This article evaluates PISA 2000 as a resource for providing information on educational attainment and inequality, focusing on France and England. Going beyond pupils' performance, it assesses levels of educational inequality by examining distributions of scores and the relationship between socio-economic status and performance. This review raises…

Correction for guessing does not fulfill its intended function when test takers who have nothing to gain from scoring will respond randomly when they could have answered correctly had they tried. Raw scores underestimate abilities. If random guessing is more prevalent in the control group, correction for guessing inflates treatment effects.…

There are difficulties in expressing effect sizes on a common metric when some studies use transformed scales to express group differences, or use factorial designs or covariance adjustments to obtain a reduced error term. A common metric on which effect sizes may be standardized is described. (Author/RL)

Supports arguments against general use of change scores and recommends the Lord/McNemar estimates of true change. Provides a numerical example illustrating the reliability problem and the problem of the prediction of true change from various linear composites of initial and final measures. (Author)

Presents both logical and empirical evidence to illustrate that the conventional scoring algorithm for ranking tasks significantly underestimates the initial level of group ability and that Slevin's alternative scoring algorithm significantly overestimates the initial level of ability. Presents a modification of Slevin's algorithm which authors…