Return to search

Assessing school effects with a two level hierarchical linear model

In this study a hierarchical linear model was applied to the analysis of data collected at the student level and the school level in an attempt to address the following issues: (1) The applicability of a two-level hierarchical linear model in assessing relative school effects on mathematics achievement of eighth graders using a large data set; (2) Within the context of a hierarchical linear model, to assess the importance of the effects of variables not under the control of the school on the outcome, to evaluate the effects of the schools on the outcome by analyzing the residuals and obtaining a ranking of the schools based on their effects, and to compare the sizes of the school effects to those of the predictors in the explicit model; (3) Compare the results obtained by the two-level analysis to those obtained by a conventional school-level analysis of the same data. / The study showed that about 11% of the total variance in the outcome scores was due to differences between the schools while 89% was due to differences within the schools. / Within the schools, about 70% of the variance was jointly explained by the students' sex, verbal ability, and minimum competency level. Virtually no variance could be related to race or ethnicity after controlling for the other variables. Between the schools, three school level variables had significant effects on mathematics achievement: The school's mean verbal ability (strong effect), the rate of disciplinary referral in the school, and the percentage of whites in the school (weak effect). / The HLM model and the traditional model explained approximately equal amounts of the between school variance in mathematics achievement. The two models also ranked schools in a fairly similar fashion on their residuals, although significant differences between the two sets of rankings did occur. However, the ranks yielded by the HLM model are based on more information than those yielded by the conventional model. / Source: Dissertation Abstracts International, Volume: 53-03, Section: A, page: 0785. / Major Professor: Jacob G. Beard. / Thesis (Ph.D.)--The Florida State University, 1992.

Identiferoai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_76613
ContributorsSubhiyah, Raja George., Florida State University
Source SetsFlorida State University
LanguageEnglish
Detected LanguageEnglish
TypeText
Format118 p.
RightsOn campus use only.
RelationDissertation Abstracts International

Page generated in 0.0016 seconds