This paper examines the estimation of two-stage clustered RCT designs in education research using the Neyman causal inference framework that underlies experiments. The key distinction between the considered causal models is whether potential treatment and control group outcomes are considered to be fixed for the study population (the…

The authors describe and illustrate 6 factors that affect the size of a Pearson correlation: (a) the amount of variability in the data, (b) differences in the shapes of the 2 distributions, (c) lack of linearity, (d) the presence of 1 or more "outliers," (e) characteristics of the sample, and (f) measurement error. Also discussed are ways to…

When errors of measurement are positively correlated, coefficient alpha may overestimate the "true" reliability of a composite. To reduce this inflation bias, Komaroff (1997) has proposed an adjusted alpha coefficient, ak. This article shows that ak is only guaranteed to be a lower bound to reliability if the latter does not include correlated…

Discretized multivariate normal structural models are often estimated using multistage estimation procedures. The asymptotic properties of parameter estimates, standard errors, and tests of structural restrictions on thresholds and polychoric correlations are well known. It was not clear how to assess the overall discrepancy between the…

This paper discusses the benefits of using generalizabilty theory in lieu of classical test theory. Generalizability theory subsumes and extends the precepts of classical test theory by estimating the magnitude of multiple sources of measurement error and their interactions simultaneously in a single analysis. Since classical test theory examines…

Previous methods for estimating the conditional standard error of measurement (CSEM) at specific score or ability levels are critically discussed, and a brief summary of prior empirical results is given. A new method is developed which avoids theoretical problems inherent in some prior methods, is easy to implement, and estimates not only a…

This Monte Carlo investigation provides some possible solutions to problems related to choosing multiple-comparison methods that maximize true rejections and minimize false ones. It has been argued that the traditional Bonferroni approach to multiple comparisons, which satisfies the statistician's family-wise Type I error concerns, could be…

Linear and nonlinear methods for incorporating score precision information when the score scale is established for educational tests are compared. Examples illustrate the methods, which discourage overinterpretation of small score differences and enhance score interpretability by equalizing error variance along the score scale. Measurement error…

Results of correlation studies of 141 adult epileptics' scores on the Background Interference Procedure (BIP) indicated that the BIP often does not agree with abnormal neurological diagnoses but often does agree with psychiatric diagnoses of Organic Brain Syndrome (OBS). Suggests that future BIP validity studies include a behavioral measure of OBS…

The literature on measurement reliability shows the consensus that group heterogeneity with regard to the trait being measured is a factor that affects the sample measurement reliability, but the degree of such effect is not entirely clear. Sample performance also has the potential to affect measurement reliability because of its effect on the…

Where the accuracy of a measurement is important, whether for scientific or practical purposes, the investigator should evaluate how much random error affects the measurement. New research may not be necessary when a procedure has been studied enough to establish how much error it involves. But, with new measures, or measures being transferred…

Several robust absolute deviation statistics have been developed recently. Two such correlation coefficients are developed and discussed, one for ranked data and another for interval level data. The standard error and range of the coefficients are given. The algebraic relationship between the coefficients and three widely used correlation…

Simulated data based on five test score distributions indicate that a slight modification of the asymptotic normal theory for the estimation of the p and kappa indices in mastery testing will provide results which are in close agreement with those based on small samples from the beta-binomial distribution. (Author/BW)