Table 2.

Overview of the proportion of data patterns that do not meet our criteria for each participant pool.

TESTING MODALITY	LAB-TESTED		WEB-TESTED
PARTICIPANT POOL	STUDENTS	FISHER	STUDENTS	FISHER	PROLIFIC	FISHER
Step 1 N	40		215		300		196
% anomalous samples	7.5		8.4		9.3		17.3
95% CI	[2.6, 19.9]	≈	[5.4, 12.8]	≈	[6.5, 13.2]	p < .05	[12.7, 23.3]
Step 2 N	37		197		272		162
% no verbal disruption effect	0		16.2		16.9		42
95% CI	[0.0, 9.4]	p < .05	[11.7, 22.0]	≈	[12.9, 21.8]	p < .001	[34.6, 50.0]
Step 3 N	37		165		226		94
No rehearsal primacy effect	5.4		5.5		5.3		27.7
95% CI	[1.5, 17.7]	≈	[2.9, 10.0]	≈	[3.1, 9.1]	p < .001	[19.6, 37.4]
Final N	35		156		214		68
% of total remaining	87.6%		72.6%		71.3%		34.7%
95% CI	[73.9, 94.5]	≈	[66.2, 78.1]	≈	[66.0, 76.2]	p < .001	[28.4, 41.6]

Note: The comparisons between participant pools consisted of Fisher exact tests. For each criterion and for each participant pool, we indicated how many participants remained in the sample on which the criterion was evaluated. For each sample, we indicated what percentage did not meet the criterion. For each percentage of data patterns that did not meet the criterion, we presented the 95% Wilson confidence interval, suitable for binomial data and small samples.