Printed in Great Britain Likelihood estimation of a simple linear regression model when both variables have error

Likelihood based methods for obtaining approximate confidence intervals for the slope in a simple linear regression when both variables are measured with error are discussed and compared. We also derive a Bartlett correction factor for the likelihood ratio statistic. The improvement of the chi-squared approximation is appreciable in the unknown variance case. Further, the distribution of the maximum likelihood slope estimator is obtained from the joint distribution of sample variances and covariance