This paper provides an alternative way of document representation by treating topic probabilities as a vector representation for words and representing a document as a combination of the word vectors. A comparison on summary data shows that this representation is more effective in document classification. [This paper was published in:…

Suppose an education researcher wants to test the impact of a high school drop-out prevention intervention in which at-risk students attend classes to receive intensive summer school instruction. The district will allow the researcher to randomly assign students to the treatment classes or to the control group. Half of the students (the treatment…

Although the importance of primary schools in the long term is of interest in educational effectiveness research, few studies have examined the long-term effects of schools over the past decades. In the present study, long-term effects of primary schools on the educational positions of students 2 and 4 years after starting secondary education are…

Data mining is known to have a potential for predicting user performance. However, there are few studies that explore its potential for predicting student behavior in a procedural training environment. This paper presents a collective student model, which is built from past student logs. These logs are first grouped into clusters. Then, an…

Learning a more distributed representation of the input feature space is a powerful method to boost the performance of a given predictor. Often this is accomplished by partitioning the data into homogeneous groups by clustering so that separate models could be trained on each cluster. Intuitively each such predictor is a better representative of…

Responses of urban school teachers to the Pupil Control Ideology questionnaire were studied using Latent Class Analysis. The results of the analysis suggest that the best fitting model to the data is a two-cluster solution. In particular, the pupil control ideology of the sample delineates into two clusters of teachers, those with humanistic and…

Tests the premise of peer cluster theory as it applies to individual alcohol use, and makes a comparative analysis between its ability to explain alcohol use and marijuana use among college students (N=1312). Results of the causal models show some support for peer cluster theory. Discusses the study's limitations and implications. (Author/MKA)

A class of closely related hierarchical grouping methods are discussed and a procedure which implements them in an integrated fashion is presented. These methods avoid some theoretical anomalies inherent in clustering and provide a framework for viewing partitioning and nonpartitioning grouping. Significant relationships between these methods and…

The purpose of this paper is to suggest a technique of cluster analysis which is similar in aim to the Interactive Intercolumnar Correlation Analysis (IICA), though different in detail. Two methods are proposed for extracting a single bipolar factor (a "contrast compenent") directly from the initial similarities matrix. The advantages of this…