In this paper we consider discrimination-aware classification of educational data. Mining and using rules that distinguish groups of students based on sensitive attributes such as gender and nationality may lead to discrimination. It is desirable to keep the sensitive attributes during the training of a classifier to avoid information loss but…

Completion rates for massive open online classes (MOOCs) are notoriously low, but learner intent is an important factor. By studying students who drop out despite their intent to complete the MOOC, it may be possible to develop interventions to improve retention and learning outcomes. Previous research into predicting MOOC completion has focused…

Our research project focuses on learning strategies and motivation among academic procrastinators in computer assisted language learning (CALL) settings. In this study, we aim to compare them according to students' levels of English proficiency. One hundred and fourteen university students participated in this research project. Sixty-four students…

C. Huberty (1994) recently noted that "It is quite common to find the use of 'stepwise analyses' reported in empirically based journal articles." Stepwise methods are used (incorrectly) by some researchers either to select variables to retain for further analyses or to evaluate the relative importance of various variables. Of course,…

This paper presents three variable selection strategies in discriminate analysis (all variables in the model, use of stepwise methods, and all possible subsets). All three methods are illustrated through examples. Although the all variables in the model and the stepwise methods are the most widely used, B. Thompson (1996) and C. Huberty (1994)…

Both predictive discriminant analysis (PDA) and descriptive discriminant analysis (DDA) require a decision to pool group covariance matrices, or alternatively, to retain separate group covariance matrices when the group covariance matrices are too dissimilar to pool. Pooling the group covariance matrices involves the so-called "linear"…

Prediction of group membership is the goal of predictive discriminant analysis (PDA) and the accuracy of group classification is the focus of PDA. The purpose of this paper is to provide an overview of how PDA works and how it can be used to answer a variety of research questions. The paper explains what PDA is and why it is important, and it…

Two nonparametric statistical methods, the inverse normal scores method and the rank order transformation, are compared for use in discriminant function analysis. The methods are compared for both normal and non-normal distributions. When the distributions are normal, the rank and inverse normal scores methods are effective substitutes for the…

The stabilities of standardized (ß) and structure (rs) coefficients in canonical (CA) and discriminant analyses (DA) were studied. Four different situations were studied--two pertaining to CA and two to DA. The situations were meant to represent "somewhat typical" and yet varying research conditions that often would not be thought to be…

Accurate classification in discriminant analysis and the value of prediction are discussed, with emphasis on the uses and key aspects of prediction. A brief history of discriminant analysis is provided. C. J. Huberty's discussion of four aspects of discriminant analysis (separation, discrimination, estimation, and classification) is cited.…

To determine if coupon collecting has any influence on product information processing like brand awareness, and to understand better the coupon collecting process, an exploratory study examined the relationship between coupon-collecting behavior and brand awareness for the coupons collected. Subjects, 152 randomly chosen respondents from a Midwest…

We consider the problem of classifying an unknown observation into one of several populations using tree-structured allocation rules. Although many parametric classification procedures are robust to certain assumption violations, there is need for discriminant procedures that can be utilized regardless of the group-conditional distributions that…

A method for comparing the cross-validated classification accuracy of Fisher's linear classification functions (FLCFs) and the least absolute deviation is presented under varying data conditions for the two-group classification problem. With this method, separate-group as well as total-sample proportions of current classifications can be compared…

The use of stepwise methodologies has been sharply criticized by several researchers, yet their popularity, especially in educational and psychological research, continues unabated. Stepwise methods have been considered particularly well suited for use in regression and discriminant analyses, but their use in discriminant analysis (predictive…

Cross-validation, an economical method for assessing whether sample results will generalize, is discussed in this paper. Cross-validation is an invariance technique that uses two subsets of the data sample to derive discriminant function coefficients. The two sets of coefficients are then used with each data subset to derive discriminant function…