Relative Efficiency of Linear Probability Model on Paired Multivariate Data
Main Article Content
Abstract
The investigation advanced the comparative efficiency of Liner Probability Model on paired multivariate data. On the basis of simulation, vector of means having common variance-covariance matrix were taken into account as data sets subjected to analyses. The linear probability model exhibited a more power tool to detect the presence of significant difference among variables compared to the multivariate paired (Hotellings’-T2) t-test. The Linear probability model is 31.33% and 44.67%., more efficient than the usual counterpart containing two and three predictor variables, respectively. It pays further, that under the regression analysis, individual variable is directly identified in relation to its significant contribution to the dependent variable. This observation is tantamount on determining that such vector of mean disparity is not significantly different from zero against the usual hypothesis. This procedure gains added advantage such that a sweeping generalization of whether vector of means are significantly or insignificantly different is avoided.