This paper explores the effect of using multitask learning for abstractive summarization in the context of small training corpora. In particular, we incorporate four different tasks (extractive summarization, language modeling, concept detection, and paraphrase detection) both individually and in combination, with the goal of enhancing the target…

An extensive review of the research literature and a study comparing over 36,000 survey responses with archival true scores indicated that one should expect a minimum of at least three percent random error for the least ambiguous of self-report measures. The Gulliver Effect occurs when a small proportion of error in a sizable subpopulation exerts…

There are two distinct but related purposes for carrying out a "discriminant analysis": (1) discrimination, and (2) classification. The primary objective of this paper was to review the outputs of selected computer programs often used to carry out a "discriminant analysis" with respect to these two purposes. Information provided by the programs on…

This paper shares the findings of an inquiry that evaluated 50 survey articles published in a refereed journal, "Educational Administration Quarterly," by examining the 53 sample designs reported in the articles. The paper presents a typology of the sample designs identified, discusses the problems of sample selection and estimation…

This paper describes a study, now in progress, of new methods for representing the sampling variability of Mantel-Haenszel differential item functioning (DIF) results, based on the system for categorizing the severity of DIF that is now in place at the Educational Testing Service. The methods, which involve a Bayesian elaboration of procedures…

Studies evaluating hypotheses about sources of differential item functioning (DIF) are classified into two categories: observational studies evaluating operational items and randomized DIF studies evaluating specially constructed items. For observational studies, advice is given for item classification, sample selection, the matching criterion,…

A linear classification rule (used with equal covariance matrices) was contrasted with a quadratic rule (used with unequal covariance matrices) for accuracy of internal and external classification. The comparisons were made for seven situations which resulted from combining three data conditions (equal and unequal covariance matrices, minimal and…

Discriminant analysis is reviewed in terms of: (1) formulations, (2) interpretations, (3) uses, (4) issues and problems in applications, (5) recent developments and conceptualizations, and (6) general references and computer programs. Four aspects of a discriminant analysis are considered. They are: (1) separation: determining intergroup…

Three new indicators of psychometric quality for objectives-based statewide assessments are proposed. These measures provide indication of the stability of reported data on item and objectives mastery, the validity of assessment items for members of various cultural groups, and the convergent validity of prescribed objectives mastery scores. The…

Political and legislative pressures have posed a number of measurement issues and challenges to the development of sound, valid voluntary national tests (VNTs). This paper focuses on what appear to be the most difficult technical issues related to the VNT proposed by President Clinton in 1997. Technical issues refer to psychometric issues, as…

A review of the literature on a variety of issues related to testing vocabulary knowledge in a second language addresses these topics: problems in estimating vocabulary size, including the related questions of what constitutes a word, how a sample should be selected, and what are the criteria for knowing a word; sampling the basic and specialized…