Factor analysis by generalized least squares

Aitken's generalized least squares (GLS) principle, with the inverse of the observed variance-covariance matrix as a weight matrix, is applied to estimate the factor analysis model in the exploratory (unrestricted) case. It is shown that the GLS estimates are seale free and asymptotically efficient. The estimates are computed by a rapidly converging Newton-Raphson procedure. A new technique is used to deal with Heywood cases effectively.The work on this project was done when the first author was Research Statistician at Educational Testing Service, Princeton, N. J. The second author was in part supported by a grant from the Research Committee of the University of Wisconsin Graduate School. The authors wish to thank Michael Browne for many helpful comments and Marielle van Thillo for valuable assistance in the numerical computations.     

The effect of memory load on reaction time in character classification

In order to investigate the effect of memory load on reaction time (RT), choice RT trials were embedded in a binary character classification task using the varied set procedure. Twelve Ss performed in experimental blocks, as well as in control blocks consisting of character classification trials only. In experimental blocks, every trial began as a classification trial with the presentation of a new positive set. On a random half of these trials, however, a choice RT stimulus was presented instead of a probe letter and S simply made the indicated response. Results indicated that memory load had no effect on the choice RT trials. Embedding choice RT trials in the classification task affected the intercept (but not the slope) of the function relating classification RT to memory load. This result implies that the increase in latency usually obtained in classification experiments is entirely due to an increase in the duration of the memory searching stage of processing.     

An analysis of individual differences in similarity judgments about complex random forms

A variation of the method of graded dichotomies was used to obtain judgments of interstimulus similarity for all pairings of 21 12-turn random shapes. Using the Tucker-Messick procedure, a six-dimensional spatial model, representing individual variation in performance on the task, was derived. Six “idealized individuals” (IIs) were isolated. A five-dimensional Euclidean spatial model was obtained for each II using the Shepard-Kruskal scaling technique. The resulting psychological dimensions of perceived form were related to two sets of physical measures on the forms: the factor space suggested by Brown and Owen and a set of more basic physical measures. These basic measures allowed interpretation of most of the psychological dimensions and revealed differences between IIs in patterns of cue utilization.     

A yoked-chamber comparison of concurrent and multiple schedules

Pigeons were exposed to alternative pairs of variable-interval schedules correlated with red and green lights on one key (the food key). In one experimental chamber, responses on a white key (the changeover key) changed the color of the food key and initiated a 2-sec changeover delay. Pigeons in a second chamber obtained food by pecking on a colored key whenever the pigeons in the first (concurrent) chamber had obtained food for a peck on that key color. There was no changeover key in the second (multiple) chamber: changeover responses in the first chamber alternated the schedules and colors in both chambers. The pigeons in both chambers emitted the same proportion of responses on each of the variable-interval schedules, and mastered discrimination reversals at the same rate. The pigeons differed only in their absolute response rates, which were greater under the concurrent schedules. In a second experiment, changes in key color occurred automatically, with different proportions of time allocated to the two variable-interval schedules. Matching of relative response frequency to relative reinforcement frequency was affected by the relative amounts of time in each component, by rate of changeovers, and by manipulations of the variable-interval scheduling.     

Sustained behavioral contrast in children

Children were exposed to a multiple schedule involving equal variable-interval schedules in each of two components and a multiple schedule involving a variable-interval schedule in one component and an extinction schedule in the other. Response rates were equal in both components when each involved a variable-interval schedule. Response rates differed in the two components of the multiple variable-interval extinction schedule. Response rates were higher in the variable-interval schedule when the accompanying schedule was extinction than when it was variable interval. The increase in response rate in the variable-interval component, simultaneous with the decrease in response rate in the extinction component, illustrated sustained behavioral contrast, and was the first evidence of this phenomenon in children.