MAGIC WORD Pg. 9
leaders emerged, based on the volume and type of words spoken by players. Thus, if a single player’s
volume was proportionally high, and if that player also drove negotiations, that player was identified as a
leader. Coalitions of leaders were subgroups of two or three players who dominated the discussion and
allied to elicit cooperation from the rest of the group. Each group evinced the emergence of some kind
of leadership, either singular or coalitional. Dissent was a dichotomous measure indicating whether one
player consistently (for more than two rounds) objected to the flow of group discussion. There were no
instances where more than one dissenter appeared. Finally, the willingness of players to keep talking
was assessed, in the repeated communication condition did players keep talking up through round 20 or
did they stop talking before round 16?
RESULTS
As a preliminary to the substantive analysis, Table 1 presents the coders observations concern-
ing the emergent aspects of group discussion for the ten repeating communication cases that are part of
the most numerous condition. As remarked, players in each of these cases reached some kind of
agreement. Concerning leadership, four of the cases featured a singular leader and six had some kind of
coalition. In seven out of ten cases, communication was dissent free. Finally, four cases evinced early
closure, in the rest players kept talking.
Place Table 1 about here.
Table 2 presents estimates of the effect of the communication manipulation using the defection
measure described above. The pattern of effects replicates those found by Ostrom et al. (1994). The
presence of communication in rounds reduces the number of observed defections by almost 37, on
average. The increased effect of repeating communication is not statistically significant, while adding a
cost to communication reduces the average expected positive effect of communicating by a statistically
significant nine defections. This specification confirms earlier findings and demonstrates the utility of our