Estimation of the Correlation Coefficient for a Bivariate Normal Distribution with Missing Data

บทคัดย่อ

This study proposes an estimator of the correlation coefficient for a bivariate normal distribution with missing data, via the complete observation analysis method. Evaluation of the proposed estimator (ˆρ _J) in comparison with the Pearson correlation coefficient (ˆρ_P ) was conducted using a simulation study. It was found that, for a higher percentage of missing data in a large sample size, the absolute bias of ˆρ _J was less than that of ˆρ _P when the population correlation coefficients (ρ) were not close to zero. In addition, the mean square error of ˆρ_J was not different from that of ˆρ_P in each situation.