Table 3.

The results of three experiments comparing the weight of feed removed and measured on a calibrated scale (control) and the measured weight reported by the automated feeding system (feeder)

	Experiment 1		Experiment 2		Experiment 3
	Control	Feeder	Control	Feeder	Control	Feeder
Paired t-test for overall relative bias
Number of observations	713	713	151	151	212	212
Mean, g	206.65	206.46	194.14	194.48	194.27	191.84
SD, g	150.09	148.65	132.25	132.74	134.61	133.44
SE, g	5.62	5.57	10.76	10.80	9.25	9.16
Mean bias, g ± SE	0.198 ± 0.515		0.344 ± 0.423		2.425 ± 1.515
95% CI for mean	−0.814, 1.209		−1.181, 0.492		−0.562, 5.411
P-value	0.701		0.417		0.111
Bland–Altman plot analysis, difference = control − feeder
Mean difference, bias	−1.805		0.383		0.723
95% CI for mean difference	−3.525, −0.08557		−1.104, 1.869		−4.54, 5.986
Slope difference vs. average, ± s.e.	0.0097 ± 0.0034		−0.003742 ± 0.0032		0.00881 ± 0.0114
95% CI for slope	0.002946, 0.01645		−0.01007, 0.002587		−0.01362, 0.03125
P-value	0.005		0.245		0.439
Limits of agreement
Lower 95% limit, slope in brackets	−11.442 (−0.01792)		−6.684 (−0.004925)		−11.835 (−0.01439)
Upper 95% limit, slope in brackets	7.831 (0.03731)		7.449 (−0.002558)		13.280 (0.03202)
Geometric mean functional regression, x = control, y = feeder
Constant, mean bias	1.792 ± 0.8686		−0.3831 ± 0.7542		−0.7311 ± 2.655
95% CI for constant	0.08703, 3.497		−1.873, 1.107		−5.966, 4.503
Slope	0.9904 ± 0.0034		1.0037± 0.0032		0.9913 ± 0.0112
95% CI for slope	0.9837, 0.9970		0.9974, 1.010		0.9691, 1.013
P-value	0.005		0.246		0.438
Correlation and reproducibility
Correlation coefficient, r	0.9958		0.9992		0.9865
P-value	<0.001		<0.001		<0.001
Lin’s concordance correlation coefficient	0.9958		0.9992		0.9863
95% CI	0.9951, 0.9963		0.9989, 0.9994		0.9821, 0.9895
Bias correction factor; Cb	1.0000		1.0000		0.9998