Table 2.

Group means or medians for each outcome measure and the results of ANOVAs and post hoc tests comparing study groups.

	Group means or medians a			ANOVA results a			P̂ (X > No Voucher) b
	Abstinence & Training Reinforcement	Training Reinforcement	No Reinforcement	F or H	df	p value	Abstinence & Training Reinforcement	Training Reinforcement
Overall steps achieved c	91 **	88 **	10.5	13.4	2	.001	.71	.70
Typing steps achieved c	49 **	48 *	9	12.2	2	.002	.70	.69
Keypad steps achieved c	42 ***	42 ***	3	17.9	2	< .001	.74	.73
Total hours attended d	106 **	148 **	36	14.6	2	< .001	.71	.73
Steps completed per hour in workplace d	0.64	0.52	0.61	0.4	2	n.s.	.52	.51
Timings initiated per hour in workplace d	36 ***	33	29	13.1	2	.001	.74	.64
Characters typed per timing initiated	63	63	63	0.0	2,115	n.s.	.53	.52
Characters typed per min. in workplace	38 **	34	27	6.8	2,115	.002	.73	.64
Overall accuracy (% correct) d	97.8 *	97.3	96.0	9.2	2	.01	.67	.67
Typing accuracy (% correct) d	97.1 *	97.2	95.4	7.7	2	.02	.65	.66
Keypad accuracy (% correct) d	98.2	97.9	97.2	3.7	2	n.s.	.61	.61

Note. Asterisks indicate statistically significant differences from the No Reinforcement group with post hoc tests (* p < .05, ** p < .01, *** p < .001).

The Abstinence & Training Reinforcement and Training Reinforcement groups were not different from one another on any of these outcome measures. n.s. = not statistically significant.

^aMeans and one-way ANOVAs (F values) are reported for ratio scale outcome measures that passed Levene’s Median Test, while medians and Kruskal-Wallis ANOVAs (H values) are reported for ordinal measures or those that did not pass Levene’s Median Test.

^bThe probability that a random member of the Abstinence & Training Reinforcement or Training Reinforcement group performed better than a random member of the No Reinforcement group.

^cOrdinal outcome measure.

^dDid not pass Levene’s Median Test.