Table 4.

Ten split-half replications of full population

Part 1: Number of subjects in training and test sets and top five factors chosen per trial
Trial	Number of training set subjects	Number of test set subjects	Number of factors used	Top five factors chosen
1	462	504	20	13, 4, 1, 12, 40
2	494	472	20	13, 4, 12, 1, 3
3	512	454	23	13, 12, 4, 1, 2
4	475	491	18	13, 3, 12, 1, 2
5	480	486	20	13, 4, 2, 12, 7
6	487	479	20	13, 12, 4, 9, 3
7	479	487	23	13, 1, 4, 12, 2
8	492	474	25	13, 4, 1, 12, 3
9	497	469	21	13, 4, 12, 2, 7
10	488	478	17	13, 4, 12, 3, 7
Part 2: Ten test set classification accuracies and t-test results
Trial	Num CON	% CON	Num ADD	% ADD
	Correct	Correct	Correct	Correct	t	df	P
1	285/336	84.82	150/168	89.29	19.76	502	0.0001
2	273/311	87.78	195/218	85.71	18.55	470	0.0001
3	252/296	85.14	145/158	91.77	20.34	452	0.0001
4	275/316	87.03	158/175	90.29	21.96	489	0.0001
5	277/318	87.11	151/168	89.88	20.68	484	0.0001
6	264/307	85.99	155/172	90.12	19.49	477	0.0001
7	274/314	87.26	155/173	89.60	22.13	485	0.0001
8	272/302	90.07	155/172	90.17	20.90	472	0.0001
9	258/291	88.66	148/178	83.15	19.18	467	0.0001
10	261/312	83.65	151/166	90.96	20.32	476	0.0001
Mean		86.75		88.49

Num number of, CON normal control, ADD attention deficit disorder, t t-test, df degrees of freedom, P probability value. Results are the number and percent of correctly classified test set subjects; t values are determined for each test set using the corresponding training-set-developed discriminant function scores