. Author manuscript; available in PMC: 2017 May 5.

Published in final edited form as: J Res Educ Eff. 2015 Nov 23;9(4):556–569. doi: 10.1080/19345747.2015.1086916

Table 4.

Scenario 3: Manipulating Participant Selection- Selection Based on Multiple Measures

	Selection on Both Measures					Selection on Either of Two Measures

Cut Point	N^a	Pretest- Posttest r	MDES^b	Participants Needed^c	Estimated Power^d	N^a	Pretest- Posttest r	MDES^b	Participants Needed^c	Estimated Power^d
No Cut Point	100,000	.85	.21	142	.80	100,000	.85	.21	142	.80
≤ 0	38,561	.69–.71	.28–.29	250–264	.54–.56	61,408	.74–.78	.25–.27	202–230	.60–.65
≤ −.33 SD	26,252	.66–.68	.29–.30	270–284	.51–.53	47,939	.71–.76	.26–.28	214–252	.56–.63
≤ −.5 SD	20,778	.65–.67	.30–.30	280–294	.49–.51	40,945	.70–.76	.26–.28	218–260	.54–.62
≤ −.67 SD	16,018	.64–66	.30–.31	290–302	.48–.50	34,103	.68–.75	.26–.29	224–270	.53–.61
≤ –.75 SD	14,026	.63–.65	.30–.31	294–308	.48–.50	31,128	.67–.75	.26–.29	226–276	.52–.60
≤−1 SD	8,976	.60–.63	.31–.32	304–322	.46–.48	22,630	.66–.74	.27–.30	232–288	.50–.59
≤ −1.25 SD	5,472	.58–.63	.31–.32	308–336	.44–.48	15,544	.65–.73	.27–.30	236–296	.49–.58
≤ −1.5 SD	3,112	.56–.60	.32–.33	324–348	.43–.46	10,191	.63–.73	.27–.31	238–304	.48–.58

N is the mean number of observations simulated for the both upper and lower limit parameters;

Calculated at p < .05, N = 200;

Calculated at p < .05, d = .25;

Calculated at p < .05, d = .25, N =142;

Note: Bold = Baseline (Scenario 1); Test 1–Test 2 r = .75; pretest - posttest r = .85 (both Test 1 and Test 2); Test 1 pretest - Test 2 posttest r = .61 – .75. For selection based on two measures, correlations and power for both Test 1 and Test 2 are equivalent. All statistics reported in Table 4 are for simulations of Test 1.