A Chi-Squared approach to Obtaining Missing Values on Egg Production

摘要

In this paper, we estimate the missing values in the outcome of an experiment that involved four different types of poultry birds fed over a period of eighteen months in four feed weight replicates per month. The outcome which is the number of eggs produced monthly by the birds is stochastically dependent on two inputs (feed weight and months). The study assumed a cubic polynomial interaction that follows a binomial series expansion between the inputs. A regression equation was used to link the outcome and the input interaction. The regression coefficients of the link function was determined by the method of ordinary least square via multiple imputation technique while replacing the missing values with the available outcomes in turn. So for each imputation, there is a corresponding likelihood value obtained as estimate of the missing data from the regression equation. The probability of the chi-square goodness of fit was then used to determine the maximum likelihood value of the missing data by plotting the graph of the obtained chi-square density against the available outcomes.