A METHOD FOR COMPUTING THE EFFECT OF AN ADDITIONAL OBSERVATION ON A PREVIOUS LEAST-SQUARES ESTIMATE