TWENTY-SIX GROUPS OF FIVE SUBJECTS EACH WERE STUDIED TO DETERMINE THE EFFECTS OF FEEDBACK ON SUBSEQUENT GROUP BEHAVIOR. EACH GROUP WAS GIVEN TEN PROBLEMS, EACH TO BE SOLVED INDIVIDUALLY, BEFORE AND AFTER A DISCUSSION PERIOD. THE USE OF AN ANALOG COMPUTER PERMITTED THE EXPERIMENTER TO GIVE FEEDBACK ON THE FIRST SOLUTION BEFORE THE DISCUSSION BEGAN ON EACH PROBLEM. IN HALF OF THE GROUPS, EACH MEMBER WAS TOLD THE EXTENT OF HIS DEVIATION FROM OTHER MEMBERS AS WELL AS HOW ACCURATE HIS SOLUTION WAS. THE REMAINING GROUPS WERE GIVEN THE SAME PROBLEMS BUT NO FEEDBACK. THE CRITERION MEASURES CONSISTED OF FINAL ACCURACY, FINAL AGREEMENT BETWEEN MEMBERS, STABILITY FROM THE FIRST SOLUTION TO THE SECOND, AMOUNT OF TIME EACH MEMBER TALKED DURING DISCUSSION, ATTRACTION TO THE GROUP, AND ESTEEM OF OTHER MEMBERS. IT WAS FOUND THAT MEMBERS OF FEEDBACK GROUPS WERE MORE ACCURATE IN THEIR FINAL SOLUTION AND SLIGHTLY MORE ATTRACTED TO THE GROUP. DEVIANT MEMBERS WITH INACCURATE SOLUTIONS WERE MORE RELUCTANT TO ACCEPT THE MAJORITY OPINION UNLESS THEY HAD BEEN INFORMED OF THEIR POSITION. MORE ACCURATE GROUP MEMBERS WERE MUCH MORE LIKELY TO ATTEMPT LEADERSHIP IF THEY HAD KNOWLEDGE OF THEIR ACCURACY. THIS DOCUMENT, L.C. CARD NO. MIC 60-1464, IS AVAILABLE FROM UNIVERSITY MICROFILMS, ANN ARBOR, MICH. MICROFILM $2.50. XEROX $3.00. 54 PAGES. (AUTHOR/AJ)