TABLE 2.

Differences from the Concept2 in mean systematic error across the stages for Peach, EmPower, and Weba. Data are mean (%), ±90% compatibility limits, with observed magnitude and p-values for inferiority and superiority tests (p_–/p₊).

	Peach	EmPower	Weba
30-s test	−2.0, ±3.2; Small 0.72/0.06	–38.3, ±9.6; e.large**** 0.998/0.001	−0.2, ±4.3; Trivial 0.35/0.27
Stage 1	3.9, ±4.2; Moderate** 0.03/0.89	−21.4, ±8.0; e.large*** 0.98/0.01	−1.9, ±6.7; Small 0.61/0.20
Stage 2	5.9, ±5.0; Large** 0.02/0.94	−22.9, ±7.9; e.large*** 0.99/0.01	−0.5, ±4.3; Trivial 0.41/0.24
Stage 3	4.0, ±4.6; Moderate** 0.04/0.88	−22.8, ±10.8; e.large*** 0.98/0.02	−0.1, ±4.6; Trivial 0.32/0.28
Stage 4	2.1, ±5.2; Small 0.14/0.66	−24.2, ±12.5; e.large*** 0.97/0.03	1.6, ±2.8; Small 0.06/0.65
Stage 5	0.0, ±6.6; Trivial 0.39/0.38	−25.2, ±15.9; e.large*** 0.96/0.03	1.9, ±3.5; Small 0.08/0.69
Stage 6	1.4, ±5.8; Small 0.23/0.55	−26.2, ±10.9; e.large*** 0.98/0.02	3.8, ±3.4; Moderate** 0.02/0.93
Stage 7	0.4, ±4.8; Trivial 0.29/0.42	−26.9, ±9.6; e.large*** 0.99/0.01	2.2, ±3.3; Small 0.05/0.79

Scale of magnitudes: <1 and >−1%, trivial; ≥1 or ≤−1%, small; ≥3.0 or ≤−2.9%, moderate (mod); ≥5.5 or ≤−5.2%, large; ≥8.5 or ≤−8.0%, very large (v.large); and ≥14.2 or ≤−12.4%, extremely large (e.large).

Reference-Bayesian likelihoods of substantial change: *possibly; **likely; ***very likely; and ****most likely.

The symbols *** and **** indicate rejection of the non-superiority or non-inferiority hypothesis (p_N– or p_N+ <0.05 and <0.005, respectively).

Likelihoods are not shown for effects that were unclear at the 90% level (failure to reject any hypotheses: p > 0.05).

Effects in bold have adequate precision at the 99% level (p < 0.005).