Table 2.

DLBCL data: Method comparison for estimating the Δ Δ C _q-value

	Estimate	se	t-value	df	p-value	LCL	UCL
mir127 vs rnu6b
EC	2.67	1.13	2.37	22	2.68·10−2	0.336	5.01
EC&VA1	2.67	1.13	2.37	22	2.71·10−2	0.331	5.01
EC&VA2	2.67	1.13	2.36	22	2.75·10−2	0.325	5.02
Bootstrap	2.68	1.05			1.00·10−3	0.876	4.82
mir127 vs rnu24
EC	2.38	1.08	2.2	22	3.87·10−2	0.136	4.63
EC&VA1	2.38	1.09	2.19	22	3.91·10−2	0.13	4.64
EC&VA2	2.38	1.09	2.19	22	3.94·10−2	0.126	4.64
Bootstrap	2.42	1.18			1.00·10−2	0.416	5.02
mir143 vs rnu6b
EC	1.17	0.846	1.38	22	1.82·10−1	-0.589	2.92
EC&VA1	1.17	0.846	1.38	22	1.82·10−1	-0.59	2.92
EC&VA2	1.17	0.847	1.37	22	1.83·10−1	-0.592	2.92
Bootstrap	1.15	0.794			1.44·10−1	-0.341	2.7
mir143 vs rnu24
EC	0.878	0.81	1.08	22	2.90·10−1	-0.801	2.56
EC&VA1	0.878	0.81	1.08	22	2.90·10−1	-0.802	2.56
EC&VA2	0.878	0.811	1.08	22	2.90·10−1	-0.803	2.56
Bootstrap	0.897	0.822			2.67·10−1	-0.603	2.58

EC efficiency corrected LMM estimate ignoring the uncertainty of the efficiency estimates. EC&VA1 EC and variance adjusted LMM estimate using the delta method. EC&VA2 EC and variance adjusted LMM estimate using Monte Carlo integration. Bootstrap Estimate by the bootstrap described in Section “Inference for ΔΔC_q by the bootstrap method” fitting the LMM and using the EC estimate. Bootstrap shows the mean and standard deviation of 4 bootstrap samples using the EC estimate. The last two columns show the 95 % lower and upper confidence interval limits