Table 1. Comparisons of endpoint variance.

Test Type	Subject	S1	S2	S3	S4	S5	S6	S7	S8	S9
vertical drift	F(239,239)	0.87	1.10	1.23	0.95	0.86	1.26	1.05	0.68	1.15
vs	p-value	0.86	0.235	0.055	0.646	0.873	0.038	0.365	0.999	0.134
vertical no-drift	evidence (dB)	−19.7	−9.8	133	−3.4	−12.7	−0.23	−7.62	−20.0	−2.9
horizontal drift	F(239,239)	3.56	3.07	3.15	2.78	3.51	3.17	3.77	2.47	0.93
vs horizontal	p-value	<.001	<.001	<.001	<.001	<.001	<.001	<.001	<.001	0.707
no-drift	evidence (dB)	509	196	936	309	172.6	145.8	198.0	88.5	−15.4
horizontal	F(239,239)	0.66	0.90	1.03	0.96	0.83	0.67	0.60	0.56	0.78
vs vertical	p-value	0.999	0.789	0.416	0.606	0.926	0.999	0.999	0.999	0.967
(no-drift)	evidence (dB)	−19.5	−23.1	−26.5	−24.5	−13.5	−16.7	−18.3	−22.8	−28.0
horizontal	F(239,239)	2.73	2.52	2.63	2.82	3.33	1.68	2.17	2.07	0.635
vs vertical	p-value	<.001	<.001	<.001	<.001	<.001	<.001	<.001	<.001	0.999
(drift)	evidence (dB)	151	25.0	266	62.5	163.5	24.8	64.6	54.7	−13.1

F- and p-values are given for comparisons of the horizontal and vertical variances in the drift- and no-drift conditions (for zero-penalty trials only). No Bonferroni corrections were applied, although no conclusions would be changed by applying corrections. For the first and third comparison, positive evidence indicates support for a model in which the two variances are unequal vs one that assumes they are equal. For the second comparison, positive evidence supports a model that assumes horizontal variance is greater in the drift trials than in the no-drift trials (versus a model that assumes equal variances). In the fourth comparison, positive evidence values support a model that assumes horizontal variance is greater than vertical variance (versus a model that assumes equal variances).