When normality does not hold, nonparametric tests represent an important data-analytic alternative to parametric tests. However, the use of nonparametric tests in educational research has been limited by the absence of easily performed tests for complex experimental designs and analyses, such as factorial designs and multiple regression analyses,…

The Solomon four-group design (R. Solomon, 1949) is a very useful experimental design to investigate the main effect of a pretest and the interaction of pretest and treatment. Although the design was proposed half a century ago, no proper data analysis techniques have been available. This paper describes how data from the Solomon four-group design…

It is generally believed that the correction formula will yield exact correlational values only when the regression of z on x is both linear and homoscedastic. The formula is shown to hold for nonlinear heteroscedastic relationships. A simple sufficient condition for formula validity and estimation predictions is demonstrated in a numerical…

Food consumption survey data were analyzed to test the stability of household size adjustment factors used for estimating the household costs of U.S. Department of Agriculture food plans. Reported household scale factors varied slightly from former estimates, and were highly dependent on the diet quality measures incorporated in the analysis.…

This study comparing four major data aspects--expert judgment as to research performance, bibliometric variables, research output, and organizational size--in 19 pharmaceutical companies reveals a consistent tendency for bibliometric indicators to perform better than expert judgments in predicting research performance. Drug output regression…

An evaluation of open education was shown to produce misleading results due to probable regression phenomena. These questionable results are now spread throughout the literature of education, sociology, and psychology. Researchers are advised to review, not merely summarize, prior articles. (Author/PN)

If the correlation between selection test and pretest is greater than the correlation between selection test and posttest, a selected group will continue to regress to the mean in the norm-referenced model of the Title I Evaluation and Reporting System and, therefore, yield a biased estimate. (Author/RL)

Discusses the use of the least-squares regression line in determining the optimum straight line in an efficient and objective manner and the validity of its underlying assumptions in certain situations. Discusses the treatment of kinetic data, distortion from large errors, and the Guggenheim method. (JRH)

The accuracy of three analytical formulas for shrinkage estimation and four empirical techniques were investigated in a Monte Carlo study of the coefficient of cross-validity in multiple regression. Substantial statistical bias was evident for all techniques except the formula of M. W. Brown (1975) and multicross-validation. (SLD)

Sought to determine if the size of an entering freshman class at a 4-year university could be accurately predicted using linear regression, a standard statistical procedure. Analyzed certain independent variables believed to be related to the number of actual enrolled freshmen. Found that a reasonably good forecast for the fall could be obtained…

Estimates frontier production functions for South Carolina's educational process, using data from 541 classes. Classes taught by teachers who received merit awards show greater mathematics and reading achievement gain scores, as do classes with fewer free-lunch students. There was a positive relationship between achievement and (larger) class…

Trained 10 undergraduate psychology students to have the knowledge needed to solve 5 multiple choice problems on descriptive regression analysis and asked them to think aloud while attempting to solve the problems. Analysis of responses shows that failure to select relevant information in the text, failure to retrieve relevant propositional…

Examined risk and protection for adolescent problem behavior and depressive symptomatology in an average-risk sample of 243 11th graders. Regression analyses using aggregated or summary measures of risk for several contexts showed that multiple contexts of risk contributed independently to both misconduct and depressed moods, and that protective…

Discusses a study of instructional cost determination at small independent colleges. Such colleges could operate more efficiently by accelerating recruitment of part-time students, increasing use of part-time instructors, recognizing the direct relationship between available revenue and instructional costs, raising student-faculty ratios where…

One source of bias, regression artifacts, in program evaluations that use regression discontinuity design is identified, and its importance is discussed. A method is suggested for its correction. Although use of this method does not solve the problem of selection bias, it does improve the reliability of nonexperimental methods. (SLD)