IN RESEARCH concerning the human subject, the traditional experimental approaches have been to match selected subjects on a series of variables or to assign them randomly into various groupings. It is not always possible or desirable, however, to as sign subjects. Often the biasing effect of the volun teer subject has been ignored in the classical or traditional experimental design when comparing a volunteer group with a non-volunteer group. In order to control the problem of the selection factor when working with volunteer subjects, exper imenters have at times utilized only volunteers in establishing experimental and control groups. Half of the group normally receives treatment, while the remaining group from which the applied technique is withheld serves as the control. But this experi mental design is not always possible, especially when an evaluation of an operating program is re quested. This study focuses on the problem of de veloping an appropriate control group with which to compare an experimental group composed of volun teer subjects. During the Fall of 1966, The University of Tenn essee initiated a tutoring program designed to act as a supplementary aid to corresponding par ent courses. All students who were enrolled in a parent course were eligible to enroll i n a tutoring course; furthermore, all students who participated in the sessions were volunteers. Due to the volun tary nature of the program, a modified design (as opposed to the traditional design) was considered essential for evaluation. Several researchers in the social sciences have grappled with the effect of volunteer subjects in ex perimental research. A lack of uniformity of ap proach and disagreement in results indicate that there remains room for improvement in experimen tal designs employed. The field of counseling has provided extensive research in this area, and some of it appears relevant to this discussion. Sear les (6), working with volunteer and non-vol unteer students in counseling sessions, has sug gested that research in evaluating counseling pro grams should be limited to volunteer subjects. In situations where students despite their wishes have been arbitrarily selected to be counseled, the ac crued results reflect dubious interpretative value at best (1, 9, 10). In past studies utilizing volunteer subjects, exper imenters have been aware that the selection factor represents an unknown variable influencing their re sults. Montgomery and Whitlock (5), studying the effects of pre-freshman courses on student academ ic achievement, compared an experimental group electing to take a how-to-study course with a matched set of controls who did not take the prepara tory courses. In that study, after one quarter of regular coursework the experimental group experi enced lower academic performance than did the con trol group. The authors wondered to what extent their findings were influenced by uncontrolled selec tion factors. Using selected variables in a study of tutoring at Parsons College, Etters (2) compared college-level student volunteers with students who elected not to attend the tutor ing sessions. The results re vealed that tutored students significantly improved their academic standing over non-tutored students; however, no attempt was made to measure the effect of volunteering as a selection factor in the improved performance. In attempting to measure the effect of the volun teer subject selection factor, Spielberger and Weitz (8) discovered that the act of volunteering can influ ence the relationship between anxiety and academic performance. Ivey (4) compared college student vol unteers with a matched group of students who elect ed not to attend a counseling program. The w ith drawal rate of the counseled group was significantly higher than among non-counseled students. When students attending four or more counseling sessions were compared with those attending three or fewer sessions, the withdrawal rate of the frequent attend ers was significantly higher than the group which