihood ratio test for testing the equality of the ranks of two non-negative definite covariance matrices arising in the area of signal processing is derived. The asymptotic distribution of the test statistic follows a Chi-square distribution from the general theory of likelihood ratio test.

Jmasm29: Dominance Analysis Of Independent Data (Fortran), Du Feng, Normal Cliff

Journal of Modern Applied Statistical Methods

A Fortran 77 program is provided for an ordinal dominance analysis of independent two-group comparisons. The program calculates the ordinal statistic, d, and statistical inferences about δ. The source codes and an executable file are available at http://www.depts.ttu.edu/hdfs/feng.php.

Analysis Of Multifactor Experimental Designs, Phillip I. Good

Journal of Modern Applied Statistical Methods

In the one-factor case, Good and Lunneborg (2006) showed that the permutation test is superior to the analysis of variance. In the multi-factor case, simulations reveal the reverse is true. The analysis of variance is remarkably robust against departures from normality including instances in which data is drawn from mixtures of normal distributions or from Weibull distributions. The traditional permutation test based on all rearrangements of the data labels is not exact and is more powerful that the analysis of variance only for 2xC designs or when there is only a single significant effect. Permutation tests restricted to synchronized permutations …

Assessing Trends: Monte Carlo Trials With Four Different Regression Methods, Daniel R. Thompson

Journal of Modern Applied Statistical Methods

Ordinary Least Squares (OLS), Poisson, Negative Binomial, and Quasi-Poisson Regression methods were assessed for testing the statistical significance of a trend by performing 10,000 simulations. The Poisson method should be used when data follow a Poisson distribution. The other methods should be used when data follow a normal distribution.

Approximate Bayesian Confidence Intervals For The Mean Of A Gaussian Distribution Versus Bayesian Models, Vincent A. R. Camara

Journal of Modern Applied Statistical Methods

This study obtained and compared confidence intervals for the mean of a Gaussian distribution. Considering the square error and the Higgins-Tsokos loss functions, approximate Bayesian confidence intervals for the mean of a normal population are derived. Using normal data and SAS software, the obtained approximate Bayesian confidence intervals were compared to a published Bayesian model. Whereas the published Bayesian method is sensitive to the choice of the hyper-parameters and does not always yield the best confidence intervals, it is shown that the proposed approximate Bayesian approach relies only on the observations and often performs better.

Semi-Parametric Of Sample Selection Model Using Fuzzy Concepts, L. Muhamad Safiih, A. A. Kamil, M. T. Abu Osman

Journal of Modern Applied Statistical Methods

The sample selection model has been studied in the context of semi-parametric methods. With the deficiencies of the parametric model, such as inconsistent estimators, semi-parametric estimation methods provide better alternatives. This article focuses on the context of fuzzy concepts as a hybrid to the semiparametric sample selection model. The better approach when confronted with uncertainty and ambiguity is to use the tools provided by the theory of fuzzy sets, which are appropriate for modeling vague concepts. A fuzzy membership function for solving uncertainty data of a semi-parametric sample selection model is introduced as a solution to the problem.

Performance Ratings Of An Autocovariance Base Estimator (Abe) In The Estimation Of Garch Model Parameters When The Normality Assumption Is Invalid, Daniel Eni

Journal of Modern Applied Statistical Methods

The performance of an autocovariance base estimator (ABE) for GARCH models against that of the maximum likelihood estimator (MLE) if a distribution assumption is wrongly specified as normal was studied. This was accomplished by simulating time series data that fits a GARCH model using the Log normal and t-distributions with degrees of freedom of 5, 10 and 15. The simulated time series was considered as the true probability distribution, but normality was assumed in the process of parameter estimations. To track consistency, sample sizes of 200, 500, 1,000 and 1,200 were employed. The two methods were then used to analyze …