Testing the Mean Vector and the Correlation Coefficient in Life-Testing Under Bivariate Normality

In life-testing situations under bivariate normality of (X, U), a few smallest or a few largest Y-observations may not be available. Tests for μ = 0 (mean vector) and o = 0 (correlation coefficient) are developed from the available Y-observations and their concomitant X-observations. The robustness of these tests to departures from normality is investigated.