Effects of ignored subpopulations' growth trajectories on estimates of school value-added scores

UNCG Author/Contributor (non-UNCG co-authors, if there are any, appear on document)
Cecilia Sisk (Creator)
The University of North Carolina at Greensboro (UNCG )
Web Site: http://library.uncg.edu/
Robert Henson

Abstract: School value-added scores classify schools into performance categories that are linked to rewards and sanctions. Because value-added scores claim to measure the schools' effectiveness on student growth, inferences of the quality of services provided are made. However, the widespread use of these scores has not yet been sufficiently supported by research as a sound accountability index, particularly when it pertains to its accurate interpretation and its ensuing appropriate use for high stakes decisions. Research shows that several factors can change the classification of schools such as methodology, constructs used, variables used and others. In order to add to the body of evidence of whether the inferences derived from value-added scores can be supported, this research will investigate the effects of un-accounted for latent subpopulations, LEP and student SES at level-1 and school SES at level-2 on the classification of schools' value-added scores and its precision estimates in multilevel data utilizing the multilevel growth mixture model and multilevel linear growth model. This research found that the number of schools identified for special treatment were cut in half when value-added scores were extracted from a multilevel growth mixture model in conjunction with the specification of school SES at level-2, in comparison to a multilevel linear growth model without school SES at level-2. Particularly, the value-added scores' magnitude were less extreme for the more homogeneous schools, the very high SES and the very low SES schools. In addition, precision estimates were improved as well. This suggests that using the methodology that sanctions the larger number of schools would be premature because there are other factors that can affect the value-added scores estimates.

Additional Information

Language: English
Date: 2018
Subpopulations, Trajectories, Value-added
Students $x Socioeconomic status
English language $x Study and teaching $x Foreign speakers
Academic achievement $x Testing

Email this document to