Using computer simulation, parametric analysis of covariance (ANCOVA) was compared to ANCOVA with data transformed using ranks, in terms of proportion of Type I errors and statistical power. Results indicated that parametric ANCOVA was robust to violations of either normality or homoscedasticity, but practiced significant power differences favored…

Randomized response measurement techniques have been proposed to overcome subject unwillingness to answer embarrassing or threatening questions truthfully. Many of the applications to date have dealt with health-related issues. This article demonstrates the application of the unrelated question randomized response design with one such question.…

Analyses for stress patterns of the first 1,000 nouns and 500 verbs from a high-frequency Russian vocabulary list. The most irregular patterns are seen in the most frequently occurring items and, therefore, should be learned early in the study of Russian. (Author/LMO)

Two methods of constucting equal-interval scales for educational achievement are discussed: Thurstone's absolute scaling method and Item Response Theory. Alternative criteria for choosing a scale are contrasted. It is argued that clearer criteria are needed for judging the appropriateness and usefulness of alternative scaling procedures.…

Frequency-of-circulation distributions from 16 United Kingdom libraries participating in initial period of operation of the Public Lending Right are discussed. Differences between the libraries are highlighted, and, for each, the adequacy of various refinements of the simplest model (geometric distribution, negative binomial distribution, or mixed…

Describes the Archive Retrieval and Analysis System (ARRAS), which provides rapid access to textual data and fast, flexible, open-ended analysis of individual texts; explains why the system should be viewed as a new kind of working environment; and suggests that such systems may influence users' intellectual point of view. (MBR)

The use of Bayesian theory to connect ordinal data and ordered scale points with the theory of order statistics is presented. Exact and approximate multivariate and marginal densities for the scale points are derived. (Author/JKS)

Review of recent developments in statistical theories for social sciences highlights Haitun's statistical distributions, Laplace's "Law of Succession" and distribution, Laplace and Bradford analysis of book-index data, inefficiency of frequency distribution analysis, Laws of Bradford and Zipf, natural categorization, and Bradford Law and…

Fifth-grade students (n=154) were given the Group Embedded Figures Test. Distribution characteristics and sex differences were examined. Earlier norms developed for adult samples are substantially different from those of a sample of children. The norms presented here provide a guide for identification of field dependence in children. (Author)

This study investigated the validity of Hayes's formula for converting rank-ordered Rokeach Values data to a normal distribution theoretically achieved through the use of rated data. Findings support the validity of Hayes's formula and expand the analytic possibilities of data collected on the Rokeach instrument. (Author/PN)

The interpretation of significant left-tailed analysis of variance (ANOVA) F-ratios is supported by considering the case of a fixed effects ANOVA model. The conclusions of this case are generalizable to other standard ANOVA models. (Author/PN)

Develops an empirical model for predicting the size of an information store necessary to satisfy specified levels of demand. A method for selecting items for retirement with the least adverse impact on the store is suggested. Three tables, three figures, and a 10-item reference list are included. (Author/JL)

Demonstrates, with an example, a relatively simple method for fitting Bradford's law to empirical data to estimate the number of journals and articles in a subject collection. An exact discrete formulation illustrates Bradford's law as a special case of the Zipf-Mandlebrot "rank frequency" law. (Author/RAA)

Proposes an error gravity/distribution factor to give a mathematically consistent evaluation method to error analysis. Such an approach assumes that "seriousness" of error is related to frequency of output and not to notions of degree of communicative difficulty. Distribution of error type is also taken into account. (PJM)

Data from national samples from the National Education Longitudinal Study of 1988 were used to study gender differences in mathematics achievement. Gender differences were not found when total group means were compared, but noteworthy differences favoring males were found at the high end of the score distributions. (SLD)