Robiah Adnan

A simple more general boxplot method for identifying outliers

Abstract The boxplot method (Exploratory Data Analysis, Addison-Wesley, Reading, MA, 1977) is a graphically-based method of identifying outliers which is appealing not only in its simplicity but also because it does not use the extreme potential outliers in computing a measure of dispersion. The inner and outer fences are defined in terms of the hinges (or fourths), and therefore are not distorted by a few extreme values. Such distortion could lead to failing to detect some outliers, a problem known as “masking”. A method for determining the probability associated with any fence or observation is proposed based on the cumulative distribution function of the order statistics. This allows the statistician to easily assess, in a probability sense, the degree to which an observation is dissimilar to the majority of the observations. In addition, an adaptation for approximately normal but somewhat asymmetric distributions is suggested.

Extraction of dynamic features from hand drawn data for the identification of children with handwriting difficulty

Studies have shown that differences between children with and without handwriting difficulties lie not only in the written product (static data) but also in dynamic data of handwriting process. Since writing system varies among countries and individuals, this study was conducted to determine the feasibility of using quantitative outcome measures of childrens drawing to identify children who are at risk of handwriting difficulties. A sample of 143 first graders of a normal primary school was investigated regarding their handwriting ability. The children were divided into two groups: test and control. Ten children from test group and 40 children from control group were individually tested for their Visual Motor Integration skills. Analysis on dynamic data indicated significant differences between the two groups in temporal and spatial measures of the drawing task performance. Thus, kinematic analysis of childrens drawing is feasible to provide performance characteristic of handwriting ability, supporting its use in screening for handwriting difficulty.

The use of graphic rules in grade one to help identify children at risk of handwriting difficulties.

Previous researches on elementary grade handwriting revealed that pupils employ certain strategy when writing or drawing. The relationship between this strategy and the use of graphic rules has been documented but very little research has been devoted to the connection between the use of graphic rules and handwriting proficiency. Thus, this study was conducted to investigate the relative contribution of the use of graphic rules to the writing ability. A sample of 105 first graders who were average printers and 65 first graders who might experience handwriting difficulty, as judged by their teachers, of a normal primary school were individually tested on their use of graphic rules. It has been found that pupils who are below average printers use more non-analytic strategy than average printers to reproduce the figures. The results also reveal that below average printers do not acquire the graphic principles that foster an analytic approach to production skills. Although the findings are not sufficient to allow definitive conclusions about handwriting ability, it can be considered as one of the screening measures in identifying pupils who are at risk of handwriting difficulties.

Model comparison of Bayesian structural equation models with mixed ordered categorical and dichotomous data

Abstract The purpose of this paper is to describe the mixed variables (ordered categorical and dichotomous) in Bayesian structural equation models. Markov chain Monte Carlo simulation (MCMC) via Gibbs sampling method is applied for estimation the parameters. Statistical analyses, which include parameters estimation, standard error, higest posterior density and Devience information creterion for testing the prposed models, are discussed. Hidden continuous normal distribution with censoring is used to handle the problem of mixed variables (ordered categorical and dichotomous). Comparison between Bayesian linear and non-linear SEMs are discussed. The proposed models are illustrated by a case study for breast cancer patient’s which obtained from the hospital. Analyses are done by using WinBUGS program. The results showed that the results of non-linear Bayesian SEM is better than the results of linear Bayesian SEM.

Estimation parameters using bisquare weighted robust ridge regression BRLTS estimator in the presence of multicollinearity and outliers

This study presents an improvement to robust ridge regression estimator. We proposed two methods Bisquare ridge least trimmed squares (BRLTS) and Bisquare ridge least absolute value (BRLAV) based on ridge least trimmed squares (RLTS) and ridge least absolute value (RLAV), respectively. We compared these methods with existing estimators, namely ordinary least squares (OLS) and Huber ridge regression (HRID) using three criteria: Bias, Root Mean Square Error (RMSE) and Standard Error (SE) to estimate the parameters coefficients. The results of Bisquare ridge least trimmed squares (BRLTS) and Bisquare ridge least absolute value (BRLAV) are compared with existing methods using real data and simulation study. The empirical evidence shows that the results obtain from the BRLTS are the best among the three estimators followed by BRLAV with the least value of the RMSE for the different disturbance distributions and degrees of multicollinearity.

A Multivariate Control Chart for Expected Changes Over Time

A new procedure is proposed that merges the methodologies of regression control charts and multivariate control charts. The simulation of this procedure shows that this new control chart can be very effective in detecting even modestly larger than expected changes in several monitored variables which are subject to some naturally occurring changes over time.

Model comparison of linear and nonlinear Bayesian structural equation models with dichotomous data

ABSTRACT In this article, dichotomous variables are used to compare between linear and nonlinear Bayesian structural equation models. Gibbs sampling method is applied for estimation and model comparison. Statistical inferences that involve estimation of parameters and their standard deviations and residuals analysis for testing the selected model are discussed. Hidden continuous normal distribution (censored normal distribution) is used to solve the problem of dichotomous variables. The proposed procedure is illustrated by a simulation data obtained from R program. Analyses are done by using R2WinBUGS package in R-program.

Robust PC with wild bootstrap estimation of linear model in the presence of outliers, multicollinearity and heteroscedasticity error variance

The regression model estimator is considered efficient if it is robust and resistant to the presence of heteroscedasticity variance, multicollinearity or unusual observations called outliers. However, in regard to these problems, the wild bootstrap and robust wild bootstrap are no longer efficient since they could not produce the smallest variance. Hence this research investigates the use of robust PC with wild bootstrap techniques on regression model as an estimator for real and simulation data in a situation where multicollinearity, heteroscedasticity and multiple outliers are present. This paper proposed a robust procedure based on the weighted residuals which combined the Tukey bisquare weighted function, principal component analysis (PCA) to remedy the multicollinearity problems, least trimmed squares (LTS) estimator, robust location and scale, and the wild bootstrap sampling procedure of Wu and Liu that remedy the heteroscedasticity error variance. RPCWBootWu and RPCWBootLiu were obtained through a modified version of RBootWu and RBootLiu. Finally, based on the real data and simulation study, the performance of the RPCWBootWu and RPCWBootLiu is compared with the existing RBootWu, RBootLiu and also with BootWu, BootLiu using the biased, RMSE and standard error. The numerical example and simulation study shows that the RPCWBootWu and RPCWBootLiu techniques have proven to be a good alternative estimator for regression model with lower standard error values.

Comparison between Bayesian structural equation models with ordered categorical data

In this paper, ordered categorical variables are used to compare between linear and nonlinear Bayesian structural equation models. Gibbs sampling method is applied for estimation and model comparison. Statistical analyses, which involve estimation of parameters and their standard deviations for testing the selected model, are discussed. The proposed procedure is illustrated by a simulation data obtained from R program. Data results are obtained from WinBUGS program.