A study of the effects on responses to a word association task of varying the modality of the input of the stimulus word was used as a means of doing step-wise analysis of word association data. Subjects were 70 males and 70 females randomly selected from adult basic education classes in the Atlanta, Georgia, area and assigned to one of the following stimulus presentation groups: (1) oral, (2) visual or pictorial, (3) graphic or printed, (4) oral-visual combination, (5) oral-graphic combination, (6) visual-graphic combination, (7) oral-visual-graphic combination. A list of 20 words was used for presentation in each of the treatment combinations. The seven variables were intercorrelated by the product-moment method and were factor analyzed by the principal component method. A single large factor emerged, accounting for 58 percent of the variance. A screen test suggested the existence of other minor factors. The results of three different methods employed to remove the effects of the first factor were compared. The oral and graphic treatments seemed to identify a factor. Other factors were not completely identified. It was noted that the quality of the effect of the modality on the response requires further research. Tables and references are included. (WB)