Estimating and testing the significance of correlation coefficient obtained from truncated bivariate normal distribution

Abstract

In this paper we have given a methodology to estimate the correlation coefficient be- tween Combined Admission Test (CAT) score and Cumulative Grade Point Average (CGPA) of the students by truncating the CAT score at different cut-off points. An expression for asymptotic variance of estimated value of correlation coefficient has been derived for testing the hypothesis H0: ρ = 0 against the alternative H1: ρ > 0.

Keywords:

• Correlation coefficient
• Asymptotic variance
• Bivariate normal population