The important elements of this paradigm are (a) the administration of a pretraining test of those skills to be measured on both the test of training effectiveness and the test of transfer; (b) the creation of equivalent experimental and control groups on the basis of these pretraining scores; (c) the presentation of the training program (auding) to the experimental group during a period when the control group experiences some activity non-related to either the training given to the experimental group or the measure of transfer; (d) the administration of a post-training measure of A to both groups to determine the effectiveness of the training program; and finally, (e) the administration of the post-training measure of transfer (reading to both groups).

As pointed out by Cooper (1966), much of the transfer research is contaminated by the omission of a post-training auding measure. Because of this omission, an important element in the transfer paradigm is bypassed. Transfer could be expected between A and B only if this training actually produced an effect on A. Without a post-training measure of A it becomes extremely difficult to interpret the measure of B. Is lack of change in B due to a failure in transfer or a failure of original training to affect A? Conversely, a change in B is opened to multiple explanation without evidence that training affected A.

With these conditions in mind, we turn now to the literature bearing on the transfer problem. As mentioned earlier, a number of college and university libraries have a policy of nondissemination of theses and so some studies could not be obtained for examination. However, 23 solicited reports were eventually received and reviewed as primary sources. In addition to the 23 studies reviewed in their entirety as primary sources, Dissertation Abstracts and Duker's (1968) abstracts provided limited information on eight additional tests of the transfer hypothesis. Thus, information is included in Table 6 for a total of 31 studies bearing on the transfer hypothesis.

Eleven studies (Withrow, 1950; McPherson, 1951; Lewis, 1951; Marsden, 1951; Kelty, 1953; Lubershane, 1962; McCormack, 1962; Taylor, 1964; Harris, 1965; Wygand, 1966; Reddin, 1968) fell subject to Cooper's (1966) criticism given above; that is, they failed to report a post-training auding measure. Five of these studies actually reported a significant effect of training on reading, but due to their omission of post-training auding measures, all 11 studies were dropped from further consideration as tests of the transfer hypothesis.

The study by Childers (1963) concluded that a group of subjects given auding training demonstrated greater gain in both auding and reading than did a group of controls. Because of the omission of Ii formal analysis of the data, however, we excluded this study from serving as a measure of the strength of the transfer hypothesis.