Table 2.

Mean Within- and Between-Objects Error Correlations (With Standard Deviations in Parentheses)

Experiment	Turn size	Within-object error correlation	Between-objects error correlation	Powera
1 (n = 16)	70°	.375 (.066)	.369 (.035)	.50
1 (n = 16)	200°	.395 (.076)	.386 (.069)	.75
1 (n = 32)	70° and 200°	.385 (.071)	.378 (.055)	.89
2 and 3 (n = 12)	70° and 200°b	.430 (.117)	.413 (.093)	.42
All (n = 44)	70° and 200°c	.398 (.087)	.387 (.068)	.94

Note. None of the comparisons showed significant differences. The n values refer to the number of correlation pairs used to calculate the means, and in the paired t tests, not the number of participants. Notice that these correlations are low by the standards of circular statistics, in which both positive and negative correlations are calculated, expressed on the scale 0 to 1, and the lower of the two is discarded (Batschelet, 1981).

^aCalculated as ability of one-tailed paired t test to detect medium effect sizes (Cohen's d = .5) at α = .05.

^bPost-disorientation turns only.

^cCombined data from pre- and post-disorientation turns.