We examine the challenges of teaching probability through the use of modelling. We argue how an integrated modelling approach might facilitate a coordinated understanding of distribution by marrying theoretical and data-oriented perspectives and present probability as more connected to the social lives of modern-day students. Research is, however,…

It is common in regression discontinuity analysis to control for high order (third, fourth, or higher) polynomials of the forcing variable. We argue that estimators for causal effects based on such methods can be misleading, and we recommend researchers do not use them, and instead use estimators based on local linear or quadratic polynomials or…

This paper makes the point that descriptive and inferential measures of nonnormality and graphic displays of the frequency distribution of important variables should be routine in research reporting. This point is particularly true for research involving measures with nonarbitrary metrics where the distribution shape is unaffected by measurement…

Methods by which structural models correct for the effects of attenuation due to measurement error are reviewed, and implications of such disattenuation for interpreting the results of structural equation models are considered. Recommendations are made for improving the practice of disattenuation, and caution is urged in drawing inferences based…

A Bayesian alternative to interpretations based on classical reliability theory is presented. Procedures are detailed for calculation of a posterior score and credible interval with joint consideration of item sample and occasion error. (Author/SLD)

L-scaling is introduced as a technique for determining the weights in weighted averages or scaled scores for T joint observations on K variables. The technique is so named because of its formal resemblance to the Leontief matrix of mathematical economics. L-scaling is compared to several widely-used procedures for data reduction, and the…

A brief, practical overview of "design effects" (DEFFs) is presented for users of the results of sample surveys. The overview is intended to help such users to determine how and when to use DEFFs and to compute them correctly. DEFFs are needed only for inferential statistics, not for descriptive statistics. When the selections for…

A 2-facet measurement model is identified, and its coefficient of generalizability (CG) is examined. Three other multifaceted measurement models and their CGs are identified. An empirical investigation of all four procedures is conducted using data from a study of the psychopathology of 71 prison inmates. (SLD)

In clinical trials, adaptive allocation means that the therapies assigned to the next patient or patients depend on the results obtained thus far in the trial. Although many adaptive allocation procedures have been proposed for clinical trials, few have actually used adaptive assignment, largely because classical frequentist measures of inference…

Bayesian methods for estimating dose response curves in quantal bioassay are studied. A linearized multi-stage model is assumed for the shape of the curves. A Gibbs sampling approach with data augmentation is employed to compute the Bayes estimates. In addition, estimation of the "relative additional risk" and the "risk specific…

A Bayesian approach to sensitivity of inferences to possible outliers involves recalculating marginal posterior distributions of parameters of interest under assumptions of heavy tails. This strategy is implemented in the hierarchical model setting through Gibbs sampling, a Monte Carlo technique, and illustrated through a reanalysis of data on…

Four measurement designs are presented for use with correlation coefficients corrected, in one variable, for attenuation due to unreliability (partially disattenuated). Associated asymptotic variance/covariance expressions are presented. Empirical simulation results illustrate the satisfactory Type I error control and statistical power of the…

P. O. Johnson and J. Neyman (1936) proposed a general linear hypothesis testing procedure for testing the null hypothesis of no treatment difference in the presence of some covariates. This is generally known as the Johnson-Neyman (JN) technique. The need for the hypothesis testing step (often omitted) as originally presented and the…

Procedures based on latent trait models and the Rasch model are described for computation and asymptotic statistical inference for two decision consistency indexes often used in mastery or competence testing. Simulations were conducted for typical mastery testing situations to illustrate the procedures. (SLD)

The use of sampling weights in deriving population estimates for the National Assessment of Educational Progress (NAEP) and the effects of nonresponse and undercoverage on those estimates are described. The estimation of sampling variability from complex sample surveys is also reviewed, concentrating on the jackknife repeated replication…