Factor analysis has been proposed and used as a method of statistical analysis of several measurements made on one individual repeatedly over time. This paper discusses some difficulties in applying factor analysis to multiple time series and attempts to indicate to what extent such methods can accomplish the goals of time series analysis. Some other methods are suggested.

A subject in a two-choice situation characteristically makes several observing responses before performing the final choice. This behavior can be described by means of a random walk model. The present paper explores some possibilities as to how this model can be extended to include choice time. The assumption is made that the duration of each step in the random walk is a random variable which is exponentially distributed. With this assumption, one can predict the probability distributions of the choice times as well as the moments of these distributions.

A generalization of direct ratio scaling methods to multidimensional ratio scaling is described. This method requires an observer to report the proportion of a standard percept that is contained in a given percept and vice versa. The method was developed to meet requirements for experimentation in such areas as color vision, gustation, and olfaction.

A method is illustrated for estimating correlation coefficients and mean criterion scores in a full-range population from bivariate distributions available in a selected sample when the criterion is a dichotomy. The proposed method requires only the assumptions needed for use of formulas for correcting correlation coefficients for restriction of range when both variables are continuous and is suitable for use when restriction is due to either direct or indirect selection.

The linear discriminant function and the generalized distance function, two special cases of discriminant technique, require multivariate normality and homogeneous variance-covariance matrices, and hence utilize only mean differences among groups. The more general methods can also utilize differences in variances and/or covariances. Tables are given showing the discriminatory value of differences in means, variances, and intercorrelations, taken singly. Equations which utilize all such differences are given for the normal multivariate distribution.

The study compares the extent to which naturally structured data and artificial, relatively random data (both with the same basic parameters) produce simple structure factors which are uniquely determined. Two examples of unstructured matrices were compared with the ball problem matrix. The results show that an oblique position of maximum hyperplane count in the structured data differs from that in the unstructured by reaching a significantly more unique position in terms of the exactitude with which it is re-discoverable when starting from different positions, and by reaching (at the maximum) a significantly higher hyperplane count.

In this method of analyzing learning data, entire learning curves are described quantitatively by single numbers which are used in a statistical test to determine whether two or more groups of learning curves are significantly different. The method has some logical advantages over prevailing methods in that it avoids the use of average learning curves and of arbitrary measures of slope and asymptote. Its disadvantage is computational. Since it involves the use of factor analytic procedures, it may be tedious to apply unless computation is carried out on a high-speed computer.

Brogden's coefficient of selective efficiency and Clemans' λ are more efficient than biserial r when the correlation is high. Some empirical sampling studies are reported comparing the standard errors of these statistics.

This paper presents a way of factoring the Guttman circumplex to get factor weights whose pattern is close to that postulated by Guttman for the circumplex. An empirical example, including a crude procedure for diagonal estimation, is shown. Other sources of nonuniqueness are discussed, and the limitation to odd-numbered complexity for the tests is indicated. An orthodox factor analysis is applied to the example, and this provides a basis for comparison of the two opposing models.

It is shown how results of Prokhovnik on the number of pattern variants that may be formed by k markers on a square network of m2 positions may be derived more simply by means of a combinatorial theorem of Pólya's, which may also be used to solve systematically many other problems of this type.

The purpose of this note is to reconsider the Kelley-Cureton definition of optimal extreme groups for estimating item-criterion correlations. Optimal tail per cents are derived, using the criterion of minimum sampling variance of the tetrachoric correlation coefficient, and the findings are related to earlier work of Mosteller. It is shown that upper and lower 27 per cent groups yield the most precise estimate of the tetrachoric coefficient only when the population correlation is close to zero. When the population value is .4, extreme 20 per cent groups provide estimates with the smallest sampling error variance. It is further shown, however, that 27 per cent extremes yield highly efficient estimates. Thus no change is recommended in traditional item analysis procedures.

The Ui statistic, discussed by Creager and Valentine in a recent paper, is defined as a standard error of estimate for the regression of a composite, consisting of n components, on the n − 1 components, exclusive of the jth component. It is shown that this standard error equals the standard error of estimate for the prediction of the jth component from the other n − 1 components. This equality is used to suggest an efficient procedure to compute all primary uniquenesses Ui.