When three’s a crowd: how relational structure and social history shape organizational codes in triads

Journal of Organization Design

Table 3 t-tests of correlation between communication effectiveness (speed and accuracy) in the last round of game played with a target image and the kind of code that was elicited at the end of the experiment for that target image

Test performed on set =>	All codes (N = 1515)			Some dyadic or group level code exists (N = 1389 for time, N = 1405 for accuracy)			At least one group level code exists (N = 846 for time, N = 853 for accuracy)			There is only one group code (N = 641 for time, N = 647 for accuracy)
	Completely divergent labels	Some dyadic or group level code exists	Difference	Dyadic codes only	At least one group level code exists	Difference	There are multiple codes	There is only one group code	Difference	The label is longer than one word	The label is a single word	Difference
mean time to complete last round in seconds (st.dev in parentheses)	13.85 (8.61)	11.37 (10.44)	2.48**	13.2 (11.61)	10.2 (9.44)	3**	11.81 (8.35)	9.68 (9.71)	2.14**	9.78 (9.86)	6.73 (1.24)	3.06†
average number of mistakes made in last round (st. dev in parentheses)	0.15 (0.40)	0.08 (0.40)	0.06†	0.12 (0.55)	0.06 (0.27)	0.07**	0.07 (0.30)	0.05 (0.26)	0.02	0.05 (0.26)	0.05 (0.21)	0.01