An analysis of repeated measures designs is proposed that uses an empirical Bayes estimator of the covariance matrix. The proposed analysis behaves like a univariate analysis when sample size is small or sphericity nearly satisfied, but behaves like multivariate analysis when sample size is large or sphericity is strongly violated. (SLD)

H. Levene (1960) proposed a heuristic test for heteroscedasticity in the case of a balanced two-way layout, based on analysis of variance of absolute residuals. Conditions under which design imbalance affects the test's characteristics are identified, and a simple correction involving leverage is proposed. (SLD)

Summarizes the existing practices of Internet research and suggests extensions to them (e.g., consideration of new capabilities, such as adaptive questions and higher levels of flexibility and control) based on a large-scale, national Web survey. Lessons learned include the use of a modular design, management of Web traffic, and the higher level…

Introduces an new approach for measuring the impact of whole school reforms. The approach, based on "short" interrupted time-series analysis, is explained, its statistical procedures are outlined, and how it was used in the evaluation of a major whole-school reform, Accelerated Schools is described (H. Bloom and others, 2001). (SLD)

Presents an overview of the most common types of single-subject research designs that can be used by school counselors, with examples of the application of each design. Suggests that using single-subject research design methods such as these will enable school counselors to validate the role they play in the lives of students with sound empirical…

Summarizes themes of articles in this special issue: developmental research must be longitudinal and would benefit from interdisciplinary collaboration; career development research should incorporate both personal and contextual variables; and designs and analyses that provide normative information but obscure interactions among these variables…

This paper demonstrates that pretest-posttest designs provide more information than is commonly assumed. What is typically overlooked is that even when the posttest is the outcome measure, such designs also reveal if different interventions or experiences account for variations in pretest-posttest gains. (Contains 12 references.) (GCP)

Points out major problems with the scientific method as a model for learning about methodology in science and suggests teaching about the scientists' toolbox to remedy problems with the conventional scientific method. (KHR)

Provides guidelines for conducting representative case research (i.e., research that examines response of specific individuals to events, situations, circumstances, or conditions important in human life). Describes phases of investigation. Concludes representative case research offers a way of knowing persons holistically through application of…

Examines two metaphors for reading research approaches: the "vertical grave metaphor" and the "architectural design metaphor." Claims that while "vertical grave" researchers only add to the body of knowledge by publishing articles, "architectural" researchers actually deal with real world problems. Notes…

Linear equating methods for the common-item non-equivalent populations design were compared when true-score correlation between the test and anchor was less than unity. Scores from two groups of approximately 300 examinees illustrated three methods: (1) the Tucker equating method; (2) the Angoff-Levine method; and (3) the Congeneric-Levine method.…

Advocates the traditional controlled message, crossed design for the study of language variables in communication research. Analyzes and rejects arguments (presented in the article, "Messages as Replications: Toward a Message-Centered Design Strategy," in the same issue) in support of the multiple-message design. (SR)

Advocates the multiple-message design (message replication within experiments) for message research in communication. Analyzes and rejects arguments (presented in the article, "The Design and Analysis of Language Experiments in Communication," in the same issue) in support of the single-message design. (SR)

Articles on equating test forms in this issue are reviewed and discussed. The results of these papers collectively indicate that matching on the anchor test does not result in more accurate equating. Implications for research are discussed. (SLD)

Describes representative single-subject research designs and related statistical procedures for group counselors to study the effects of interventions in groups. Concludes that despite limitations, single-subject designs are useful for obtaining information about intervention effects with groups in direct service contexts that can be compared to…