Assuming chance to be fully operative, the predicted effect of chance success on test validity when answer options are supplied depends on the number of options, the difficulty of the test and the variance of test scores. Predicted validity values are compared with empirical validity values in an experiment which used the same mathematics test items with and without answer options.

An attempt is made to elaborate upon Lev's results concerning the reliability of the point biserial coefficient of correlation in a manner that will be helpful to the psychological statistician. Procedures required in the use of the non-central t tables prepared by Johnson and Welch are described as they relate to the determination of the fiducial limits for a point biserial coefficient. A normal approximation technique for the estimation of fiducial limits is also suggested. Numerical evidence is presented which shows that relative to a given level of significance the width of the fiducial interval estimated from a point biserial coefficient of any size is smaller than that of the fiducial interval corresponding to an ordinary Pearsonian coefficient of the same magnitude.

This note revises the figures in the original article concerning the accuracy of the tetrachoric estimates involved. These estimates are better than previously noted and are very satisfactory. Some minor known errors in the original article are also noted.

This paper describes the construction of a new item analysis table for the high-low-27-per-cent group method. The table provides a ready means of translating the observed proportions of success in the two extreme groups (pH, pL) into measures of item difficulty and item discrimination (p, Δ, and r). The tabled values of both the difficulty index, p, and the discrimination index, r, have been derived from Karl Pearson's tables of the normal bivariate surface.

This article reports a second-order factor analysis of the 13 interpretable first-order factors of Adkins and Lyerly based on 66 variables for 200 Army men. The approach to simple structure for 6 second-order factors was unusually good. Loadings of the original 66 variables on the second-order factors were also obtained. Five of the factors are tentatively interpreted as Precision in Formation and Use of Verbal Concepts, General Verbal Fluency, Visualizing Spatial Constancy During Movement, Speed in Analysis, and Flexibility in Analysis.

Scores obtained at eight different stages of practice on the Complex Coordination Test together with scores on 18 reference tests were subjected to a Thurstone Centroid Factor Analysis. Nine meaningful factors were identified in the experimental battery. The results indicated considerable, but systematic, changes in the factor structure of the Complex Coordination Test as practice on the task was continued. The test became less complex (factorially) as practice was continued. Moreover, there was a change in the nature of the factors contributing variance at early and later stages of practice. Implications of the findings are related to certain problems of learning theory, psychomotor test development, and criterion analysis.

The purpose of this article is to show that a relationship can be established between Osgood and Suci's D based on raw scores and scores obtained from factor analysis data without equalizing the means and variances of each individual's set of scores.

Attempts to develop improved methods of selecting a panel of judges for psychometric work are presented. The applications reported are in the sensory field and are limited to the assumption of a unidimensional ability underlying the judgments in question. Some aspects of cost efficiency are also considered.

An electronic computer with large capacity memory, and employing binary numbers, is ideally suited for large-scale item analysis. The adaptation and use of one computer, SWAC, is described.