Date of Award
Master of Science (MS)
Mathematics and Statistics
Rex L. Hurst
Ronald V. Canfield
Missing data in regression is often a problem to research workers because standard regression methods are applicable only to complete data sets. At present there are three general methods for solving the problem of missing data.
At first, the reduced data method, reduces the incomplete data set to a complete data set before analyzing. Although this method is very simple to apply, substantial amounts of information are sometimes lost when data is eliminated. This results in less precise estimates of the regression parameters.
The second method, generalized least squares, estimates the missing values through least squares techniques, thus obtaining a complete set of data to which regular regression techniques can be applied. This method is practical but relies on estimates to obtain other estimates, thus again creating some loss in precision. Also, it may require multistage processing which could be very time consuming. Afifi and Elashoff (1966), Yates (1933), Bartlett (1937), Wilkinson (1958), and Goldberger (1964) have all given examples of the generalized least squares method for estimating the missing data
Yamasaki, Gayle M., "The Evaluation of Glasser's Maximum Likelihood Method on Missing Data in Regression" (1973). All Graduate Plan B and other Reports. 1160.
Copyright for this work is retained by the student. If you have any questions regarding the inclusion of this work in the Digital Commons, please email us at .