One study found that individuals tend to agree in their testimony following group discussion and are liable to change their original replies to bring themselves into agreement with group leaders' recall. A second experiment focused upon the actual items which witnesses agreed on following discussion. Changing the mode of questioning from interrogative to narrative eliminated differences in magnitude of agreement between discussion and no-discussion groups. As narrative recall leads to reporting of 'easy' items, it was suggested that only certain 'different' items are susceptible to discussion effects. The need for caution regarding the reliance on testimony following witness discussion was emphasized. Tables, 20 references, and 2 notes are supplied. (Author abstract modified)