Table 1.

Average Test Length for Different SPRT Settings and Item-Selection Methods.

$\rho =.05$	$\alpha =0.05$			$\alpha =0.10$
	$\delta =0.1$	$\delta =0.2$	$\delta =0.3$	$\delta =0.1$	$\delta =0.2$	$\delta =0.3$
Random item selection
0.0	50.000	48.404	43.598	49.916	45.303	38.431
0.3	49.999	47.778	42.383	49.850	44.217	36.949
0.6	49.998	47.186	41.324	49.752	43.218	35.725
Item selection by maximization at the ability estimate
0.0	49.996	43.165	35.065	48.792	37.411	28.624
0.3	49.993	41.819	33.541	48.341	35.918	27.318
0.6	49.989	40.785	32.310	47.887	34.674	26.188
Item selection by maximization at the projected estimate
0.0	49.990	43.362	35.359	48.898	37.673	28.893
0.3	49.992	43.911	35.965	49.165	38.261	29.824
0.6	49.981	43.045	34.890	48.952	37.160	28.739
Item selection by maximization at the cutoff point
0.0	49.531	40.851	32.337	47.269	34.733	25.840
0.3	49.234	39.264	30.660	46.268	33.022	24.230
0.6	48.901	37.815	29.120	45.399	31.541	22.956

Note. SPRT = sequential probability ratio test; $\rho$ = correlation between the abilities; $\alpha =\beta$ = acceptable error rates; $\delta$ = distance between the cutoff point and the boundary of indifference region.