Inferential Procedures for Testing Assumptions on Observations for Applications of Biometric Techniques

Abstract

Biometrical Techniques are often used in Genetic Statistics involving plants and animal studies for assessing their genetic potential in selection trails for genetic material improvement. For such purposes, genetic parameters namely means, variances, variance components, heritability parameters, genotypic correlations etc., have been often estimated by using genetic statistical methods. In the Biometrical research analysis, the applications of most of the advanced experimental statistical tools based on certain crucial assumptions such as the assumptions of independence, homoscedasticity of observations on study variable and assumption of normality of observations in the data. Departures from these assumptions may lead to biased and inconsistent estimators; and incorrect conclusions. Thus, the Biostatistician has to test these assumptions on observations rather than to presume that they are correct. In the present article, an attempt has been made by developing test procedures for testing hypotheses about population’s symmetry and population’s kurtosis by using some modified Beta measures. Further, a test for normality of errors in linear regression model has been developed by using modified Fisher’s g-statistics.