This manuscript is an unpublished dissertation.  The abstract is below.  It is not available for purchase via ProQuest....the digital dissertation source the WMF HCA project uses to secure copies of such manuscripts.  If anyone can find a copy, please contact Kevin McGrew.

VISUAL IMAGERY: AN ANALYSIS OF STRUCTURE

by BLACKWOOD, LYNN CARSON, JR., Ph.D., The University of North Carolina at Chapel Hill, 1980, 171 pages; AAT 8114789

Abstract (Summary)

This study was designed to explore the structure of the process of visual imagery by means of factor analysis. Self- report measures of imagery vividness, physiologic measures of breath rate and eye movement accompanying imagery, and performance tasks requiring visual recall and image manipulation were administered to 50 female undergraduates. A self- report personality measure, a self- report measure of cognitive habits, and a vocabulary test were also included in the battery.

Exploratory factor analysis produced eight factors which accounted for approximately 73% of the common variance. These factors, presented in the order of emergence, were interpreted as follows: self- report method, visual-spatial ability, verbal ability, habitual visual thinking, physiologic method, visual memory, paired associate learning, and tendency towards objective thinking without visual imagery. It is significant that visual and verbal factors emerged as separate dimensions and that more than a general visual ability factor was required to account for the variance among the imagery tasks. Also, the contribution of self- report measures to this analysis was negligible most likely due to the perceived "social desirability" of experiencing vivid imagery in the context of this study. Another interesting aspect of this analysis was the emergence of two polar opposite personality or cognitive style factors. The tendency towards objective or verbal thought and intellectualization without visual imagery was contrasted by the self-described creative thinker with visual thinking habits.

Four hypothetical factor models were formulated prior to data gathering. These models were subsequently tested for "goodness of fit" to the unrestrained factor analysis by the technique of Confirmatory Maximum Likelihood Factor Analysis. Model I postulated the emergence of two orthogonal factors: general visual ability and verbal ability. Model II predicted visual memory, visual constructive ability, and verbal ability as three orthogonal factors sufficient to explain the common variance. Model III proposed three orthogonal "method" factors: self-report, physiologic, and performance. Model IV called for six orthogonal factors which included visual memory, visual constructive ability, visual spatial ability, visual stability, perceptual speed, and verbal ability. Formulation of each hypothetical model involved the restriction of factor loadings, factor correlations and unique variances of the tests to represent the factor structure proposed by the model based on the experimenter's understanding of the measures. The "free" or unspecified parameters of the model were then estimated by the maximum likelihood method. A large sample Chi square statistic was used to test the degree of congruence between the hypothesized and unrestrained factor matrices. None of the four models as formulated showed significant fit to the data. In the case of Model I, and most likely II and III, poor fit was due to the specification of an insufficient number of factors, based on the results obtained when hypothetical models restricting only the number of factors were tested for fit to the data. However, in the case of Model IV, it was clear that the poor fit obtained was due to overly and inaccurately restricted parameters. These results set the stage for future research efforts which may refine Model IV to a point of good fit and then test that model's fit to new data.