Improving Livestock Statistics using Balanced and MPPS Sampling Designs

This paper aims to illustrate the efficiency of balanced and Multivariate Probability Proportional to Size (MPPS) sampling designs for improving livestock estimates. Using data from the Brazilian Municipality’s Livestock Survey, the study compares the statistical performances of sampling designs employing Probability Proportional to Size (PPS), MPPS and Balanced Sampling (BAL). The goal is to select samples of municipalities using auxiliary information from 2019 to estimate livestock figures for 2020. Results indicate that MPPS designs present a better overall performance than PPS when considering multiple variables and that, in general, Balanced Sampling contributes to a further reduction in the estimates' variance.