Critérios para amostragem de frangos de corte na avaliação do rendimento de carcaça e cortes

Abstract

From chicken carcass yield and broiler chicken cuts, data value were obtained from chicken at the aged of 42 days to estimate the number of repetitions needed in an experiment, as well as to evaluate the best way to select the chicken for slaughter and the quality of data regarding the presence of discrepant data (outliers). In a first experiment, 236 samples were collected at random, and 236 samples based on the average weight of the experimental unit, in which at 42 days of age each poultry in the experimental unit was weighed to obtain the average weight and selection of 4 poultry with varying weight by 2.5% more and less in relation to the average of the experimental unit. Four poultry were randomly selected per experimental unit. For the second experiment, 480 poultry were selected to obtain carcass and cut yields. From descriptive statistics, analysis of variance and the help of the EPGP.xls program, the necessary number of repetitions to obtain differences between means for different magnitudes was estimated. The results obtained in the first experiment indicated that the way the poultry were selected did not affect carcass characteristics (P>0.05), both with and without the presence of outliers. The random selection of poultry proved to be effective for producing reliable data in the estimation of carcass and cut yields, especially after the removal of outliers. In the second experiment, the VC found were stable. The differences between CV and standard deviation found made the number of repetitions necessary to obtain a difference between means to be different for each variable. The smaller data variation allows finding smaller differences between means with fewer repetitions. Experimental planning should be used as an indispensable tool, and prior knowledge of possible results, as well as possible data adjustments, through transformations or removal of discrepant data, which allow obtaining more adequate and accurate results, making research more relevant and reliable.