Calculating Missing Student Data in Hierarchical Linear Modeling: Uses and Their Effects on School Rankings
Timothy H. Orsak, Robert L. Mendro, Dash Weerasinghe · 1998
In the age of student accountability, public school systems must find procedures for identifying effective schools, classrooms and teachers that help students continue to excel academically. As a result, researchers have been modeling schools to calculate achievement indicators that will withstand not only statistical review but political criticism. One of the numerous issues encountered in statistical modeling is the management of missing student data. This paper addresses three techniques that elucidate the effects of absent data and highlight consequences on school achievement indicators. The outcomes of each technique are estimated data and School Effectiveness Indices (SEIs). A set of criteria is established from an original data set to determine a baseline to which the analyses will be compared in determining the most appropriate approach in estimating missing data. ompleteness of any data base should be considered a rarity when managing educational data. Numerous factors, not limited to lack of student attendance, data misinterpretation, and mistakes in data entry, all affect the accuracy of any educational database. While incorrect data scores are difficult, if not impossible, to detect, missing scores