The purpose of this study was to investigate the relative effectiveness of three categories of variables--sociodemographic, personality, and social support system--in predicting persistence in Adult Basic Education. A particular emphasis was focused upon social support variables and the extent of their capacity to improve the level of prediction success which could be achieved by the other two categories of variables. This study was also concerned with identifying that combination of variables from all three categories which enabled the best possible prediction of persistence. The sample population for this study included 163 adult students enrolled in GED preparatory classes in three urban Virginia school districts. These students all attended classes two nights a week for a total of 5-6 hours. Sociodemographic data pertaining to age, race, sex, employment status, and time since last school attendance, were collected by means of a short questionnaire. Personality measures were acquired relating to self-confidence, achievement need, and affiliation need through use of the appropriate subscales of the Adjective Check List developed by Gough and Heilbrun (1952). Selections of particular sociodemographic and personality variables were based on their frequent inclusion in past studies of Adult Basic Education persistence and participation. This assured meaningful comparisons of the prediction effectiveness of these variables with that of the social support variables emphasized in the present study. Social support data were obtained pertaining to family, work, and church environments through use of selected subscales of the Social Climate Scales designed by Moos (1974). Composite measures were computed, consisting of the average of subscale scores pertaining to each environment area. A series of multiple linear regressions were performed to determine the order of best prediction among various combinations of variables and variable categories. The composite social support (environment average) variables proved poor predictors, a fact which raised doubts about the meaningfulness of the averaging procedure used to derive them. Selected individual subscales of the Social Climate Scales, by contrast, were better predictors of persistence than any of the sociodemographic and personality variables employed in the study. This was particularly true of the group environment subscales used to measure church leader support and member expressiveness. Despite the relative prediction effectiveness of some social support measures as compared to other variables, it should be noted that none of the predictor variables or variable categories employed in this study explained a substantial portion of persistence variance. The results of the study are thus of questionable practical value for Adult Basic Education teachers and administrators. They do, however, suggest some directions for future research. It is recommended that researchers continue to investigate social support systems of Adult Basic Education students. Social support measures are, on the whole, probably as effective predictors of ABE persistence as sociodemographic and personality measures, two categories of variables more frequently focused upon in previous studies. At the same time, many environmental factors can be addressed more directly and effectively by ABE practitioners than relatively fixed sociodemographic and personality characteristics. / Ed. D.