Stepwise regression is not an adequate technique to provide the best set of variables with which to predict the dependent variable. By using the stepwise regression method, one who attempts to select the best set of predictors of a given dependent variable will face more problems than he or she attempted to resolve. This is illustrated with an…

This study examined the robustness of the estimation of fixed-effects in multilevel analysis, as might occur in the conduct of school-effects studies with outlying schools. Outlying values for both intercepts and slopes for individual schools were modeled separately to determine the effects of extreme values of second-level variables on the…

Norm-referenced tests yield information regarding a student's performance in comparison to a norm or average of performance by similar students. Norms are statistics that describe the test performance of a well-defined population. The process of constructing norms, called norming, is explored briefly in this paper. Some of the most widely reported…

The Job Satisfaction Survey (JSS) (P. Spector, 1985 and 1992) is a 36-item survey instrument designed to measure 9 aspects of job satisfaction, including: (1) pay; (2) promotion; (3) supervision; (4) benefits; (5) contingent rewards; (6) operating procedures; (7) co-workers; (8) nature of work; and (9) communication. In addition to measuring the…

Item response theory (IRT) model parameter estimates have considerable merit and open up new directions for test development, but misleading results are often obtained because of errors in the item parameter estimates. The problem of the effects of item parameter estimation errors on the test development process is discussed, and the seriousness…

Statistics used to estimate the population correlation ratio were reviewed and evaluated. The sampling distributions of Kelly's and Hays' statistics were studied empirically by computer simulation within the context of a three level one-way fixed effects analysis of variance design. (Author/RC)

Some basic statistical concepts relevant to the teacher--mean scores, standard deviation, normal and skewed distributions, z scores, item analysis, standard error of measurement, reliability--and their use by the teacher are explained. (RM)

The nature of the criterion (dependent) variable may play a useful role in structuring a list of classification/prediction problems. Such criteria are continuous in nature, binary dichotomous, or multichotomous. In this paper, discussion is limited to the continuous normally distributed criterion scenarios. For both cases, it is assumed that the…

Studies have revealed significant problems in correctly scoring ambiguous verbal responses to test items on the Wechsler Intelligence Scale for Children-Revised (WISC-R). This study evaluated the effectiveness of an instructional design procedure developed to reduce examiner scoring errors on the WISC-R. Data concerning frequent sources of error…

The estimation of mean and standard errors of the eigenvalues and category quantifications in generalized non-linear canonical correlation analysis (OVERALS) is discussed. Starting points are the delta method equations. The jackknife and bootstrap methods are compared for providing finite difference approximations to the derivatives. Examining the…

It has often been argued that all techniques of standard setting are arbitrary and likely to yield different results for different techniques or persons. This paper deals with a related but hitherto ignored aspect of standard setting, namely, the possibility that Angoff or Nedelsky judges misspecify the probabilities of the borderline student's…

Canonical correlation analysis is a powerful statistical method subsuming other parametric significance tests as special cases, and which can often best honor the complex reality to which most researchers wish to generalize. However, it has been suggested that the canonical correlation coefficient is positively biased. A Monte Carlo study…

Determination of statistical power for analysis of variance procedures requires five elements: (1) significance level; (2) effect size; (3) number of means; (4) error variance; and (5) sample size. Significance levels are traditionally chosen to be 0.5, .01, or .001. Effect size is not discussed in this paper. The number of means is determined by…

An assumption underlying multiple linear regression is that the standard error of measurement is equal for all observations. The literature has not addressed the procedures to be used when this assumption is violated. It was clear that data analysis to be performed on districts in California would severely violate this assumption, since district…

Assuming that the density p of the true ability theta in the binomial test score model is continuous in the closed interval (0, 1), a Bernstein polynomial can be used to uniformly approximate p. Then via quadratic programming techniques, least-square estimates may be obtained for the coefficients defining the polynomial. The approximation, in turn…