Evaluation of statistics to be used to quantify the magnitude of errors in the sorting of pigs for market via simulation

The BW growth curves for twenty-five 4,000-head finishing barns were simulated to (1) evaluate methods to quantify the errors in the estimation of BW for market pigs from available data and (2) estimate the effect of sorting errors on the mean and variance in BW and carcass weight. Two types of errors were evaluated, BW estimation error (BWEE) and percentage of pigs not visually evaluated (PNVE). Four levels of BWEE with SD of 0, 4, 6, and 8% of BW and 4 levels of PNVE (0, 8, 16, and 24%) were simulated. Pigs were marketed in 3 marketing cuts (MCUT): 25% at 169, 25% at 179, and the remaining 50% at 193 d of age. The percentage of pigs sold correctly for each MCUT was determined. Two types of sorting errors, pigs not marketed that should have been or pigs marketed that should not have been, were evaluated. The magnitude of the sorting errors was estimated as the pig BW minus the cutoff BW for that day. Statistics identified that differed with (P < 0.01) and quantify the accuracy of sorting are percentage of pigs sold correctly, especially for the second MCUT; magnitude of the errors for pigs sold incorrectly; distribution of the sorting errors; and the SD for carcass weight for pigs of the second MCUT. These statistics can be estimated from currently available data to quantify the accuracy of sorting market pigs. The magnitudes of BWEE and PNVE affect the distribution of carcass weight.